Beyond the Köppen-Geiger Climate Classification System, Part II: A New Bioclimate Classification System

This is the second part of a short series in which I'm investigating alternate approaches to climate classification to the popular Koppen-Geiger system. In Part I, we investigated various existing classification systems proposed by researchers in the past, either modifying the Koppen system or trying an alternate approach. This time, I want to take some of the lessons learned in that overview and apply them to constructing a new climate classification system that serves our particular needs. We might call it the "Hersfeldt Bioclimate System" in keeping with climatology conventions, or "Pasta Bioclimate System" in keeping with this blog's conventions; whatever you prefer.

• koppenpasta 2.0
• Goals and Reference Data
• Constructing the Pasta Bioclimate System
• Tropical and Cold Forests
• Arid and Semiarid Climates
• Mediterranean and Pluvial Climates
• Quasitropical and Dark Climates
• Hot Climates
• Extraseasonal Climates
• Oceans
  
• Climate Parameters
• Minimum Temperature (MinT)
• Maximum Temperature (MaxT)
• Growing Season (GDD, GDDz, GDDl)
• Growth Interruption (GInt)
• Aridity (Ar, GAr)
• Growth Supply (GrS)
• Evaporation Ratio (Evr)
• Ice Cover (MinIce)
• The Pasta Bioclimate System
• Unproxied Koppen-Geiger 

Back to Part I

koppenpasta 2.0

Before we get to all that, a bit of bookkeeping: as mentioned last time, to prepare for implementing a new climate classification system, I have substantially overhauled the koppenpasta script, reworking much of the internal functions to make it much easier to add algorithms for new climate systems and options to tweak the outputs, essentially converting it from a tool specifically designed around the Koppen system to a more general climate classification tool.

From the user side, the functionality is much the same but with a few tweaked options and extra toys:

• Land zone options have been sorted into main types for the separate systems described in Part I and subtypes with slightly different approaches or subsets of zones.
• Sea climate zones can now be selected separately, with its own subtypes and color options.
• A "seasonless" approach, feeding in averaged annual data to the classification algorithm rather than monthly data, is now a separate option that can be applied to any climate zone (though it may not work all that sensibly for some of them).
• There are now a few different options for how data from different files is combined to produce a climate map:
• "Average data" is the same procedure as old versions of the script: basic climate data for each month (e.g. temperature, precipitation) is averaged across all files first, and then important parameters (e.g. hottest month, total precipitation) is calculated based on those averages, treating them as if they came from one file.
• "Average parameters" calculates parameters for each file separately, and then averages those parameters together across files; this takes a bit longer but might better reflect cases where, say, the hottest month is at different points of the year in each file.
• "Sequential" treats the files as separate portions of a single year, lining them up in order of input (which should be alphabetical if loaded from a folder); so you could have 4 files each representing different quarters of a single year, and then use this option to treat them properly as one long year.
• Interpolation, including temperature adjustment by topography, is properly implemented as an option for all climates, can now rescale to any given resolution (computer memory allowing), and can use a couple alternate routines where the usual spline method proves unreliable.

I've also added an option for generating additional images beside the main map:

First, the script can now automatically generate a map key showing the appropriate color and name for all zones that appear on the map.

 

 

Second,  it can also generate a chart of the average temperature and precipitation of every point on the map, colored by the zone of that point.

 

 

The most substantive change is that a lot of the more esoteric options for things like varying climate zone definitions or internal routines have been moved out of the default command line interface. Most people probably won't need to alter these options at any point, but for those that do these can now be set a few different ways:

• Within the koppenpasta script itself, there's a clearly marked "OPTIONS" section near the start of the script containing a list of all the options with short explanations.
• The download includes a separate configuration file, "kpasta_options.cfg", which is not required to run the script but will be used if present in the same folder, overriding the default options in the script.
• You can also copy this configuration file, modify it, and give it a different name, and then during setup of koppenpasta there's an option to select such a file, overriding both of the above.
• During setup, there's also a final option that allows you to enter option names and values directly through the command line, overriding all of the above.

Note that the options lists in koppenpasta and kpasta_options.cfg also contain the default values for the options set through the command-line setup, like land type and color, so by altering these you can alter the script's default behavior when you choose to skip setup. In general, options are prioritized in the order of command-line input >  custom config file > kpasta_options.cfg > script defaults, with the former overriding the latter. If you want to modify the script, additional options can be added to the script or configuration file, and will then be available to any functions within the script; incidentally you also don't have to include every option in the configuration file, if it lacks any it will default to the options in the script.

 

Most of these extra options are technicalities of zone definitions or designed to deal with different data types other than ExoPlaSim outputs—making it easier for me to continue doing public data explorations in the future—but there's a few worth highlighting that you might not use often but on occasion:

• const_lapse_rate: when using temperature adjustment by topography, usually the program tries to determine a regional "empirical" lapse rate (change of temperature with elevation) based on variation between neighboring cells, but this isn't always entirely reliable; if you set this value to a number other than 0 or None, then that number will be used as a single global lapse rate (it should be in units of °C decrease per km elevation, positive for decreasing temperature with elevation; 5 °C/km is a typical value for Earth).
• image_scale: This rescales the final image, without any interpolation involved. You can either input a single scaling factor to multiply the size of the image or a specific (x,y) resolution (in that format, with a comma and parentheses). This might be useful if you're producing uninterpolated images but want them at a more readable resolution like (1024,512) rather than (128,64).
• debug_file: Setting this to True causes the script to save all data extracted from the input file, parameters calculated from that file, and all option settings to a netCDF file, which might help identify any issues within the script.
• temp_adjust, precip_adjust: These are the factors used to convert all temperature- and precipitation-related measures in the input files to the standard Celsius and mm/month scales used within the script. By default they're set to the necessary conversions from ExoPlaSim's usual output scales (Kelvin and m/s), but if you've used my eps_avg script to convert the units in your data files, you may need to adjust these.
• pet_gascon: this is the effective gas constant for use calculating potential evapotranspiration with any climate system that needs it (e.g. Holdridge, Prentice, my new system). It depends on the average molar mass of the atmosphere, so if you have a nitrogen-dominated atmosphere similar to Earth (regardless of total atmospheric pressure, only composition matters) you shouldn't need to change it from the default of 287.05. If you have a substantially different atmospheric composition, you may need to adjust it for accurate PET: usually in your ExoPlaSim outputs there will be a file with "diag" in its name that can be read as a text file, which should have a line near the front listing the "GASCON".

There are a few others that may be worth tweaking related to my new classification system, but I'll mention those later.

 

Custom color lists are also still supported, for all climate zones including any you might add, and it's no longer necessary to include all zones in each color list, only those you'll actually use; you can separately decide if land and sea zones should reference the color list, but if both are doing so they will use the same file.

 

Note that all this overrides the old config file system from previous versions, and any files made with that system are no longer supported. Old color lists should still work, as they're formatted the same way.

 

All the classification systems I showed generated maps for last time are available as options. Some specific notes on some of them:

• Koppen-Geiger climate zones work in much the same way, with the same full, no-As, 14-zone reduced, and group-only variants as subtypes, and also a new 2-letter option, categorizing zones by only the first 2 letters in the Koppen designation, as I've seen in some papers. For colors, you have the same set of regular (blue rainforest), alternate (red rainforest), and true color approximation as before. The various configuration options once in the command line input have been moved to the extended options, and I've added options to use the aridity threshold calculation and Xs/Xw definitions from Trewartha.
• For Koppen-Trewartha I've implemented a modest 2-letter, 14-zone system, based on Belda et al. 2014, but with group E split into Eo and Ec. Importing Am or the Xxa/Xxb subtypes from Koppen-Geiger wouldn't be too difficult, I just haven't wanted to bother coming up with enough colors; the default color scheme is also taken from the paper, with some modification, but there's also a "Koppen-alike" option copying the standard colors for the most similar Koppen zones for each type. Many of the configuration options for Koppen-Geiger are applied here as well.
• Holdridge life zones are still here, but as mentioned I've properly incorporated PETR now, though for simplicity's sake I'm approximating the hexagonal grid with a staggered rectangular grid and prioritizing average biotemperature over the other two factors. In addition to the naturalistic green-yellow shades from wikipedia, there's also a more vibrant color option taken from Audebert et al. 2024. There is still an option to use the old scheme of indexing zones by precipitation and biotemperature only, as well as a configuration option to attempt to roughly predict average biotemperature if only average temperature is available, as I've been doing for my public data explorations, though don't regard that as a particularly reliable approach. So far I haven't implemented a warm temperate/subtropical division, as I'd have to decide how best to approximate the frost line with data available from ExoPlaSim (and choose appropriate colors), but I might try that in the future.
• For Thornthwaite-Feddema I've implemented the primary 36-zone classification by annual average PET and Pr/PET ratio and an alternate 12-zone classification by climate variability, with colors taken from the paper.
• The Prentice et al. 1992 BIOME1 model has been implemented, mapping each vegetation type and then defining zones by their overlap.
• Whittaker biomes, Woodward vegetation types, IPCC climate zones, and the Sayre et al. World Climate Regions are all implemented, as described in the last post.
• Sea zones include the usual options for 4 zones, no separate tropical zone, or one sea zone, and there's also options to assume no sea zones at all, or color all seas white.

I've done my best to document the script and indicate how it should be edited to alter the climate zone scheme or add a new one, but the more flexible nature of how it operates has also made it a bit less linear. There's a guide near the start on how data generally flows through the script.

Goals and Reference Data

Now that we have the necessary tools, let's review some of our main goals and challenges. For all the different approaches we reviewed in Part I, I still want a system that fills the same sort of role as Koppen; it should attempt to predict the distribution of biomes based on climate parameters. Indeed, prediction of biomes is more specifically the goal here, taking priority over any broader considerations of climate conditions, though those will hopefully be clear along the way as well. But compared to Koppen, there are three primary directions of improvement I want to make:

• First, definitions should be based on the climate parameters with the most direct functional link to biome boundaries, rather than indirect proxies.
• Second, definitions should be adjusted and expanded to better accommodate a broader range of worlds with different temperature ranges and seasonal planets, at least to the range that could conceivably support Earthlike life.
• Third, potential differences in biology should be accounted for to some extent, at least in the sense that boundaries should reflect ecological reactions to fundamental environmental barriers that we might reasonably expect to appear on alien worlds, rather than the particular evolutionary quirks of modern Earth.

Most suggestions I see for modifying the Koppen system are focused on that second goal, essentially just tacking on extra categories to cover greater extremes of summer heat or overall seasonality (by the way, if you do something like that I'd suggest not using "x" as the letter designation for any of these additional categories because it's common practice to use "x" as a stand-in for any unspecified letter; e.g. I can say "Xxa" to refer to all zones ending in "a", meaning Cwa, Cfa, Csa, Dwa, Dfa, and Dsa). But, with a mind to that first goal, I want to make some more fundamental changes here, essentially borrowing some of the methodology of the Woodward and Prentice et al. models while still trying to retain some of the flexibility and greater detail of the Koppen system.

I also want any such extension into exotic climate to have some similar connection to ecology and biomes, rather than picking arbitrary temperature thresholds. Because we have no ecology to observe from such environments, this may be quite difficult to achieve, but there is some research to go on that at least suggests some potential challenges and ecological barriers that life may encounter.

Similarly, the third goal requires making a fair few assumptions about the nature of life on other worlds, even if we explicitly limit the scope of this system to life with broadly similar biochemistry to Earth. Ultimately, though, the less assumptions we make regarding common ecological barriers across worlds, the less we can say about their ecology at all, so in many cases we're faced with a choice between making highly contingent biome predictions or making no predictions at all. This is essentially the same quandary that many climate classification systems face just predicting biomes on Earth: there are confounding factors outside of climate, so these predictions must always have some inaccuracy, but if we're determined to attempt some prediction then we just have to accept that inevitability. Suffice it to say that I will generally tend to make assumptions here where they are required to allow for specific predictions, but there are a few cases where biome boundaries on Earth are clearly linked to the specific context of our current planet rather than fundamental ecological mechanisms.

 

Now, if we want this system to correspond to real biomes and we'll be judging it largely on its success doing so, we need some reliable data on where those are on earth. I've picked out 5 primary sources to draw on:

• The WWF Terrestrial Ecoregions, based on an influential 2001 study by Olson et al. which classifies land areas into biogeographic ecoregions that are in turn sorted into biomes and biogeographic realms. It uses 15 biomes, and 4 of these are at least partially based on terrain or elevation factors that couldn't be identified from climate data, so it's a fairly coarse categorization but still a good place to start, and the ecoregions give some sense of finer biome boundaries.
  
  

  

  This original version has been a tad adjusted now and excludes the "Rock and Ice" category from the key.Olson et al 2001
  

  
  
• The IUCN Global Ecosystem Typology, which classifies land areas into 7 broad biomes, subdivided into 34 functional groups with distinct ecology. 1 of these biomes is areas of intensive land use by humans and a few of the functional groups are defined mostly by terrain features (as are the various terrestrial/marine/freshwater transitional biomes also included in the typology), but that still leaves us with around 25 ecosystem types to work from, including some not always mentioned in biome categorizations like heathlands and dry woodlands. However, these biomes and functional groups are not defined as exclusive areas, but mapped out as often overlapping regions where they're more or somewhat less likely to occur, acknowledging that many areas have a patchwork of different ecosystems, so though it helps suggest some potential ways to add more detail, it's not as useful for the exact placement of important boundaries.
• Haxeltine and Prentice's 1996 paper describing BIOME3, one of the followups to Prentice et al.'s 1992 model, includes a fairly detailed map with 18 biomes based on a variety of sources, constructed for gauging the success of their model but useful for us as well, with particularly good detail on semiarid regions often left a bit neglected in biome breakdowns.
  
