Beyond the Köppen-Geiger Climate Classification System, Part III: Assessing the Pasta Bioclimate Classification System

This is Part III of a short series I've been doing on creating a new climate classification system, the Hersfeldt or Pasta Bioclimate Classification System, as an alternative to Koppen-Geiger that might better suit some of the more unusual climates I've been exploring through ExoPlaSim modelling. In Part II, I explained the details of this scheme and my rationale for constructing it; this time, I'll look at how well this system applies to Earth, and then go over all our past climate explorations to see how they look in this new scheme.

• koppenpasta Tweaks
• The Pasta Bioclimate System on Earth
• North America
• Europe
• South America
• Africa
• Asia
• Australia
• Modelled Earth
  
•  Climate Re-Explorations
• Temperature
• Hotter Climates
• Colder Climates
  
• Obliquity
• Lower Tilt
• Higher Tilt 
• Day Length
• Shorter Days
• Longer Days
• Eccentricity
• Tidal-Locked Earth
• Teacup Ae

Back to Part II

koppenpasta Tweaks

First off, a few little updates on the koppenpasta script; aside from various bugfixes (let me know if you still have any more issues), I've updated it to work properly with the default fonts in PIL, and added a font_size option to adjust the font size in map keys and charts.

 

As you can tell from the header, I've also added a "true color" approximation color option for the Pasta scheme, using much the same approach as for Koppen: comparing a satellite map of Earth to a map of bioclimate zones and then finding the average true color in each zone, though with a bit of careful filtering in some cases to keep the colors from being too affected by snow or hillsides. For zones that don't appear on Earth, I just copied what seemed like the nearest analagous Earth zone; that's something you can play with if you want something a bit more exotic.

 

Now for some technical matters: there was an issue with calculation of GDD for hot climates which I've corrected, and on further testing with correct GDD calculation I've decided to adjust the GDD calculations for a better tropical-hot transition: Where before GDD accumulation began declining above 35 °C and stopped at 45 °C, and the same for GDDz at 35 °C and 55 °C, I've increased all these thresholds by 5 °C. This perhaps feels a bit of artificial, but it's well within uncertainties about photosynthesis responses to high temperatures—and if we're concerned with hot bioclimates above tropical at all we're probably speculating some degree of adaptation to high temperatures—and avoids oddities like tropical regions with steady temperatures between 45 and 50 °C having no growing period and so being classified as TF.

 

I've also adjusted the way GDD is calculated, such that if GDD falls to zero in a month, this only stops accumulation if it occurs during a period of GInt accumulation where it reaches over that 1250 threshold to count as a substantial growth interruption dividing XT from XD zones; this shouldn't affect most climates, it's mostly for certain cases of planets with long days and short sampling periods to prevent the GDD count being interrupted by relatively short nights (because this means this GInt threshold is now used in two places in the script, for counting GDD and for defining zones, I've added a single pas_gint_thresh option to set both). To avoid unnecessary complexity and recursion issues, this is only applied to regular GDD, not GDDz or GDDl for oceans, and there's no analogous requirement for GInt interruption. As another measure for planets with long days, I've adjusted the default procedure for binning months in koppenpasta to preserve the extremes of maximum and minimum temperature in each "bin" of months rather than just finding an average; so you can use shorter sampling periods for these cases, and then bin months together in the script such that daily temperature variation doesn't affect GDD and GInt counts at all but the extremes of midday high and night low temperatures are properly recorded.

 

Finally, I've tweaked a few of the options for deriving the Pasta bioclimate system from monthly average temperature and precipitation data alone: based on some testing with Earth data I've tweaked the "kalike" estimation of potential evapotranspiration from from 4.5 to 7 mm/month/°C, and I've replaced the evap_estimate_sea option with the more general estimate_evap option, which can be set to never to never estimate evapotranspiration from precipitation and PET, sea to estimate it over sea cells when interpolating, and all to force it to estimate evapotranspiration everywhere without even consulting the input file's evap data (but the script should still properly interpret evap_estimate_sea in old config files). I've also added an extra pas_simple_input option which automatically sets all the other options to use only these two inputs, for both land and sea, and while I was at it added the pas_med_thresh to allow for direct tuning of the GrS threshold for Mediterranean zones, as that seems to be fairly sensitive to different evaporation estimations. This is how that all looks applied to real Earth data, with the GrS threshold set to 1.0:

 

Not too bad a match overall; applied to ExoPlaSim data (with the usual 0.8 GrS threshold), it creates an odd proliferation of polar pluvial zones, but otherwise again mostly matches up:

 

Still, I don't recommend actually using these options for ExoPlaSim data; better to use the more accurate evaporation data and PET estimation methods where they're available. This all mostly just makes it easier for myself to use different data sources in the future for public data explorations.

 

I've edited Part II to reflect all these changes, just to avoid any future confusion.  

The Pasta Bioclimate System on Earth

We'll start today by taking a closer look at the bioclimate zones based on real Earth climate data (with the proper PET and evapotranspiration data), going continent-by-continent (sans Antarctica, which as you can imagine isn't much of a challenge to classify), highlighting both how I tuned the system to best match the distribution of biomes on Earth and a few problem areas I couldn't quite get to match up as well as I would have liked.

North America

One consistent theme we'll see is that biomes on the west side of the Atlantic tend to be wetter than their counterparts on the east side, so many of the aridity boundaries have been tuned as a bit of a compromise between the two. Here, the extent of semiarid (CA) zones seems to somewhat underestimate the extent of the Great Plains, but that might reflect the strong influence of both grazing herbivores and anthropogenic fires in restricting tree growth around its edges