  

  

  Haxeltine and Prentice 1996
  

  
  
•  The book Habitats of the World: A Field Guide for Birders, Naturalists, and Ecologists, by Iain Campbell et al. It somewhat follows the Olson et al. 2001 study in broadly categorizing biomes, but then breaks them down into biogeographic regions somewhat larger than the WWF ecoregions, and so more directly corresponding to distinct biomes, though it's similar to the IUCN approach in that it doesn't mark these regions as wholly exclusive with each other. Because it's written as a travel guide, it specifically focuses on areas with a different feel or appearance, and so highlights some areas like taiga savanna or miombo woodland that aren't always identified in more rigorous studies.
• This 2014 map by Pfandenhauer and Klötzli, , produced for the German textbook Vegetation der Erde, based on several previous sources; I've attempted an English translation of the key here, helped along by a version of this map in Fischer et al. 2022. It's detailed, including 32 exclusively marked biomes (plus wetlands as essentially a modifier), and also hits an interesting balance between mapping global biomes based on shared functional plant types, while also sometimes highlighting biomes that are more unique to particular regions and so may reflect local evolutionary quirks, which helps in gauging how to interpret similar biomes across continents.
  
  
  

The amount of variation that exists between these systems goes to show how subjective these classification systems can be, and all show a world substantially influenced by human activity. They all make some attempt to reconstruct the "natural" state that areas would have if they weren't covered by cropland, pasture, or other development, but thousands of years of agriculture, herding, lumber use, and fire setting have substantially shifted biome boundaries in some areas in a way that can be difficult to reconstruct, not to mention more recent climate change. So even aside from any other confounding factors, we'll have to accept some problem areas.

That all settled, it's time to put the system together. I won't go through every iteration and detour of the process, but I do want to lay out the main logic and motivations behind the resulting system on a conceptual level, after which I'll list out a bit more rigorously the required climate data and resulting bioclimate zone definitions.

Constructing the Pasta Bioclimate System

Let's start with the broadest biome concepts. When unhindered by major environmental barriers, life on land tends towards the end state of a forest, filling the space to the limit of available photosynthetic area (ecologists might call this a climax community, but we'll leave that sort of discussion to another time). However, various areas of the planet do have environmental restrictions, in some cases to the opposite end state of preventing the growth of vegetation entirely. Leaving aside areas with poor terrain stability or soil quality (neither of which cover much area), on Earth such areas occur where it is either too dry for growth or too cold, but we can easily envision other habitable worlds with substantial areas too hot or too dark as well. Between forests and these barren extremes are a number of transitional biomes, but exactly how those biomes look depends on the opposing end states. We'll start with forests, then, and then work outwards to the biomes that emerge as additional restrictions are added.

Tropical and Cold Forests

Life on Earth generally seems to be happiest at temperatures around 20-30 °C, forming tropical rainforests that are the most biodiverse and productive (in terms of rates of carbon fixation) regions on modern Earth. Outside of this range, growth rates decline, but even tropical plants can weather occasional growth interruptions: the real issue is that low temperatures can damage vegetation, so the extent of tropical biomes is mostly bounded by thermal tolerance limits.

Based on our previous sources and other work, here's a range of potential tolerance limits for minimum temperature encountered for more than a brief period in a typical year, for a range of life forms—do note that some of these are very approximate:

• 18 °C: tolerance limit for shallow marine coral.
  
• 10 °C: tolerance limit for some particularly warm-adapted tropical plants.
• 0 °C: the frost point, when ice formation may damage exposed soft tissues; tolerance limit for tropical plants; minimum growth temperature for even cold-adapted plants.
• -15 °C: tolerance limit for frost-tolerant broad-leaved evergreens
• -20 °C: minimum growth temperature for extremophile photosynthetic microbes.
• -40  °C: tolerance limit for many non-boreal plants.
  
• -60 °C: tolerance limit for needle-leaved evergreens; approximate freezing point of water-H₂O₂ mix.
  
• -80 °C: minimum temperature with observed metabolic activity; potential atmospheric freezeout in CO₂-rich atmospheres.
• -97 °C: freezing point of water-ammonia mix.
• -180 °C: atmospheric freezeout in N₂/O₂-rich atmospheres; freezing point of Li-ammonia brines.
  

Of these, the 0, -15, and -60 °C limits generally seem to be the most important in dividing forest types, at least on Earth, with the first generally marking the limits of the tropical zone. So to start off with, we'll divide the bioclimate system into two first categories:

• Tropical (group T), which retain clement temperatures year-round.
  
• Cold (group C), which have periods of winter frost.

I'll explain the exact thresholds using for dividing zones later

The other tolerance limits do have some relevance within group C, but the most notable shifts in forest composition relate more to the overall seasonal cycle and length of the growing season. Forests on Earth show a curious back-and-forth pattern of evergreen trees, which retain their leaves year-round, dominating in the tropics and subtropics (we might also call them evergrowth if we don't want to make any assumptions about the dominant color); deciduous trees, which drop their leaves in fall and regrow them in spring, dominating in the mid-latitudes; and then back to evergreen trees in cold polar forests (and then one final reversion to deciduous conifers in the most seasonal parts of Siberia, but this is likely due to cold tolerances we'll get back to shortly).

There's a number of factors that play into this pattern, but for our purposes the simple version is that each transition reflects how different parts of the seasonal cycle impose different stresses on a tree and so favor different optimal growing strategies. Each leaf can be understood as a sort of economic investment on the part of the tree: there is an upfront "cost" in terms of energy and nutrients required to grow the leaf, and then during each leaf's lifetime it generates "income" in terms of energy from photosynthesis but still has ongoing maintenance "expenses" to repair damage and replace water losses due to evapotranspiration through the leaves. For a tree with limited energy resources and space, the objective is to maximize the rate of total "profit", in terms of income minus expenses relative to the lifetime of the leaf, in order to maximize growth and so outcompete its neighbors.

Within the warm tropics, consistent high temperatures (and moisture within heavily forested areas) provide good growing conditions year-round means that leaves can always produce more income than expenses. As a leaf ages, it may begin to degrade (and there is some stochastic chance of damage due to herbivory, fires, etc.) but each leaf can just be individually replaced at the optimal point to maximize profit.

 

In colder climates, there will be a winter period where photosynthesis slows, but maintenance must still be maintained, so income may fall below expenses and leaves cause an overall energy loss (frozen ground also yields less water, so evapotranspiration losses through leaves may dangerously deplete the plant's water). In subtropical climates, winters are still short and mild enough that the cost of maintaining leaves through this unproductive period is still small relative to the cost of growing whole new leaves in spring (there's also an added opportunity cost to losing and regrowing leaves, because the tree misses out on some potential photosynthesis in late fall after losing leaves and early spring before growing new ones), so trees still mostly retain their leaves year-round, though they may need to make some adaptational compromises to tolerate the occasional frost.

 

But in even colder climates, that winter maintenance cost mounts. The tree can adapt to reduce those expenses by having thinner leaves with thicker protective coatings to reduce frost damage and evapotranspiration losses, but these leaves will also be less productive in summer. In areas with mild winter temperatures or poor summer growing conditions (generally because of limited water or heavy cloud cover), this tradeoff may be worth it, but in most temperate climates on Earth the optimal strategy is to become deciduous, dropping all leaves in fall and growing new leaves in spring. The cost of growing new leaves is still generally higher than that of retaining old leaves, but because these leaves don't have to survive any frosts or last for more than a year, they can be optimized for maximum productivity in their short summer lifetimes, returning a greater overall profit.

 

Thus, evergreen subtropical forests dominate where winters are short and the cost of retaining leaves is quite low, but once winters become long and cold enough to significantly interrupt growth, they generally give way to deciduous temperate forests (though some temperate areas instead have hardier evergreens).

But as we go on to colder climates, even longer winters don't much increase the maintenance expenses for retaining leaves; as temperatures drop below freezing, plants can go largely dormant and minimize losses to metabolic activity or evapotranspiration, especially hardy frost-adapted evergreens (i.e. conifers). The more important change is that the growing season shortens, such that deciduous leaves have an increasingly short period to recoup the cost of their growth before winter returns. Even if optimized for maximum production in that short summer, eventually the production over the course of each summer barely covers the cost of growing new leaves each year. Evergreens may produce less each year but accumulate a higher net profit over successive years thanks to lower average costs (and again can take better advantage of early spring and late fall, which are a larger portion of the total potential growing period). Thus, temperate deciduous forests transition to evergreen boreal forests.

 

The overall upshot here is that the subtropical-temperate transition is mostly related to the length of unproductive winter, regardless of the exact length of the growing season, and the temperate-boreal transition is mostly related to the length of the productive summer, regardless of the length of unproductive winter.

 

Thus, the group C is primarily split into 3  main subgroups:

• Subtropical (CT), with short winter interruptions to growth.
  
• Temperate (CD), with a long winter interruption but still a long growing season.
  
• Boreal (CE), with a short growing season.
  

It is worth emphasizing again that we shouldn't get too attached to thinking of these subgroups purely in terms of evergrowth and deciduous trees. As mentioned, there are some more evergreen-dominated CD areas that are still clearly distinct from their subtropical and boreal counterparts, lacking the broad leaves of the former but with faster growth and lusher undergrowth than the latter. On Earth, these tend to be associated with especially heavy rains (temperate rainforests) or conversely summer droughts (Mediterranean), but the exact balance of necessary conditions is tricky to pin down with simple climate parameters and could conceivably vary considerably with different year lengths and seasonal cycles. Even in more deciduous-dominated areas, CD forests tend to have some evergreen trees taking advantage of their exclusive access to early spring and late fall light or their better performance in poor soils due to their lower nutrient requirements, and for their part CT forests may shade to semideciduous in drier areas while CE regions may have some deciduous trees where winters are particularly harsh.

To avoid overtuning this system to Earth, then, it's perhaps better to think of these zones not strictly in terms of evergrowth and deciduous habits, but more broadly of CT as regions warm enough year-round to support tropical-like vegetation, despite the inconvenience of frosts; CE as regions with only intermittent short periods of growth that can support only the most conservative growth patterns; and CD as a range of intermediate climates that have substantial periods of growth but fall short of the heat or consistency required for thick tropical vegetation. In this looser sense, these ecological boundaries are also more likely to have some relevance to alien biospheres with different photosynthetic structures.

 

Past the boreal forests, eventually the growing season gets too short to cover even the hardiest evergrowth's costs, though we can subdivide the polar regions a bit based on their suitability for at least some low shrubs:

• Tundra (CF), with a very short but still significant growing season.
• Cold Barren (CG), with no substantial growing season but no permanent ice cover.
  
• Ice Sheet (CI), with permanent ice cover.

The tricky bit is how well these patterns will apply to other worlds with different seasonal patterns:

If a world has shorter years but a similar range of seasonal temperature variation, both summers and winters will shorten, which should tend to shrink CD zones until CT transitions directly to CE; in such cases I'll divide them based on winter growth interruption, on the presumption that so long as winters are brief and mild, subtropical plants should be able to simply continue to grow across years with little issue; but once winters are severe enough to substantially interrupt growth, the short growth periods between winters becomes a more serious issue. Extremely short years could perhaps eliminate even CE zones, but again we'll take CT and CF zones to be divided based on growth interruption on the same logic.

Very long years conversely favor CD climates, though always with some transition to CT and CE climates so long as there are still tropical and polar regions. In subtropical climates, long summers aren't likely to change the equation much in terms of when winter growth interruptions are long enough to force a switch to deciduous or hardier evergrowth leaves. In boreal climates, however, even if maintenance costs remain very low for the bulk of winter, a substantially longer winter may cause more degradation of leaves between growing seasons, increasing the relative benefit of growing new leaves. However, I have no real way to predict at what exact effect this might have without making several assumptions about the nature of the vegetation involved, so I'll make no attempt to account for it here. Extremely long years might also allow for larger plants to complete their entire lifecycle in a single year before dying in fall, opening up more potential niches, but again there's not much basis to speculate on exactly how that might influence biome transitions.

Worlds with very low seasonality present a bit more of a challenge, as they create climates with year-round mild temperatures, which we don't see too often on Earth. I'll discuss the specific numbers later, but on Earth there seems to be significant overlap between the range of temperatures cold enough to interrupt growth in subtropical climates but warm enough to support growth in temperate climates, so the most straightforward extrapolation is to classify these consistently mild climates as CD, but rather than deciduous plants we'd expect them to be dominated by evergrowths similar to temperate rainforest or Mediterranean regions. Exactly how seasonal a world would have to be for deciduous plants to appear is difficult to determine from Earth data alone, and might depend on the specifics of local evolution anyway.

 

At any rate, for further subdividing cold C climates we can now return to minimum temperature tolerances: around -15 °C is cold enough to damage plants without more substantial adaptations to frost tolerance, causing a subtle shift in forest tree and undergrowth composition, and -60 °C is cold enough that even the most frost-tolerant plants cannot prevent ice formation in their exposed soft tissues, forcing a final switch from evergrowth back to deciduous plants. We can thus subdivide our groups based on winter temperatures:

• Oceanic (Cxa) zones have cool winters that don't require intensive frost tolerance.
• Continental (Cxb) zones have cold winters requiring greater frost tolerance.
• Percontinental (Cxc) zones have frigid winters too harsh for even frost-tolerant evergrowths.

But to keep down the total zone count, we'll trim this system in a few places:

• Subtropical (CT) zones will be restricted to cool winters, on the presumption that even a short but cold winter will damage subtropical evergrowths and so force a switch to deciduous (or more frost-tolerant evergrowth) plants.
  
• The percontinental category will only be applied to boreal (CE) zones, as the frigid winter threshold should only be relevant to frost-tolerant evergrowths.
• Tundra (CF) will be divided into cool and cold variants, to divide milder coastal and highland tundra that might support more shrubs and grasses from harsher continental tundra, but CG and CI zones have too little life for minimum temperature to matter much and are unlikely to have cool winters at any rate.

Arid and Semiarid Climates

Earth's other main gradient from lush forests to lifeless deserts is, of course, linked to water availability. Photosynthetic production is limited by soil water availability both because water is directly consumed in photosynthesis and because there's no practical way for a plant to allow CO₂ to passively diffuse into its photosynthetic structures without also allowing some water to diffuse out (both CAM and C₄ photosynthesis limit this water loss in different ways, but impose their own productivity costs—we'll discuss them both more another time). Plants in dry conditions must then maintain a tricky balance between growing as much as they can to outcompete their rivals while never exhausting their water supply completely, which would cause them to desiccate and sustain damage or die.

Some plants may respond to long droughts by becoming deciduous, just as temperate plants respond to winter, but others may instead respond by developing hardier leaves or reducing their leaf area (through which water is lost) relative to their root area (through which water is gained). This creates a somewhat complex tangle of semideciduous forests, where a short dry season prompts some trees to become deciduous while others can remain evergreen, or some may be optionally deciduous depending on drought severity in a given year; dry forests, where more severe droughts prompt all plants to become regularly deciduous; savannas, where trees are present but with an open-enough canopy to allow light for substantial grass to grow between them; and more exclusive grassland or shrubs, though this is fairly rare in tropical climates (note that grasses are a specific phylogenetic group on Earth that rose to prominence fairly recently, but for convenience I'll be using the term to refer broadly to low-growing vegetation that grows in the open and has little or no permanent above-ground structure, which is likely to be a common niche). Then within savanna there's also a lot of variation along fairly continuous spectra, both from moist to dry varieties with increasingly severe droughts requiring more substantial adaptations for drought tolerance, and from more open savannas with only scattered or very low trees to woodlands where trees are as densely packed as forests but with individually small or thin canopies such that enough light still reaches the ground to support grasses (and there's variation as well in terms of dominance of grasses or low shrubs between the tall trees).

 

There are some parallels here with cold climates, because drought and winter pose some similar challenges—frozen ground yields little water to trees in winter—but there's no direct analogue to boreal forests in dry climates; evergrowth trees can become more common in cooler semiarid climates, and in very dry climates cacti and other succulents might be considered a form of evergrowth, but there is no range of drought conditions that specifically favors dense forests of evergrowth trees, likely because hot, dry conditions impose a much higher maintenance cost for retaining leaves, reducing the potential benefit compared to losing them and growing new ones when rain returns.

Many classification systems, especially Koppen, make little effort to distinguish many of these categories, and generally have trouble dividing grasslands and savanna from forests in cooler climates as well. This may be due to a couple reasons: First, definitions of these biomes can vary considerably, in particular with regards to which woodlands and temperate grasslands can be counted as savanna, and really these may be quite gradual transitions in many cases. Second, the distribution of these biomes may be strongly influenced by factors outside of climate, like soil quality. Frequent fires or grazing by large herbivores in particular may limit tree growth, forming savannas where forests might grow otherwise, and these factors have some relation to climate but are harder to predict than simple thresholds of water availability or growing temperature.

 

For my part, I could not find any measure to consistently divide open savanna and grassland, woodland, and dry forest, but I did at least attempt a somewhat finer breakdown of semiarid regions based on the aridity factor, here defined as the ratio of actual evapotranspiration to potential transpiration, as used in the Prentice et al. model. Where the factor is high, a small drop indicates some period of dry conditions which in tropical climates will require some adaptation regardless of exactly when it occurs; and where the factor is low, a small rise shows some period of rain, which will support some marginal plant life, again regardless of exactly when it occurs. At intermediate levels, however, the best match to biome boundaries comes from measuring the aridity factor specifically during the growing season, as the capacity for a semiarid region to support thick woodland or lush shrubs depends more strongly on whether water is immediately available during the optimal growth period.

We'll start with tropical zones, which have the most clear gradations between wet and dry conditions:

• Tropical Rainforest (Tr) has a very high aridity factor, allowing for thick evergrowth.
  
• Tropical Forest (Tf) has a lower but still high aridity factor, shifting towards a more semideciduous composition.
  
• Tropical Moist Savanna (Ts) has a moderate overall aridity factor but still somewhat high in the growing season, allowing for lusher woodlands and savanna.
• Tropical Dry Savanna (TA) has a low growing season aridity factor, favoring more open savanna or deciduous dry forests.

 

Even drier climates cannot support much in the way of trees or lush undergrowth, and so share little life with wet tropical zones; we'll thus split off our third top-level group:

• Arid (group A), with a low aridity factor too low for forests.
  

Many desert regions do actually support a fair amount of grasses and shrubs, so we can subdivide this group a bit as well:

• Semidesert (Ad) has a moderately low aridity factor still allowing for widespread vegetation.
  
• Hyperarid Desert (Ah) has a very low aridity factor, with little vegetation.

 

This overall scheme can then be generally carried across into colder climates, but with decreasing variety as there are fewer viable strategies to tolerate both harsh winters and drought, such that diverse savanna and woodland transitions to more uniform grassland and steppe.

• Subtropical (CT) zones lack a distinct rainforest category, but still partially parallel tropical zones:
• Subtropical Forest (CTf) has a high aridity factor and so thick evergrowth forests.
  
• Subtropical Moist Savanna (CTw) has a lower aridity factor and so tends towards less lush semideciduous forest or woodland.
• All C zones warmer than tundra will remain as forests so long as the aridity in the growing season is high, but then transition into semiarid zones depending on temperature:
• Cool Semiarid (CA) has a low growing season aridity factor.
  

The distinction between hot and cold arid zones in many systems seems to relate mostly to the transition between C₃ and C₄ grasses; C₄ plants use CO₂ more efficiently, and many plants lose water mostly in the process of gaining CO₂, so they're at an advantage in hot, dry environments with high evapotranspiration. However, the temperature at which C₄ plants gain that advantage depends strongly on the atmospheric CO₂ level, and in general C₄ photosynthesis evolved fairly recently on Earth, so there's little guarantee that this distinction will consistently apply across planets. However, the cool/cold winter temperature line does seem to roughly correspond to the transition from savannas and dry forests parallelling tropical climates to more typically treeless grasslands in cooler climates:

• Cool Dry Savanna (CAa) has cool winters.
• Cool Steppe (CAb) has cold or frigid winters.

The tropical/cold distinction doesn't have much impact on deserts, but again cold winters do have some more impact on vegetation and wildlife and generally seem like a good way to divide up what will be very large zones.

• Warm Arid (Axa), with cool or warm winters.
• Cold Arid (Axc), with cold or frigid winters.

As we approach the coldest climates, the aridity factor becomes generally less important as growing seasons become too short for some plants, evapotranspiration remains low, and frozen ground and cold air in winter yields little water regardless; so we'll give some precedence to cold zones over arid zones:

• Tundra (CF) will take precedence over semiarid (CA), so CA terminates where the growing season is too short for trees or large grasses regardless of aridity.
• Cold Barren (CG) will take precedence over Arid (A), so A terminates where the growing season is too short for almost any vegetation regardless of aridity.
  

Mediterranean and Pluvial Climates

The final element to consider for Earthlike land climates is a finer breakdown based on seasonal precipitation patterns. The Koppen system of course has its dry-summer Xs and dry-summer Xw zones, but these are defined by seasonal extreme months in a scheme that makes them somewhat awkward to apply for worlds with a different number or length of months or substantially different seasonal patterns. The overlap of these zones with Koppen's various thermal categories also adds 14 zones to the scheme, accounting for almost half of the total despite covering less than 11% of Earth's land area altogether; that in itself isn't necessarily a problem given that Earth is also a bit skewed in the relative area of different biomes, but many of these Koppen zones don't have a clear connection to Earth's actual biomes.

We'll start out with Mediterranean or chapparal climates, which have the peculiarity of having relatively dry summers and wet winters. The reverse is usually the norm, because evapotranspiration is higher in summer and moisture-laden winds tend to converge on warm landmasses, but Mediterranean climates appear in areas awkwardly caught between circulation cells or where shifting winds around mountains create seasonal rainshadows. These areas should also ideally be somewhat semiarid, neither too wet—such that summer rains are still sufficient for regular forest growth despite being relatively drier than winter—or too dry—such that there's insufficient water for much vegetation at all. This particularly combination of conditions creates biomes where there is sufficient moisture for substantial vegetation but it doesn't coincide with optimal temperatures for growth, so plants have to rely on stored water, groundwater, or flowing water; or adapt to take advantage of sporadic growing conditions rather than a single concentrated growing season. This fosters a diverse evergrowth/deciduous mix, and often causes a "mosaic" landscape, with groundwater-dependent vegetation clustering along streams and in wet soil, while ridgelines remain more bare.

 

Koppen's Xs category actually does a fairly decent job of identifying these areas, but again is too reliant on a specific sampling scheme and assumed seasonal cycle. Trying to find an alternative, more flexible measure was difficult, but after testing dozens of potential parameters, the most reliable indicator I've found for Mediterranean biomes is what I'll call the growth supply factor, which compares precipitation in the growing season to total evaporation for the year. This essentially represents how much of the total water available for photosynthesis comes from rains that actually occur during the growing season; a low growth supply factor indicates that while a region has a significant source of water, it doesn't coincide well with growth and so plants have to rely largely on soil or stored water.

Thus, we'll break off Mediterranean zones as a separate group in the C class, as the wet winter/dry summer pattern seems to largely obfuscate the relative advantages of deciduous and evergrowth plants and instead favor a fine mosaic based on how local soil and topography affects water availability. Unlike the other zones, we'll also extend this group into semiarid zones, and in fact this accounts for most of Earth's Mediterranean biomes; moist submediterranean zones are relatively rare but a few notable patches make them worth including.

• Submediterranean (CM), with a low growth supply factor but high growing season aridity factor, indicating a summer drought but still enough summer rain for thick vegetation.
• Mediterranean (CAM), with a low growth supply factor and low growing season aridity factor, indicating a more substantial summer drought.

These zones will take priority over CT, CD, CE, and CA, but still give way to CF and A. 

On the other hand, Koppen's dry-winter zones don't have as clear a utility; though they do give some sense of the extent of the East Asian monsoon, they don't actually seem to consistently correspond to any particular biomes: especially in the cooler areas, these ares are distinguished mostly by their dry winters rather than wet summers, but vegetation will be mostly dormant in winter and not get much water out of frozen soil anyway, so a dry winter isn't a particular hazard and the vegetation in these areas ends up not much different from their European and North American counterparts, many of which actually have moderately drier summers than winters. 

 

Instead, I've found more utility from the evaporation ratio, comparing total evapotranspiration to total precipitation, essentially representing the portion of precipitated water that remains in the soil and evaporates in the same place; if the ratio is below 1, that shortfall indicates runoff in periods where the soil was already saturated with water, so a low ratio indicates at least part of the year where rains far exceeding potential evapotranspiration and cause soil saturation and significant runoff.

Thus for most of the zones I'm marking out a parallel set of pluvial zones:

• Pluvial (Xxp) has a low evaporation ratio, indicating periods of excessive rain.

This adds quite a bit of extra zones given how little area they end up collectively covering, but this neatly accounts for a range of notable biomes and climate types:

• For rainforest (Tr), a pluvial zone may seem redundant, but it indicates where rainfall is so heavy that soil nutrient leaching may become a serious issue, causing some patches of forest to give way to shrubby heathland (though exactly where this happens depends strongly on local soil geology and topography, so it's more an indication of regions at risk of soil depletion than a direct match to such areas).
• For other tropical and subtropical regions, it indicates particularly heavy monsoon rains, which may support denser forests than the categorization by aridity factor alone suggests.
• For temperate and boreal regions, it indicates temperate rainforests, which tend to be particularly lush and also somewhat more favorable to evergrowth/deciduous mixes.

Semiarid pluvial zones are quite rare on Earth, but worth including in case they help show very strong monsoons on other worlds. Just to keep the count of zones down a bit, though, I have excluded pluvial subtypes from:

• Arid (A) zones, which we expect to get rain mostly in sporadic storms anyway.
• Tundra (CF) zones and colder, where low evaporation, frozen soil, and significant ice melt make the evaporation ratio a poor indicator of wet conditions.
• Mediterranean (CM) zones, where the severity of the heaviest rains in winter doesn't particularly shift the overall nature of the climate and vegetation.

Perhaps I'm repeating the sins of Koppen by adding so many extra zones to account for one additional factor, but they seem to have more relevance to biomes and we can largely think of pluvial conditions as a modifier on other zones, with a simple color scheme to reflect this.

 

That about covers the range of land climates we expect to encounter on Earth, with a total of 38 zones (25 excluding the pluvial zones), which fits fairly well with more detailed counts of Earth's biomes. However, the challenge now is to try to extend this out to account for climates well outside Earth's norm, extrapolating as best we can from the ecological principles and tolerances we've established for life here.

Quasitropical and Dark Climates

To start out with, there's a few quick modifications we can make to account for climates that might have a similar range of temperatures as Earth but a different pattern of seasons or light coverage. On a tidal-locked planet or other world with weak seasons, for example, substantial areas might never have winters colder than the tolerance limits for tropical plants but still have generally milder temperatures than regularly experienced in Earth's tropics and so slower growth. There could also be worlds with warm poles or nightsides that maintain tropical temperatures but simply lack the light for photosynthesis and growth for part of the year.

To cover both of these cases, tropical climates will be split into two main subgroups:

• Eutropical (TU), with good growing conditions year-round.
• Quasitropical (TQ), with a period of growth interruption similar to temperate climates.

We could perhaps imagine various special cases with different patterns of seasonal light cover or temperature variation overlapping with different precipitation patterns, but I'll keep things simple here. Most of the moisture gradations will by mirrored across both groups, save that I'll exclude rainforests from quasitropical zones on the presumption that growth interruption from limited light or moderate temperatures affects forest dynamics in a similar way to the mild winters of subtropical zones. Low light may inhibit growth, but will also reduce evapotranspiration, so there's less need for a deciduous habit to limit maintenance costs so long as temperatures remain mild.

Areas with very little light will then transition to more marginal zones, in line with cold regions:

• Tropical Twilight (TF), with too little light for large vegetation. 
• Tropical Dark (TG), with too little light for almost any vegetation.

Much as in cold climates, TF will take priority over semiarid climates and TG over arid climates, as we expect them to have little vegetation regardless, and in low light evapotranspiration and aridity will be less of a concern anyway.

Hot Climates

Handling hotter and extremely seasonal climates may be trickier. Tolerance limits for maximum temperature are not often considered in climate classifications because few areas on Earth seem to hit any such limits, aside from perhaps some deserts where water availability is the overriding concern anyway. Some tundra plants may respond poorly to tropical temperatures, but we'd expect them to be displaced by other plant types long before temperatures rose that high anyway. The hottest sustained climates on Earth since the evolution of land plants have been about 10-15 °C hotter than today, but this relative warming was mostly concentrated towards the poles; temperatures may have spiked higher for brief periods during the End-Permian mass extinction, and there is some indication of widespread vegetation die-offs, but this could reflect shifting precipitation patterns, ecological shock from rapid climate change, or any number of other potential stressors at that time, and even if it was primarily temperature-driven this doesn't necessarily reflect the capacity of life to adapt to a sustained warm climate. Stable Earthlike worlds might be able to reach averages of over 40 or even 60 °C, and as we've seen even worlds with similar average temperatures to Earth might have extreme summers.

 

There is some literature exploring the absolute limits of temperature at which plants are irreversibly damaged even in moist conditions that we can go off of here:

• 35 °C: bleaching temperature of even heat-tolerant corals.
• 40 °C: tolerance limit for aquatic and undergrowth plants.
• 50 °C: tolerance limit for most plants with exposed leaves.
  
• 70 °C: tolerance limit for desert-adapted plants, and complex life generally.
• 100 °C: water boiling point at 1 atm; tolerance limit for any life on land.
• 120 °C: maximum observed growth temperature for microbes in high-pressure water.
• 150 °C: limit of stability for Earth-like biochemistry.
  
• 200 °C: maximum observed survivable temperature for microbes with brief exposure.
  
• 374 °C: boiling/critical point of water even at high pressure.
  

The actual temperature of the plant can depend on conditions like shading, surface moisture, and time of exposure, which aren't always clear from climate data, but we'll go by decent approximations anyway.

Even when not causing direct damage, high temperatures can also inhibit growth: Photosynthetic productivity tends to rise with temperature because it makes enzymes in the cell move and react faster, but at very high temperatures two issues arise:

• Transpiration rates become very high, such that even if there is sufficient water in the soil, plants may not be able to transport it to their leaves fast enough to keep them moist and continue producing sugars.
• Enzymes and other molecules break down, both slowing photosynthesis and causing various other inefficiencies and requiring energy expenditure for repair, to the point of offsetting energy gains from photosynthesis.
  

These can be compensated for to varying extents by different adaptations—such as C₄ photosynthesis, which more efficiently uses CO₂ inside leaves to reduce the need to open pores to intake more CO₂, which is when most water loss occurs—but only to an extent.

The result is that most plants tend to achieve optimal growth at around 30 °C, and then growth rates sharply decline with higher temperatures until reaching the compensation temperature, when energy loss to repair or inefficiencies outweighs gains from photosynthesis and the plant can no longer sustain growth. This is typically around 40 °C but potentially over 50 °C for heat-tolerant plants.

 

All this in mind, the safest approach may be to assume a degree of symmetry with cold climates. We will assume that tropical plants on a hot world can survive without significant damage up to a maximum temperature of around 50 °C, beyond which more extensive heat tolerance adaptations are necessary, forming a new climate group:

• Hot (group H), with periods of intense heat.

Sufficiently high temperatures may then inhibit growth even among heat-tolerant plants, creating a period of interrupted growth in summer similar to that in winter for cold climates, and so a parallel sequence of increasingly growth-restricted climates:

• Supertropical (HT), with only brief growth interruptions, allowing for evergrowth plants, named as the converse to subtropical zones.
• Swelter (HD), with longer growth interruptions potentially favoring a more deciduous lifestyle.
• Parch (HF), with a brief growing season to short for large vegetation.
• Hot Barren (HG), with a growing season too short for almost any vegetation.

You'll note that, much as in arid climates, there's no direct analogue to boreal climates here: Whereas plants in cold climates can keep relatively low maintenance costs through winter thanks to low transpiration and metabolic dormancy, plants in hot climates with long summers must face high transpiration and heat stress throughout, and so retaining soft tissue through summer would likely never be competitive with deciduous plants even given very short growing periods. There might still be some niche for highly conservative growth patterns in hot climates, but because it'll be based on different dynamics of profits and costs, I can't predict where exactly that will be, and we probably shouldn't put too much faith in my estimates for hot growing temperature limits anyway.

These zones can then be subdivided based on additional thermal tolerances, in particular the 70 °C and 100 °C boundaries:

• Hot Summer (Hxa) climates have summers within the range of survival for complex life on Earth.
• Torrid (Hxb) climates have summers beyond Earth's complex life tolerances, requiring extreme adaptations to survive.
• Boiling (Hxc) climates reach temperatures above boiling, fatal for all exposed life incapable of surviving total desiccation.

Much as in cold climates, we'll keep the zone count down by trimming some of the potential combinations:

• Much like subtropical climates, supertropical (HT) climates will be restricted to hot summers.
• Swelter (HD) and Parch (HF) will both be split into all 3 categories, which may be expressing a lot of optimism in the capacity for complex life to adapt to extreme heat beyond the boiling point, but helps to cover all our bases.
• Hot Barren (HG) will remain undivided like the other barren categories; it's unlikely to have only hot summers anyway.
  

Even if most of these assumed biome categories are broadly accurate for alien worlds with hot summers, there's every chance that the specific temperature thresholds won't apply consistently; other than the boiling point, which few habitable climates are likely to reach and could vary considerably with atmospheric pressure and altitude anyway, these thresholds aren't linked to substantial physical barriers like the freezing point; it's entirely possible that slightly different biochemistry or evolutionary history might allow some alien life to tolerate much greater (or less) heat or evapotranspiration, though because many chemical processes tend to scale roughly exponentially with temperature, there's at least some reason to hope the shifts won't be too great.

The same pattern of semiarid, Mediterranean, and pluvial zones can also be carried over, with 3 thermal categories for all because boiling summers are likely to more fundamentally affect all types of hot biomes than percontinental winters affect non-Boreal cold biomes. The interpretation of "Mediterranean" is a bit peculiar here, though, because in H climates the growing period is in the cooler part of the year, so our definition of Mediterranean climates based on growing season precipitation favors them in dry-winter climates here rather than the dry-summer climates of cold Mediterranean regions. Still, in testing this system against some of the high-temperature climate models I've made in the past, I have found that this definition does correspond to restricted areas with a specific pattern of particularly harsh winter or spring droughts with heavy summer rains, which might indeed favor Mediterranean-like biomes, so I'll include this "paramediterranean" category.

 

We can also add a couple of hot arid zones, but in parallel to cold deserts, these will have a higher temperature threshold—torrid rather than hot summers—and here I will skip a separate boiling category, on the continued presumption that desert life is going to be inherently more tolerant to temperature extremes and will perhaps tend to have fewer exposed wet, soft tissues exposed to high exterior temperatures and so will be less affected by boiling temperatures.

Extraseasonal Climates

Finally, of course, a region could both have extremely hot summers and extremely cold winters. There would likely be a separate set of adaptations required to tolerate both extremes in the same year, marking out a distinct set of biomes, forming another distinct group:

• Extraseasonal (group E), with both periods of frost and periods of intense heat.
  

We will, again, presume some overall symmetry with our cold climates, though with the extra wrinkle here that these climates may face growth interruptions in both summer and winter, with 2 growing seasons in between. For both growth interruptions and growing season, I will presume that the longest is always the most relevant to questions of biome transitions. In any case, this implies our usual gradient of groups:

• Extratropical (ET), with no substantial growth interruption in either summer or winter.
• Extracontinental (ED), with some period of extended growth interruption.
• Pulse (EF), with only brief periods of growth between seasonal extremes, too short for large vegetation.
• Extraseasonal Barren (EG), with growing conditions insufficient for almost any vegetation, due either to temperature or light restrictions.
  

There are many possible combinations of different tolerances or growing seasons we could consider here, but we're rapidly running short of visually distinct colors, so I'll keep this category small: there will be no "extraboreal" category, and we'll use only two thermal tolerance categories:

• Superseasonal (Exa), with hot summers and cool winters.
• Hyperseasonal (Exb), with either torrid summers or cold winters, requiring more substantial adaptations to extreme temperature tolerance.

Thus, there's some indication of more severe seasons, even if it may not be obvious at a glance which season is the worst; hopefully in most cases this will be reasonably clear from context. As per usual extratropical (ET) will be restricted to superseasonal climates and we won't bother breaking up extraseasonal barren (EG).

 

We can then carry over semiarid, extramediterranean, and pluvial zones as per usual, and for arid zones we'll add an extraseasonal category but with the particular requirement for both torrid summers and cold winters (because arid zones with just one are already covered by the existing categories).

Oceans

Though the main focus of this bioclimate system is on land areas, I do want to give some consideration for oceans, so I might as well give them their own designations

• Ocean (group O)
  

Ocean biomes are heavily influenced by factors such as seafloor depth, deep water upwelling, and nutrient runoff from land areas that aren't well represented by an atmosphere-only climate model like ExoPlaSim. We might be able to guess at some of these factors in various ways, but to keep things simple we'll stick to what we can determine based directly on climate data, which ultimately comes down largely just to sea surface temperatures and ice cover.

 

First off, of course, permanent and seasonal ice cover are fairly straightforward data points that may influence both aquatic and semiaquatic life:

• Permanent Ocean Ice (Ofi), with permanent sea ice cover.
  
• Seasonal Ocean Ice (Ofd), with seasonal sea ice cover.
  

Within ice-free waters, open oceans are largely uniform across climate regions, but coastal waters are clearly divided into a tropical region where coral reefs dominate and a cooler region where kelp forests and shellfish beds dominate. Tropical corals can tolerate sea surface temperatures above 18 °C, though I've never found a clear answer for why, perhaps something to do with water chemistry and nutrient availability. For what it's worth, the extent of glass sponge reefs in the Jurassic seems to have reached to about the same temperature boundary, as well as our geological record can attest, so this doesn't seem to be a trait specific to modern Earth corals.

 

Heat-tolerant corals can survive up to 35 °C, with the mechanism there more clearly being damage to their symbiotic algae, but corals or similar analogues seem to have been common in tropical waters during substantially periods in the past (and ExoPlaSim again tends to err high on equatorial temperatures) so I'll be more permissive and set the high end of tropical ocean temperatures at 40 °C. To cover our bases, I'll add a zone where temperatures reach above 60 °C, about the maximum we've seen complex life tolerate for extended periods, and a zone where temperatures vary between below 18 °C and above 40 °C (the former taking precedence over the latter):

• Temperate Ocean (Oc), with sea surface temperatures dropping below 18 °C but no sea ice.
• Tropical Ocean (Ot), with sea surface temperatures remaining between 18 and 40 °C.
• Hot Ocean (Oh), with sea surface temperatures reaching above 40 °C.
  
• Torrid Ocean (Or), with sea surface temperatures reaching above 60 °C.
  
• Extraseasonal Ocean (Oe), with sea surface temperatures reaching both below 18 °C and above 40 °C, but not above 60 °C.

In colder waters, you might expect ocean vegetation to show some progression towards more growth-limited vegetation in colder climates just as on land, but in practice kelp forests and open ocean plankton activity reach essentially right up to the edges of the permanent ice sheets (or their former distribution, anyway), with no clear zonation by latitude. Perhaps because frost damage and evapotranspiration aren't issues, maintenance costs through winter simply aren't a concern, especially compared with the strong influence of water nutrient distribution. Still, photosynthetic life will need some light, so for tidal-locked worlds or similar we can mark out permanently dark regions, which will take priority over our temperature-determined zones (save that I won't bother distinguishing dark and lighted permanent ocean ice).

• Dark Seasonal Ocean Ice (Ofg), with seasonal ice cover and too little ice-free light exposure for growth.
  
• Dark Ocean (Og), with no ice and too little light for growth.
  

That completes the system, bringing us to a grand total of 95 land zones (65 without the pluvials) and 9 ocean zones. That's perhaps a bit more sprawling than I originally intended (and trying to give all of these recognizable, distinct colors was about as much of an ordeal as you might imagine), but we're unlikely to see all these zones on a single world, so that perhaps mitigates the complexity a little.

 

Climate Parameters

Now that we've got a conceptual sense of how we want to divide up our climate zones, it's time to dig a bit more into the technical aspects of how we can define them based on specific data. For a lot of the parameters we want to use, availability and exact measurements may vary between sources, so I'll mostly be focusing on how we can use ExoPlaSim data, which we'll presumably be using the most, but I'll also discuss some potential alternative approaches when working with more restricted data sources, to the extent that in principle the whole system can be approximated from just monthly average temperature and precipitation, the same as Koppen-Geiger.

 

In general, I've tried to pick parameters with a mind to flexibility and potential sampling issues. I've mentioned before that when working with different year lengths, it's hard to know whether the months used as sampling periods in Koppen-Geiger should best be interpreted as corresponding more to a portion of the year or to an absolute length of time; some definitions like the length of summer for subarctic zones should probably be based on absolute lengths of time, but others based on extremes of temperature should probably be scaled to the length of the seasonal cycle such that short summer heat waves or winter chills aren't averaged out in too large a sample. We've also seen a number of cases where the Xs/Xw definitions become unreliable because they assume a simple seasonal cycle with a single summer and winter or single wet and dry season (there's sampling length issues there too, because they sample the wettest and driest months of each season; too short a sample could be biased by a single heavy rainstorm or dry spell, but too long a sample may miss real seasonal variation). I don't think I can obviate the potential for sampling biases or perfectly account for every seasonal cycle, but—aside from with temperature extremes where the length of these conditions seems unimportant—I've tried to pick parameters that are a bit more continuous, looking at variation in temperature or precipitation across the year rather than trying to pick a single month or specific number of months as representative of the entire cycle.

 

That said, though the motivation for the use of particular parameters has been based on theoretical considerations, the choice of specific definitions and thresholds will have to be empirically determined to best match actual biome distributions, and there's even a degree to which definitions may have to be tuned to specific data sources, which I'll make a note of in each case.

There are a number of assumptions here that might be questionable, especially where we're extrapolating to non-Earth-like climates, but because I'm trying to avoid the use of indirect proxies, it's fairly easy to adjust these assumptions if we like, and the koppenpasta script is also designed to  make that fairly easy (the thresholds are all listed at the front of the Pasta_Alg function, save for the GDD parameters which are in Biome_Param); if you're more optimistic or pessimistic about the ability for vegetation to tolerate high temperatures, for example, you can directly tweak the thresholds used, rather than having to work through any odd indirect measurements.

Minimum Temperature (MinT)

Based on previous studies, the critical temperature thresholds in group C should align with absolute minimum temperatures (MinT) of 0, -15, and -60 °C. However, actual measurements of minimum temperature can vary considerably depending on the exact methodology. For example, the Earth climate data I've been using for reference seems to show some averaged daily low for each month, rather than an absolute minimum. For ExoPlaSim, my best guess based on what I know of the internal memory management is that data is collated roughly every 5 days in-model, with the maximum and minimum temperatures being the extremes within those periods, but then when data is collated into months for output, the monthly values for max and min temperature are the average for the results of these 5-day periods within each month. Thus, in most cases they are also best regarded as representing typical daily temperature variation rather than absolute extremes for the whole month. One can force ExoPlaSim to output data in shorter intervals and then collate that into months externally, but that's generally inconvenient in practice. Even if we can properly measure the exact absolute minimum temperature in a single year, occasional cold snaps every, say, 5 years could still impact biome boundaries.

Thus, the thresholds of minimum temperature in these datasets that best correspond with actual biome boundaries are generally higher than we theoretically expect them to be, but we can accept that as a degree of inherent error to the approach or perhaps even some error in our theory. At any rate, I've decided on a set of tuned thresholds to be used for each data source (which we can also use for the Woodward and Prentice et al. models). I've also found monthly average temperature minimums that seem to correlate to these absolute minimum thresholds, for cases where that's the only data available.

Winter Type	ExoPlaSim MinT	TerraClimate MinT	Average Temperature
Mild	>10 °C	>10 °C	>17 °C
Cool	-10 – 10 °C	-4 – 10 °C	0 – 17 °C
Cold	-40 – -10 °C	-35 – -4 °C	-30 – 0 °C
Frigid	<-40 °C	<-35 °C	<-30 °C

To avoid potential confusion, for defining zones I'll use the names specified above, saying e.g. "cool winters" to indicate a minimum temperature that should be in the theoretical range of -15 to 0 °C, but can be tuned as appropriate for the data source. Thus, these categories can be applied for various zones:

• Mild winters for tropical (T) and hot (H) zones
• Cool winters for oceanic (Cxa) and superseasonal (Exa) zones
• Cold winters for continental (Cxb) and potentially hyperseasonal (Exb) zones
• Frigid winters for percontinental (Cxc) zones

One additional quirk for ExoPlaSim data is that it samples maximum and minimum temperatures from the surface temperature, not the 2-meter air temperature usually prefered for climate classification, so by default these measures are all adjusted based on the difference between monthly average surface temperature and 2-meter air temperature, but this can be toggled with the temp_adjust_ts setting.

For oceans, we shouldn't expect much daily or even monthly temperature variation, so we can judge zones just by monthly averages of surface temperature.

Maximum Temperature (MaxT)

Critical temperature thresholds in group T are set mostly by the absolute maximum temperature (MaxT), similar to the above. There is, however, not much good tuning data to go off of. We might expect that, as with minimum temperature, the measured maximum temperature might fall short of the actual absolute maximum, but ExoPlaSim seems to err a bit high in its maximum temperatures and our estimated thresholds are imprecise anyway, so we'll just go with our theoretical thresholds of 50 °C for hot summers, hostile to most plant life, and 70 °C for torrid summers, hostile to even heat-tolerant life. For cases where we have only monthly average temperature data, setting these thresholds 10 °C lower seems like a decent approximation based on looking through previous ExoPlaSim experiments I've done with high temperatures. Much as with minimum temperatures, we can use more generic names for summer temperatures to avoid confusion:

 

Summer Type	ExoPlaSim MaxT	Average Temperature
Warm	< 50 °C	< 40 °C
Hot	50 – 70 °C	40 – 60 °C
Torrid	70 – 100 °C	60 – 90 °C
Boiling	> 100 °C (or boil point)	> 90 °C (or boil point)

The boiling point is around 100 °C at sea level on Earth, but can vary significantly depending on atmospheric pressure, which varies with altitude and potentially between planets. Thus, rather than assigning a single boiling temperature, if data on surface pressure is available we can use the temperature and the Antoine equation to estimate a boiling pressure, below which water will boil. Sufficiently low pressures can significantly lower the boiling point, e.g. to about 45 °C at 0.1 atm, so on low-pressure worlds some areas may skip directly from warm to boiling summers. In such cases I'll also take any definitions requiring hot or torrid summers, e.g. Axh or Exb zones, to also be met if temperatures exceed the local boiling point, regardless of how low it is.

 

The above temperature thresholds can still be used if no pressure data is available, and this is toggled by the pas_boil_pres option.

• Warm summers for tropical (T) and cold (C) zones
• Hot summers for hot-summer (Hxa) and superseasonal (Exa) zones
• Torrid summers for torrid (Hxb) and potentially hyperseasonal (Exb) zones
• Boiling summers for boiling (Hxc) zones.
  

Growing Season (GDD, GDDz, GDDl)

For measuring growing seasons, I'll again take inspiration from the Prentice et al. model and rely on growing degree-days (GDD), where every degree Celsius of the daily average temperature above a certain base temperature is counted as one GDD, and this is summed across the whole year (with no negative values counted for days averaging below the base temperature). The idea is that plants will grow faster at higher temperatures, so a short, hot growing season will support about as much growth as a long, mild growing season, with the critical factor for biomes being the total accumulated amount of growth possible in a single season. The growth-temperature relation for most plants tends to be a somewhat more complex curve, but this linear approximation works reasonably well when considering whole biomes, averaging across different plants with different optimal growth conditions.

 

GDD is usually measured on a daily basis, but of course ExoPlaSim and many similar data sources only give us monthly averages for temperature. There are ways to interpolate reasonable estimates of daily average temperature from monthly averages, but they're somewhat computationally expensive, so to keep things simple I'll measure growing degree-months (GDM) based on the average 2-meter air temperature for each month, and then assume that 1 GDM corresponds to 30 GDD. This generally isn't exactly true, but it'll be fairly close, and if we're tuning our definitions based on measures of GDM then it shouldn't much matter.

The Prentice et al. model used a base temperature of 5 °C for most of its GDD measures, so I'll do the same. However, we'll also have to handle declining growth at high temperatures. Different plants have different optimal growth temperatures, but again when considering whole biomes and presuming that any particular region will already be occupied by the plants best adapted for that climate, peak growth rates tend to be achieved at around 25-30 °C. Many measures of GDD will account for this by capping GDD counts at 30 °C, counting no additional GDD for temperatures higher than that; i.e., with a base temperature of 5 °C, any day with a temperature above 30 °C will always have exactly 25 GDD. But even higher temperatures of 35-50 °C can cause growth rates to decline or even stop, which generally isn't accounted for.

 

A typical temperature-GDD curve

 

To try to keep things simple, I will start with a base of 5 °C and then assume that growth plateaus at 25 °C; i.e., growth rates are capped at 20 GDD/day. All temperatures between 25 and 35 °C will be counted as a uniform 20 GDD, and then GDD will be taken to decline by 2 per °C above 35 °C, dropping to no growth above 45 °C. In reality growth rates at high temperatures could depend greatly on atmospheric CO₂ levels or the evolution of particular photosynthetic pathways, but this will do for a simple approach.

 

The Pasta Bioclimate GDD curve

 

Though there are some ambiguous hemiboreal forests, the temperate-boreal transition generally falls at about 1300 GDD. The extent of tundra varies a bit between sources, corresponding to either 350 or 500 GDD depending largely on whether it is considered to include a strip of sparse trees on the edges of the boreal forest, which I've also sometimes seen referred to as its own distinct biome, "taiga savanna". I've opted to go with the lower number here, if only because it somewhat compensates for ExoPlaSim's tendency to underestimate polar temperatures and so overstate the extent of tundra.

• Below 1300 GDD for boreal (CE)
• Below 350 GDD for tundra (CF)
• The same 350 GDD boundary will be used for HF, TF, and EF as well.
  

But there are various special cases to account for. First off, the Prentice et al. model also included a second measure of GDD with a base value of 0 °C, for dividing tundra from polar desert, reflecting how tundra plants may be able to persist on meager growth performed at very low temperatures where more warm-adapted plants have largely gone dormant. I'll use a similar growing degree-day-zero (GDDz) measure here for much the same purpose, but extend it out a bit to use for all the barren (XG) boundaries: GDDz will be capped at 20 °C, for the same 20 GDDz/day maximum as regular GDD, and then taken to begin declining at 35 °C at 1 GDDZ/°C, dropping to no growth at 55 °C, which thus represents my broad estimate for the maximum growth temperature for even very heat-tolerant vegetation, based on that being roughly the maximum temperature for which we see sustained activity by complex life on Earth.

 

The new GDDz curve, in orange, compared to the old GDD curve, in blue.

This single factor can then be used for all groups:

• 50 GDDz for dividing XF from XG in all groups
  

This ensures that, for example, at the terminator of a tidal-locked world we should see a consistent transition to barren biomes across different temperature regimes.

 

But that brings up our next issue: photosynthetic vegetation needs light to grow as well as clement temperatures. On Earth, this isn't as important as you might think: The rate of photosynthesis is bottlenecked by the chemical process of carbon fixation, so most plants receive more light than they can actually use. Towards the poles, winters can get fairly dim, but it generally gets too cold for photosynthesis before it gets too dark. But, of course, we can easily imagine other worlds—or just periods in Earth's past—with substantial areas that remain warm through periods of darkness.

Based on other models, the best approach here is likely to calculate light-limited growing degree-days (GDDl) based on light levels, and then take the actual GDD in each month as the minimum of GDDl and the temperature-determined GDD (thus the resulting GDD count is determined by the strictest limiting factor at any time, not any combination of the two, which seems to match the best models of photosynthesis). The relationship between light levels and photosynthetic rates is somewhat complicated but we can make some rough approximations: the instantaneous rate of photosynthesis rises about linearly with light level up to about 250 W/m² of sunlight, beyond which there are diminishing returns as the process becomes bottlenecked by the rate of chemical reactions rather than energy input from light, with around 500 W/m² being about the highest level with any benefit. When considering real conditions where light varies throughout the day and exposure of individual leaves to light varies, the dynamics are more complex but it seems that there is coincidentally a similar relationship where overall productivity rises about linearly with daily average light radiation up to about 200 W/m², beyond which there are sharply diminishing returns; this conveniently means that we may not have to worry too much about the difference between rotating and tidal-locked planets. Below about 20 W/m², typical large plants may not be able to produce enough from photosynthesis to cover their respiration and maintenance costs, but some marginal life can likely manage even with very low light levels.

 

Typical relationship of net phosynthetic production to instantaneous light irradiation (left, Formighieri 2015) and daily average irradiation (right, Rosati and Dejong 2003); 1 W/m² of sunlight corresponds to about 2 μmol/m²/s or 0.17 mol/m²/d of photosynthetically active photons

 

Thus, for simplicity's sake we'll take GDDl to accumulate at a rate of 1 GDDl/day per 10 W/m² of average daily radiation, up to a maximum of 20 GDDl, above a baseline of 20 W/m² for limiting GDD and 0 W/m² for limiting GDDz. This presumes an Earthlike relationship between total light irradiation and photosynthetically useful radiation, so I'll add the gdd_par_ratio setting to adjust the ratio of photosynthetic to total radiation (which is about 0.5 for light from our sun) should anyone want to speculate about the potential difference with the light spectra from different stars.

 

The GDDlz (orange) and GDDl (blue) curves.

 

For oceans, dark zones will be determined from GDDl directly (with the 0 W/m² baseline), with no reference to temperature.

Finally, areas on Earth with cold winters consistently have one winter each year, so if we count total GDD for the year we can assume it comes in one long growing season. But on other worlds we could have more complicated seasonal cycles with multiple potential growing seasons. Based on our theoretical understanding of leaf growth patterns discussed before, generally speaking the longest continuous growing season should be the most important parameter, because it determines the maximum amount of stored energy and resources that can be accrued before a period of interruption where some resources must be spent either losing and then growing new leaves or maintaining leaves through a nonproductive period. Thus, GDD will be taken to accumulate until average monthly temperatures fall below 5 °C (or rise above 45 °C, or light falls below 20 W/m²), with the largest accumulation before such an interruption taken as the total. If average temperatures remain above 5 °C in all months, the GDD can be regarded as effectively infinite, because vegetation can just continuously grow across years without having to accrue resources to survive substantial interruptions. The same principles apply to GDDz and GDDl for oceans, with their respective thresholds.

Which then leads us to the question of overall year length; I've speculated before that longer or shorter years with correspondingly different growing season lengths might be compensated for with different winter lengths to store resources for, but our theoretical considerations so far seem to indicate this is unlikely, as longer winters don't impose a particularly higher maintenance cost. This may not hold for very long winters beyond any seen on Earth, or for growth interruptions due to high heat or low light, but that's largely just why I've excluded boreal analogues from such cases; deciduous plants or annuals might be more capable of going dormant in such conditions. At any rate, it does seem that growing seasons and growing interruptions should best be counted in terms of absolute length in time, rather than relative to the year. ExoPlaSim data doesn't contain detailed time information, so instead I've added a prompt in koppenpasta to adjust GDD counts based on a month length modifier, which represents the length of each month in the dataset relative to a typical 30-day, 720-hour, Earth month and is then used in the conversion of GDM to GDD (focusing on months because koppenpasta counts growth parameters month-by-month and then sums these without presuming any particular number of months or length relative to the year); note specifically that despite "month" here refers to the recording periods in the input data (though after any binning of months is applied), rather than having any relation to a particular planet's moon or calendar.

Finally, different environmental conditions or evolutionary history might allow for substantially different growth rates on different worlds, even without having to assume much difference in biochemistry. The chemical bottleneck on carbon fixation rates I mentioned earlier appears to be largely a consequence of Earth's high ratio of atmospheric O₂ to CO₂, making it difficult to specifically incorporate CO₂ to drive the reaction forwards rather than accidentally using O₂ and driving the reaction backwards; a world with different atmospheric conditions might thus be able to achieve higher (or lower) growth rates, at least in those marginal biomes where growth is restricted by growing season length rather than nutrient or water restrictions. To account for this, koppenpasta includes an additional option to apply a flat modifier to all GDD counts (which stacks with the length multiplier).

Growth Interruption (GInt)

For this parameter, I'm moving beyond my examples in the literature; various models do attempt to predict patterns of evergreen and deciduous trees, but generally by modelling their growth rates in detail and comparing their competitive advantages. Deciduous trees tend to use specific temperatures to trigger leaf loss, but these are just convenient indicators of the approach of winter and do not represent the root cause for deciduous behavior or the more general subtropical-temperate transition, and in general this boundary doesn't correlate well to any particular minimum temperature threshold.

 

After testing a few different parameters, I've found that the boundary does correlate well to what I'll call the growth interruption factor (GInt), which works somewhat like GDD in reverse: for each day with an average temperature below 15 °C, 1 GInt accumulates for every degree below 15 °C, to a maximum of 15 GInt/day at 0 °C (though much like GDD, I'll actually count GInt by monthly temperature and multiply by the length of month). This means that, much like GDD flexibly accounts for both long, mild summers and short, hot summers, GInt accounts for both long, cool winters and short, colder winters; in essence it's accounting for the accumulation of cold, low-growth conditions that increase stress on subtropical plants, rather than defining winter by any simple threshold.

• 1250 GInt for dividing XT from XD (or XE or XF) in all groups
• The same 1250 GInt will divide TU from TQ
  

At high temperatures, I'll consider GInt to begin accumulating above 40 °C, to a maximum at 55 °C; this conveniently means that rather than having to count GInt separately, I can simply count how much GDDz falls short of 15 GDDz/day. I'll presume the same GInt thresholds divides the hypothetical supertropical climate, with tropical-like evergrowth vegetation, from hotter climates that must adapt to prolonged periods too hot for growth. GInt is also counted after GDDz is limited by light levels, so dark periods will be counted as growth interruptions as well.

 

The temperature-GInt curve

 

GInt will also be affected by the GDD month length and productivity modifiers; longer months will accumulate more GInt just as they accumulate more GDD and GDDz. But a greater productivity modifier reduces GInt, on the presumption that if plants are innately more productive, they can tolerate lower temperatures before their growth is seriously interrupted. So, for example, if productivity is doubled, then temperatures have to fall below 7.5 °C (or rise above 47.5 °C) before GInt starts accumulating.

 

Also like GDD, if there are multiple periods of growth interruption divided by months above 15 °C (or whatever the appropriate threshold is), the GInt total will be counted as the longest accumulation of GInt; and if all months are below 15 °C, GInt will be taken as effectively infinite.

 

Note the implication that months between 5 and 15 °C accumulate both GDD and GInt, and a climate with all months in this range can have infinite counts of both. This is fine as they're marking separate biome boundaries caused by different ecological dynamics: subtropical planets may indeed still have some growth at moderate temperatures but temperate plants more specifically adapted to such conditions will likely have a competitive advantage, and a climate of constant mild temperatures may not have deciduous vegetation but still have a distinct set of planet different from subtropical and boreal climates.

Aridity (Ar, GAr)

For defining semiarid and arid biomes, I'll again be following Prentice et al. in using an aridity factor (Ar) defined as the ratio of total annual actual evapotranspiration (AET) to potential evapotranspiration (PET). Note the quirk that an arid climate actually has a very low aridity factor, but this is in line with how "aridity" factors and indices are commonly defined in research.

 

We can divide up our various levels of aridity thusly:

• Above 0.9 Ar for tropical and quasitropical rainforest (TXr)
• Above 0.75 Ar for seasonal forest (Xf)
• Below 0.2 Ar for arid (A)
• Below 0.06 Ar for hyperarid desert (Ah)

You'll note, of course, the gap in the middle there. As mentioned, dividing moist and semiarid zones is best done by measuring aridity specifically in the growing season, reflecting how the degree of growth possible in such climates depends on how well precipitation and clement growth temperatures coincide, to a greater extend than for very wet or dry climates. The boundary is thus defined by growth aridity (GAr), which, rather than comparing an equal average of AET and PET, compares averages weighted by the GDD (i.e. the value of AET or PET in each month is multiplied by the GDD count in that month, and then the resulting total for the year is divided by total GDD; for simplicity's sake this is done with the raw total GDD count without considering interruptions or multiple growth seasons); thus, the conditions in months with more growth have a greater effect on the growth aridity, and months without any growth aren't considered at all.

• Below 0.5 GAr for semiarid (XA)

Because tundra and other XF zones take priority over XA (thus XA can only occur where there is at least 350 GDD), there should always be some significant growing season to use for counting this parameter where it is relevant. The semiarid definition has priority over moister zones (an area with below 0.5 GAr cannot be moist regardless of how high Ar is) and the arid definition takes priority over semiarid (an area with below 0.2 Ar is arid regardless of how high GAr is).

I debated for some time if using total annual ratios for all zones was appropriate, as it generally seems that some biomes like rainforests are bounded by the severity of seasonal droughts moreso than year-round moisture conditions, but any measures I tried for monthly extremes of aridity or accumulated drought conditions proved to be poorer fits for biome boundaries, and didn't work terribly well in cooler climates where PET can be very low in winter and so small variations in AET can drastically change the monthly aridity factors. Using annual totals helps mitigate the influence of such oddities or biases based on sampling periods, and the use of AET in particular helps make this measure a more reliable indicator of the consistency of water availability compared to similar measures of precipitation/PET, which can be biased by heavy seasonal rains in a way that can obscure the dry season. A substantial dry season will always clearly reduce AET/PET regardless of how heavy rains are in the wet season, and conversely brief seasonal rains will raise the ratio, but only so far for a short burst of heavy rains in an otherwise dry climate (AET is also less sensitive to sampling periods, because evapotranspiration rates are buffered by soil water storage so a short sample of a moist period that happens to miss any heavy rain will still show high AET).

 

The main potential confounding factor is year length: a given period of drought will have less effect on the annual AET/PET ratio for longer years, even though it might have a similar effect on biome composition. But so far I haven't come up with a good measure to account for that, and at any rate we can probably expect longer years will usually come with accordingly longer seasonal variations.

PET can be estimated a number of ways, a few of which I've implemented in koppenpasta:

• The ASCE Penman-Monteith method is a commonly used standard approach estimating PET from average temperature, daily maximum and minimum temperatures, net surface radiation (absorbed sunlight - emitted heat), relative humidity, atmospheric pressure, near-surface wind speed, soil temperature, and vegetation density, all of which can be directly taken or estimated from ExoPlaSim output data; but should it be necessary, this method can be run with just average temperature, net radiation, and relative humidity, with all other values filled in with typical Earth averages.
• The Hargreaves method is a much simpler equation using just average temperature, net radiation, and optionally daily maximum and minimum temperature.
• Finally, a very simple Koppen-alike method estimates PET based just on monthly average temperature, as 4.5 mm/month per °C above freezing, based on Koppen's 20mm/year/°C threshold for aridity and a bit of tuning to best replicate the spread of climate zones on Earth; the results are far from perfect but they'll do as a backup of last resort.

AET is provided by ExoPlaSim data and many similar models, but where such information is lacking we can try to estimate it based on precipitation and PET with a simple model of soil moisture:

• In each month, precipitation is compared to PET
• If precipitation exceeds PET the excess is added to soil moisture, up to a maximum soil capacity equal to 50 cm of rain (in reality this varies considerably but this will do as an average); any extra water past that is presumed to be lost as runoff.
• Evaporation is then equal to PET
• If precipitation falls short of PET, the deficit is removed from soil moisture.
• Drier soil causes slower evaporation, so once soil moisture drops below 25 cm rain-equivalent, the total evaporation from soil is limited to the minimum of either the total remaining soil moisture or [ (PET - precipitation) * (soil moisture / 25 cm) ].
• Evaporation is then precipitation plus soil evaporation.
  
• Some soil moisture may carry across years, of course, so this algorithm runs through the year until there's no more than 1 cm difference in soil moisture between the start and end of the year.

This is essentially how ExoPlaSim determines evaporation rates internally, the main difference being that the model calculates evaporation every timestep while we're using monthly averages; so it'll be less accurate (particularly with regards to soil evaporation rates) but will do as an estimate.

 

Even with models that do provide evaporation data, this algorithm can be useful for interpolation; ocean cells should of course always have enough moisture to satisfy PET and so have an aridity factor of 1, but this may be unrealistic for any islands too small to have appeared at the model resolution but will appear in the interpolated map; so this routine can be used on ocean cells to estimate the appropriate evaporation that should occur in any land areas they contain, and that value used for interpolation; this is on by default in koppenpasta but can be toggled with the evap_estimate_sea option.

Growth Supply (GrS)

This is the parameter I've chosen to define Mediterranean zones, and I'm not exaggerating when I say I tested dozens of alternative approaches before I arrived at this one. Growth precipitation (Gpr) is a GDD-weighted average of precipitation, like those used for GAr, which here is compared to the unweighted average of actual evapotranspiration (AET) to determine the growth supply (GrS). The idea is that a low growth supply ratio indicates that rains during the growth season only supply a small portion of the total evapotranspiration, thus plants must rely mostly on water stored in the soil or their bodies through the growing season, creating the unique circumstances for Mediterranean biomes. This particular combination with a weighted and unweighted average may seem peculiar, and admittedly it's the most indirect parameter I'm using here, but it turned out to perform substantially better than all the alternatives:

• Comparing Gpr to a similarly GDD-weighted average of AET still somewhat identifies Mediterranean regions, but also tends to include many continental climates where AET sharply rises for a brief, hot period of midsummer, but remains low through the rest of the year; these regions can resemble Mediterranean areas in some ways, but because the period of high evaporation is brief and soil water can accumulate to high levels the rest of the year, water shortage isn't as much of a challenge in these areas.
• Comparing total yearly precipitation and AET is useful for pluvial zones as we'll discuss shortly, but not Mediterranean zones, as it can't show a seasonal deficit in precipitation, only a surplus.
• Comparing precipitation to PET (or AET to PET) tells us if a region is overall arid, but doesn't work well to show a seasonal mismatch in water supply and consumption.
• Comparing GAr to Ar (thus giving us some sense if the growing season is drier or wetter compared to the average) had some promise but I couldn't find any threshold that included patches of clear Mediterranean biomes in Africa and Australia while excluding large areas of clear temperate forests in Europe. 
• I attempted various approaches for measuring accumulated deficit of precipitation relative to AET and similar measures, under the assumption they would more directly indicate soil water conditions, but none proved to be as good a match to Earth's Mediterranean biomes.

For data on Earth, a maximum GrS threshold of 1.15 reliably identifies Mediterranean regions, implying that it is the norm for the average growing season precipitation rate to exceed the annual average evaporation; but, as has become clear through our explorations, ExoPlaSim has some innate bias towards Mediterranean-like climate patterns, so if we tune this model a bit to give the proper extent of Mediterranean regions in the baseline Earth tests, a threshold of 0.8 seems to work better (and conveniently avoids the potential for seasonless climates to be erroneously classified as Mediterranean). As with the temperature thresholds, I'll simply refer to "high" or "low" GrS in the zone definitions to allow for appropriate tuning.

• Low GrS for Mediterranean (XM) zones. 

Evaporation Ratio (Evr)

This is the parameter for defining pluvial zones, and in contrast to the above it's fairly straightforward: The evaporation ratio (Evr) is defined as the ratio of total annual actual evapotranspiration (AET) to precipitation (pr); i.e., it is the portion of precipitation that remains and ultimately evaporates rather than running off (in principal AET can also include external sources, but for pluvial zones this is likely to be a minor influence).

 

A low evaporation ratio indicates that there is at least part of the year where precipitation far exceeds PET, and this accounts for a substantial portion of annual precipitation, which implies saturated soil and some requirement to adapt to heavy rainfall and runoff and prolonged damp conditions. As mentioned, this conveniently covers a range of different special cases: in wet, warm climates, potential for soil degradation and tropical heathland; in drier climates, heavy monsoons that help support thicker forests; in colder climates, temperate rainforest or bogs that favor taller, thicker forests. All these biomes even seem to be associated with a similar evaporation ratio threshold:

• Below 0.4 Evr for hyperpluvial rainforest (TUrp)
• Below 0.45 Evr for all other pluvial zones (Xxp)

Ice Cover (MinIce)

Koppen-Geiger estimates ice cap extent by a hottest-month average temperature of 0 °C, which is generally decent at large scales, but a monthly average of 2-meter air temperature is not a perfect indication of frozen surfaces, and it's also possible for permanently subfreezing areas to lack ice if there's never enough precipitation for ice to accumulate. On Earth, this is limited to a few small valleys, but on a much drier world they could be more extensive, and the CG zone gives us the option to distinguish these areas from both proper ice sheets (CI) and slightly more hospitable tundra (CF).

 

ExoPlaSim does give us some more direct data on ice cover. The glacial model isn't entirely reliable, especially over the typically short runs of a few decades or centuries at most (large ice sheets may take many thousands of years to reach equilibrium between accumulation and ablation; one of these days I may investigate using a more dedicated ice sheet model), but if we combine that with snow depth data we can get a decent sense of which areas are likely to accumulate ice in the long term. Thus, I'll use ExoPlaSim's measure of snow depth (which includes glacial ice) and define ice sheets based on the minimum ice cover (MinIce).

• Above 10 cm MinIce for CI

For interpolation with temperature adjustment by topography, I'll have to make some awkward adjustments; I'll presume that any areas with persistent ice become none-iced if the absolute maximum temperature (unadjusted by the surface temp/2-meter air temp difference) rises above 0 °C, and I'll add ice where the maximum temperature drops below 0 C and there is at least some part of the year where ice accumulates to over 10 cm.

Sea ice cover is still defined based on fractional ice cover as reported by ExoPlaSim:

• Above 20% maximum sea ice cover for Of
• Above 80% minimum sea ice cover for Ofi 

But this also gives me an opportunity to fix a common interpolation issue, whereby sea ice doesn't appear in some small waterways too small to have appeared as sea cells at the model resolution and too far from any larger bodies of water to be covered by the "dummy ice" hack (which gives any coastal land cells the highest sea ice cover value present in any neighboring sea cells); whenever these areas have some snow cover, we can count them as having sea ice as well for interpolation purposes (for simplicity in the algorithm I'll take 1 cm of snow as corresponding to 10% sea ice cover, so 2 cm in at least 1 month is required for seasonal ice and over 8 cm is required in all months for permanent sea ice).

 

But as backup options, I've also added the pas_ice_def option, which can be set to ice for this ice-cover-based approach, ice_noadj for the same but without any of the above temperature-related adjustments when interpolating, tavg for the former definition by average monthly temperatures, or maxt for defining ice cover based on whether the maximum surface temperature ever surpasses 0 °C. Similarly, for sea zones the sea_ice_use_temp parameter can be used to toggle estimating sea ice as occuring where surface temperature is below -2 °C (which applies to both the standard and Pasta sea classifications).

 

You may note that, though I'm trying to choose the best data available from ExoPlaSim, again the whole system can be derived from just monthly averages of temperature and precipitation where necessary:

• Average Temperature
• Estimate MinT using average temp tunings
• Estimate MaxT using average temp tunings
• Take 100 °C as the boiling point (tuned as 90 °C average temp)
• Calculate GDD and GDDz (from corresponding GDM) from temperature
• Skip limitation by GDDl
• Calculate GInt from temperature
• Estimate MinIce based on 0 °C threshold for land, -2 °C threshold for sea
• Estimate PET as  4.5 mm/month/°C
• Average Precipitation
• Estimate AET from precipitation and PET with simple soil water model
• Calculate Ar from AET and PET
• Calculate GAr feom AET, PET, and GDD
• Calculate GrS from precipitation, AET, and GDD
• Calculate Evr from precipitation and AET

The Pasta Bioclimate System

Here at last is what the complete system looks like for Earth, using the same TerraClimate dataset I used for the maps in part I (which doesn't include data on oceans):

 

 

And here it is applied to our baseline ExoPlaSim model for Earth I made back in our first climate exploration. 

 

 

We'll assess how well this reflects Earth's biome distribution in the next post, but for now, for clarity and ease of reference here's the list of the resulting zones, their individual definitions, any good examples on Earth, and our general expectations for the biomes they should support. To avoid excessive repetition, here's a couple common rules across zones:

• All pluvial (Xxp) except for TUrp require below 0.45 Evr, and nonpluvial zones are presumed to have higher Evr.
• Peritropical (XT), moderate (XD), boreal (CE), and non-tropical semiarid (CA, HA, EA) zones all require high GrS, dividing them from Mediterranean (XM).
• All moist land zones, which excludes arid (A), semiarid (XA), marginal (XF), barren (XG), and ice (CI), require above 0.5 GAr.
• All non-arid land zones except for XG and CI require above 0.2 Ar
  
• All land zones except for A, XF, XG, and CI require above 350 GDD
• All land zones except for XG and CI require above 50 GDDz
• All land zones other than CI require above 10 cm MinIce.
• Ocean zones operate on an exclusive set of rules, but similarly all except for Ofi, Ofg, and Og require above 50 GDDl.
• All ocean zones except for Of require below 20% maximum sea ice cover.

Summer temperature, winter temperature, and Mediterranen GrS thresholds are tuned by data input, but here again are the values I've established so far:

 

Winter Type	ExoPlaSim MinT	TerraClimate MinT		Average Temperature
Mild	> 10 °C	> 10 °C		> 17 °C
Cool	-10 – 10 °C	-4 – 10 °C		0 – 17 °C
Cold	-40 – -10 °C	-35 – -4 °C		-30 – 0 °C
Frigid	< -40 °C	< -35 °C		< -30 °C
Summer Type	ExoPlaSim MaxT		Average Temperature
Warm	< 50 °C		< 40 °C
Hot	50 – 70 °C		40 – 60 °C
Torrid	70 – 100 °C		60 – 90 °C
Boiling	> 100 °C (or boil point)		> 90 °C (or boil point)
Growth Supply	ExoPlaSim GrS		TerraClimate GrS
Low	< 0.8		< 1.15
High	> 0.8		> 1.15

 

Here, now, is the actual list:

• Arid (A)
  -Below 0.2 Ar
• Hyperarid Desert (Ah)
  -Below 0.06 Ar
• Warm Desert (Aha)
  -Warm or hot summers and mild or cool winters
  -Sahara; Atacama; Sonoran Desert
  Too dry for most life, but may support burrowing or migratory fauna, and rare rainstorms may bring bursts of growth.
  
• Cold Desert (Ahc)
  -Cold or frigid winters and warm or hot summers
  -Tarim Basin; Atacama Plateau
  Substantial periods of frost with some potential for snow, reducing evapotranspiration stress but still providing too little moisture for most vegetation, and with significant risk of frost damage.
  
• Hot Desert (Ahh)
  -Torrid or boiling summers and mild or cool winters
  Extreme summer heat may further reduce habitability, but some life may survive through extreme dormancy.
• Hyperseasonal Desert (Ahe)
  -Cold or frigid winters and torrid or boiling summers
  Extreme seasonal temperature swings hazardous to most life, with periods of both frost and extreme heat; may only be visited by migratory fauna or require extreme dormancy.
  
• Semidesert (Ad)
  -Above 0.06 Ar
• Warm Semidesert (Ada)
  -Warm or hot summers and mild or cool winters
  -Australian Outback; Chihuauan Desert; Patagonia
  Occasional rain supports widespread grass, shrubs, and succulents, with accompanying fauna.
  
• Cold Semidesert (Adc)
  -Cold or frigid winters and warm or hot summers
  -Central Asian northern desert; Gobi; Great Basin
  Deep frosts provide additional hazard, but cool temperatures may reduce evapotranspiration, favoring seasonal vegetation.
• Hot Semidesert (Adh)
  -Torrid or boiling summers and mild or cool winters
  Extreme heat may limit life to ephemeral or migratory forms.
• Hyperseasonal Semidesert (Ade)
  -Cold or frigid winters and torrid or boiling summers
  Particular intense seasons favors ephemeral, migratory, or adaptable forms.
  
• Tropical (T)
  -Warm summers and mild winters
• Eutropical (TU)
  -Below 1250 GInt
• Tropical Rainforest (TUr)
  -Above 0.9 Ar
  -Amazon; central Congo; Singapore
  Consistent rain and growing temperatures allows for thick evergrowth forests with layered canopies.
  
• Hyperpluvial Tropical Rainforest (TUrp)
  -Below 0.4 Evr
  -Amazonas, Brazil; central Borneo
  Mostly typical rainforest, but extreme heavy rains give potential for significant soil degradation, forming more open tropical heathland or bogs.
  
• Tropical Forest (TUf)
  -0.75 - 0.9 Ar
  -Havanna, Cuba; Abidjan, Cote d'Ivoire; Hainan, China
  Moderate droughts or a slight water deficit, commonly forming a semideciduous forest.
  
• Tropical Monsoon Forest (TUfp)
  -Freetown, Sierra Leone; Kochi, India
  Heavy seasonal rains encourage a more rainforest-like structure, but there may still be a drought period.
• Tropical Moist Savanna (TUs)
  -Below 0.75 Ar
  -Serengeti, Tanzania; Cerrado, Brazil; Bangkok, Thailand
  Significant droughts but a wet growing season allows for lush savanna, woodland, or scrub depending on soil, fire, and grazing conditions.
  
• Tropical Moist Monsoon Savanna (TUsp)
  -Conakry, Guinea; Goa, India
  Heavier seasonal rains encourage thicker woodlands or full forests despite droughts.
• Tropical Dry Savanna (TUA)
  -Below 0.5 GAr
  -Sahel; Caatinga, Brazil; Deccan, India
  Drier conditions create more open savanna, woodland, or dry deciduous forest.
  
• Tropical Dry Monsoon Savanna (TUAp)
  -Mumbai, India
  Heavier seasonal rains encourage thicker woodlands or dry forests, though with a significant drought period.
• Quasitropical (TQ)
  -Above 1250 GInt
• Quasitropical Forest (TQf)
  -Above 0.75 Ar
  Heavily forested with tropical vegetation, but significant periods of low light or mild temperatures may cause periods of dormancy
• Quasitropical Monsoon Forest (TQfp)
  Heavy rains may encourage a more rainforest-like structure.
• Quasitropical Moist Savanna (TQs)
  -Below 0.75 Ar
  Significant droughts but a wet growing season; may have varied savanna and woodlands, but lower cold/dark period evapotranspiration may allow for lusher vegetation, though dormancy periods may limit tree growth.
• Quasitropical Moist Monsoon Savanna (TQsp)
  Heavier seasonal rains encourages thicker woodlands or forests, especially where they coincide with the growing season.
• Quasitropical Dry Savanna (TQA)
  -Below 0.5 GAr
  Drier conditions, but again likely to be less evapotranspiration in non-growing periods.
• Quasitropical Dry Monsoon Savanna (TQAp)
  Heavier seasonal rains encourage thicker woodlands.
• Tropical Twilight (TF)
  -Below 350 GDD
  Marginal regions near the terminator of tidal-locked worlds or other light-restricted circumstances. Might support grasses, shrubs, or other low vegetation optimized for limited growing conditions, but without the need for broad environmental tolerances.
  
• Tropical Barren (TG)
  -Below 50 GDDz
  Areas on the dark side of tidal-locked worlds still kept warm by global heat distribution; even with little precipitation, low evapotranspiration will likely maintain damp conditions, possibly helping to support some soil microbes, but likely nothing apparent above ground that could support large fauna.
  
• Cold (C)
  -Cool, cold, or frigid winters and warm summers
• Subtropical (CT)
  -Below 1250 GInt and cool winters
  
• Subtropical Forest (CTf)
  -Above 0.75 Ar
  -Louisiana; Parana, Brazil; south China
  A long, wet growing season supports thick evergrowth forest, but greater seasonality and occasional winter frosts prevent development to the scale of tropical rainforests.
  
• Subtropical Monsoon Forest (CTfp)
  -north Myanmar; Batumi, Georgia; Kagoshima, Japan
  Extremely heavy rains support lush vegetation, but seasonal cold or dry conditions still slow growth, preventing development of full rainforest structure.
• Subtropical Moist Savanna (CTs)
  -Below 0.75 Ar
  -Pampas; Miombo; Ganges plain, India
  Drier conditions transition to savanna and woodland similar to tropical savannas, but cooler conditions reduce the severity of the dry season, so may still have full forests and lush undergrowth.
• Subtropical Moist Monsoon Savanna (CTsp)
  -patches in north California, Albania, Nepal
  Heavier seasonal rains may favor thicker vegetation.
• Temperate (CD)
  -Above 1250 GInt and above 1300 GDD
• Oceanic Temperate (CDa)
  -Cool winters
  -Delaware; Belgium; Nanjing, China
  Lush deciduous forests with large-leafed undergrowth, but not as thick or tall as warmer evergrowth forests.
  
• Oceanic Temperate Rainforest (CDap)
  -coastal Oregon; west Ireland; Valdivian forests, Chile
  Heavy rains support taller trees and lusher undergrowth, potentially with more evergrowths.
  
• Continental Temperate (CDb)
  -Cold or frigid winters
  -Ohio, USA; Belarus; Seoul, South Korea
  Less diverse mixed deciduous/evergrowth forests with harsher winters and more regular snow cover.
  
• Continental Temperate Rainforest (CDbp)
  -Nova Scotia, Canada; Ljubljana, Slovenia; highland Japan
  Heavy rains may support a lusher, taller forest, but cold winters limit undergrowth density and overall diversity.
• Boreal (CE)
  -Below 1300 GDD
  
• Oceanic Boreal (CEa)
  -Cool winters
  -Falklands; Orkneys, Scotland
  Evergrowth forests with mild winters, allowing for somewhat more diverse vegetation, potentially with some deciduous trees.
• Oceanic Boreal Rainforest (CEap)
  -Magellanic forests, Chile; Reykjavik, Iceland; Ketchikan, USA
  Heavy rains and mild winters may encourage more deciduous varieties or at least lusher evergrowths and undergrowth
• Continental Boreal (CEb)
  -Cold winters
  -Edmonton, Canada; Finland; Yekaterinburg, Russia
  Typical polar evergrowth forests, with uniform, conservatively growing trees and little undergrowth.
  
• Continental Boreal Rainforest (CEbp)
  -Newfoundland, Canada; Trondheim, Norway; central Austria
  High precipitation may encourage lusher undergrowth, but may also manifest as heavy snowfall and significant spring melt.
• Percontinental Boreal (CEc)
  -Frigid Winters
  -east Siberia
  Extreme frost in winter prevents survival of evergrowth leaves, prompting switch to deciduous forests but still with very conservative growth patterns.
• Percontinental Boreal Rainforest (CEcp)
  -patches in Siberia
  Rare on Earth but may be more common in a colder or more seasonal world; heavy precipitation likely to manifest as heavy snow, may lead to boggy, more open environment due to low evapotranspiration.
• Submediterranean (CM)
  -Low GrS
• Oceanic Submediterranean (CMa)
  -Cool winters
  -Seattle, USA; Samsun, Turkiye; Christchurch, New Zealand
  Moist enough conditions to support thick forests, but drier summers still favor patchier woodlands with a mix of deciduous and evergrowth vegetation.
• Continental Submediterranean (CMb)
  -Cold or frigid winters
  -north Idaho; patches in Greece and Turkiye
  Harsher winters may encourage more exclusively evergrowth forests and more limited undergrowth.
• Cold Semiarid (CA)
  -Below 0.5 GAr
  
• Mediterranean (CAM)
  -Low GrS
• Oceanic Mediterranean (CAMa)
  -Cool winters
  -central California, USA; Santiago, Chile; Sicily
  The archetypal mediterranean climate, with substantial summer droughts favoring a mosaic landscape of trees, grass, and shrub, supporting a variety of unique fauna.
  
• Continental Mediterranean (CAMb)
  -Cold or frigid winters
  -east Oregon; Tabriz, Iran; central Tajikistan
  Colder conditions may reduce diversity, but also tend to favor thicker, more evergrowth-dominated forests.
• Cool Dry Savanna (CAa)
  -Cool winters
  -Gran Chaco, Argentina; Mopane woodlands; Jaipur, India
  Savanna and woodlands with scattered trees and brush or sometimes dry forest, but not as productive as tropical savannas; though may also have milder dry seasons.
• Cool Dry Monsoon Savanna (CAap)
  -patches in Oregon, Spain, and Iran
  Rare on Earth, indicates heavy seasonal rains; may support thicker undergrowth and more tree cover.
• Cold Steppe (CAb)
  -Cold or frigid winters
  -shortgrass prairie, USA; south Ukraine; Inner Mongolia
  Colder conditions create more completely grass-dominated plains, with very few trees.
• Cold Pluvial Steppe (CAbp)
  -patches in BC, Canada and Afghanistan
  Also rare; heavy seasonal rains may support more vegetation and reappearance of trees.
• Tundra (CF)
  -Below 350 GDD
• Oceanic Tundra (CFa)
  -Cool winters
  -south Tierra del Fuego, Argentina; patches in Iceland and Alaska, USA
  Too little warmth for trees and widespread permafrost but fairly consistent temperatures allow for tall grass and large shrubs.
  
• Continental Tundra (CFb)
  -Cold or frigid winters
  -Baffin Island, Canada; north Russia; Tibet
  More seasonal climates more typical of polar regions, with some grass, lichen, and shrubs, and potentially migratory grazers, but little activity in winter.
  
• Cold Barren (CG)
  -Below 50 GDDz
  -central Svalbard; strips around Greenland ice sheet;
  Regions of bare soil and rock too cold for most vegetation but without permanent ice cover and so possibly hosting some lichen or microbes; mostly around the edges of ice sheets or in mountains, but could cover large portions of a very dry planet.
  
• Ice (CI)
  -Permanent ice cover
  -Antarctica; central Greenland
  Ice sheets and other glaciers, with no stable ground for vegetation but potentially microbes or transient fauna.
  
• Hot (H)
  -Hot, torrid, or boiling summers and mild winters
• Supertropical (HT)
  -Below 1250 GInt and hot summers
• Supertropical Forest (HTf)
  -Above 0.75 Ar
  A long, wet growing season favors tropical-like vegetation, but adaptations to extreme heat are required.
• Supertropical Monsoon Forest (HTfp)
  Heavier rains may help mitigate heat stress of plants, though may make homeothermy difficult for animals.
• Supertropical Moist Savanna (HTs)
  -Below 0.75 Ar
  Transition to savanna or woodland with drier conditions.
• Supertropical Moist Monsoon Savanna (HTsp)
  Heavier seasonal rains may favor thicker vegetation and mitigate heat stress.
• Swelter (HD)
  -Above 1250 GInt and above 350 GDD 
• Hot Swelter (HDa)
  -Hot summers
  Long periods of extreme heat may require deciduous vegetation and dormancy, but are still generally tolerable by complex life.
• Hot Pluvial Swelter (HDap)
  Heavy rains may favor thicker vegetation and help mitigate summer heat stress.
• Torrid Swelter (HDb)
  -Torrid summers
  More extreme summer heat is more challenging to complex life, requiring either more extreme dormancy, migration, or more fundamental biochemical alterations.
• Torrid Pluvial Swelter (HDbp)
  Heavier rains may favor lusher vegetation if summer heat can be tolerated.
• Boiling Swelter (HDc)
  -Boiling summers
  Boiling summer temperatures hazardous to life, potentially preventing permanent habitation unless organisms can adapt to completely dry out or shelter underground.
• Boiling Pluvial Swelter (HDcp)
  Heavier rains may benefit any migratory or boiling-tolerant life, but high humidity may also make homeothermy more difficult.
• Subparamediterranean (HM)
  -Low GrS
• Hot Subparamediterranean (HMa)
  -Hot summers
  Drier growing periods may favor a patchier evergrowth/deciduous mix, but still allow for thick vegetation.
• Torrid Subparamediterranean (HMb)
  -Torrid summers
  Hotter summers may inhibit evergrowth habits and so reduce overall vegetation.
• Boiling Subparamediterranean (HMc)
  -Boiling summers
  Vegetation may be limited to sparse, ephemeral forms.
• Hot Semiarid (HA)
  -Below 0.5 GAr
• Paramediterranean (HAM)
  -Low GrS
• Hot Paramediterranean (HAMa)
  -Hot summers
  Droughts during growth periods may require plants to specialize either in growth during hot, wet periods or reliance on stored water.
• Torrid Paramediterranean (HAMb)
  -Torrid summers
  Hotter summers may limit viable rain-dependent growing season, increasing dependency on stored water.
• Boiling Paramediterranean (HAMc)
  -Boiling summers
  Extreme summer heat may limit permanent vegetation.
• Hot Dry Savanna (HAa)
  -Hot summers
  Summer heat and evapotranspiration may reduce but perhaps not eliminate large permanent vegetation.
• Hot Dry Monsoon Savanna (HAap)
  Heavier seasonal rains may help mitigate heat stress and support more vegetation.
• Torrid Steppe (HAb)
  -Torrid summers
  Extreme summer heat and poor water availability may limit life to annuals or migratory species.
• Torrid Pluvial Steppe (HAbp)
  Heavier seasonal rains may allow for somewhat lusher vegetation, at least seasonally.
• Boiling Steppe (HAc)
  -Boiling summers
  Vegetation strongly limited by heat and water availability.
• Boiling Pluvial Steppe (HAcp)
  Rains may favor some vegetation depending on seasonal patterns of water availability.
• Parch (HF)
  -Below 350 GDD
• Hot Parch (HFa)
  -Hot summers
  Limited cool growing periods, but moderate summers may allow for some permanent vegetation.
• Torrid Parch (HFb)
  -Torrid summers
  More extreme summers may limit life to only small annuals.
• Boiling Parch (HFc)
  -Boiling summers
  Largely uninhabitable region with only brief hospitable periods.
• Hot Barren (HG)
  -Below 50 GDDz
  Areas with no period appropriate to Earth-like vegetation growth, may be occupied only by microbes.
  
• Extraseasonal (E)
  -Hot, torrid, or boiling summers and cool, cold, or frigid winters
• Extratropical (ET)
  -Below 1250 GInt, warm or hot summers, and mild or cool winters
• Extratropical Forest (ETf)
  -Above 0.75 Ar
  Thick tropical-like forests but adaptations required for strong seasons.
• Extratropical Monsoon Forest (ETfp)
  Heavier rains encourage thicker forests, but still inhibited by temperature extremes.
• Extratropical Moist Savanna (ETs)
  -Below 0.75 Ar
  Sparser savanna and woodlands, potentially with complex growth patterns around seasons and dry periods.
• Extratropical Moist Monsoon Savanna (ETsp)
  Heavier rains support thicker woodlands and forest and may help mitigate seasonal stresses.
• Extracontinental (ED)
  -Above 1250 GInt and above 350 GDD
• Superseasonal Extracontinental (EDa)
  -Warm or hot summers and mild or cool winters
  Likely deciduous forests similar to temperate climates, but may have complex patterns of vegetation growth and migration or dormancy to best take advantage of growing periods between seasonal extremes.
• Superseasonal Extracontinental Rainforest (EDap)
  Heavy rains may allow lusher, potentially evergrowth vegetation, though may favor life more tolerant to temperature extremes.
• Hyperseasonal Extracontinental (EDb)
  -Torrid or boiling summers or cold or frigid winters
  More severe seasons may favor more evergrowth varieties or more complex patterns of deciduous vegetation depending on the particular temperature range; likely strong incentive for migration or dormancy in animals.
• Hyperseasonal Extracontinental Rainforest (EDbp)
  Heavy rains may help vegetation if timed properly to coincide with growing periods.
• Subextramediterranean (EM)
  -Low GrS
• Superseasonal Subextramediterranean (EMa)
  -Warm or hot summers and mild or cool winters
  Inconsistent availability of moisture may limit forest cover or favor more evergrowths.
• Hyperseasonal Subextramediterranean (EMb)
  -Torrid or boiling summers or cold or frigid winters
  Potentially patchier vegetation or more ephemeral life.
• Extraseasonal Semiarid (EA)
  -Below 0.5 GAr
• Extramediterranean (EAM)
  -Low GrS
• Superseasonal Extramediterranean (EAMa)
  -Warm or hot summers and mild or cool winters
  may resemble Mediterranean but with more migration, dormancy, or deciduous habits, with life adapted to best take advantage of limited windows of ideal growing conditions.
• Hyperseasonal Extramediterranean (EAMb)
  -Torrid or boiling summers or cold or frigid winters
  Severity of seasons may limit any permanent
  
• Superseasonal Dry Savanna (EAa)
  -Warm or hot summers and mild or cool winters
  May have scattered trees but more severe seasons may favor deciduous or annual forms.
• Superseasonal Dry Monsoon Savanna (EAap)
  Heavier rains may allow for lusher vegetation, at least seasonally.
• Hyperseasonal Steppe (EAb)
  -Torrid or boiling summers or cold or frigid winters
  Severe seasons and limited water availability may limit vegetation to ephemeral forms like annual grasses.
• Hyperseasonal Pluvial Steppe (EAbp)
  Seasonal rains may help bring greater blooms of growth, but temperature extremes may still limit more permanent forms.
• Pulse (EF)
  -Below 350 GDD
• Superseasonal Pulse (EFa)
  -Warm or hot summers and mild or cool winters
  Likely fairly rare, may require short years; brief periods of growth between seasonal extremes likely limit life to ephemeral and migratory forms.
• Hyperseasonal Pulse (EFb)
  -Torrid or boiling summers or cold or frigid winters
  Life even more limited by seasonal extremes, though may have little impact on already ephemeral vegetation.
• Extraseasonal Barren (EG)
  -Below 50 GDDz
  Heavily seasonal regions of the nightside of tidal-locked planets.
• Ocean (O)
  -Ocean areas
  
• Cold Ocean (Of)
  -Above 20% maximum sea ice
  
• Permanent Frozen Ocean (Ofi)
  -Above 80% minimum sea ice
  Permanent ice cover limits photosynthesis, but diverse life may still exist below ice shelf.
• Seasonal Frozen Ocean (Ofd)
  -Below 80% minimum sea ice and above 50 GDDl
  Seasonal breaks in ice cover allow for more photosynthesis.
• Dark Seasonal Frozen Ocean (Ofg)
  -Below 80% minimum sea ice and below 50 GDDl
  Ephemerally warm regions of the nightside of tidal-locked planets.
• Dark Ocean(Og)
  -Below 50 GDDl
  Unfrozen oceans in warm regions of the nightside of tidal-locked planets, lacking photosynthetic life but possibly supporting other ecosystems.
• Cool Ocean (Oc)
  -Below 18 °C minimum surface temp and below 40 °C maximum surface temp
  Lack of ice allows for uninterrupted growth, but waters are too cold in winter for coral; coastal waters dominated by flora like kelp or shellfish.
• Tropical Ocean (Ot)
  -Surface temp always between 18 °C and 40 °C
  Stable water temperatures support corals or similar life in coastal waters, forming high-diversity shallow reefs.
• Hot Ocean (Oh)
  -Above 18 °C minimum surface temp and 40-60 °C maximum surface temp
  Conditions too hot for corals, may be inhabited by more warm-adapted species.
• Torrid Ocean (Or)
  -Above 60 °C maximum surface temp
  Surface temperatures for at least part of the year exceed limits for complex life, though this might be avoided by diving into deeper waters.
• Extraseasonal ocean (Oe)
  -Below 18 °C minimum surface temp and 40-60 °C maximum surface temp
  Warm but variable oceans, challenging for sessile shallow water life.

The full scheme is implemented in koppenpasta, with a few different variants:

 

The land zones can be implemented as the full set, without pluvial zones, limited to Earthlike zones (no hot, extraseasonal, or quasitropical and no light restrictions on GDD), or Earthlike without pluvial, as shown below:

 

The sea zones can similarly be implemented as the full set, Earthlike-only (no dark, hot, torrid, or extraseasonal), or limited to only show ice cover.

 

For now, there's only a single standard color set, but I may add more in the future.

 

But that's a matter for another time. In Part III, we'll go over these maps in more detail and discuss how well this scheme represents Earth's biomes—and how well ExoPlaSim replicates those patterns—then look back at all our previous explorations and see how this new scheme helps to highlight their various peculiarities.

 

But before that, one last little detour:

Unproxied Koppen-Geiger

Though I thought the isues with applying Koppen-Geiger to other worlds ran deep enough to merit building this whole new system, I aslo figured that it would be useful to have an option to still use the same classical set of Koppen-Geiger zones, but with the definitions adjusted to more directly reflect the parameters more fundamentally linked to biome distribution rather than the proxies from the original scheme.

 

Many of the zones can share the same parameters as used for the Pasta system and and can share some of the same thresholds (and thus the same tunings for different data sources), though I've tweaked them in a few cases to better match the original distribution of Koppen zones. The tricky bit is that the Xxa and Xw categories don't clearly correspond to any of the biome boundaries I identified for tuning my system.

 

Despite some of the names used, Xxa doesn't quite correspond to subtropical climates and is perhaps best interpreted as indicating a long growing season which might better allow for multiple crop harvests, something I didn't identify in my system because it doesn't much impact natural biome makeup but can be easily enough pinned to a high GDD count here. Incidentally, note that for this scheme GDD and GDDz aren't given any high-temperature cutoffs (i.e. any temperature above 25 °C always gives 20 GDD, because there's no hot zones we could transition to.

 

Xw is a bit more of a challenge, because as mentioned it doesn't correspond too well to any actual biomes. I considered just switching it to represent pluvial zones, but after playing around with a few parameters I found I could get a decent match to the original Xw distribution by defining it in terms of the ratio of GAr to Ar; what that means practically I couldn't quite say but it should at least be a little more flexible with different seasonal patterns and sampling periods.

• Mild winters for A
• Above 0.92 Ar for Af
• Above 0.85 Ar or below 0.45 Evr for Am
• GAr > Ar for Aw
• Otherwise As
• Below 0.32 Ar for B
• Below 0.14 Ar for BW
• Otherwise BS
• Mild or cool winters for BXh
• Otherwise BXk
• Cool winters for C
• Low GrS for Xs
• GAr/Ar > 1.02 for Xw
• Otherwise Xf
• Over 2300 GDD for Xxa
• Over 1300 GDD for Xxb
• Cold or frigid winters for D
• Frigid winters for for Dxd
• Otherwise as for C
• Below 250 GDD for E
• Above 10 cm MinIce for EF
• Otherwise ET

The results definitely aren't a perfect match for the original but do generally seem to capture the spirit, if that's what you're looking for. This alternate scheme is implemented in koppenpasta with all the subtypes and color options as for regular Koppen-Geiger.

 

That'll do for today. Pick up the koppenpasta script here if you want to play around with it. See you in Part III!

 

Buy me a cup of tea (on Patreon)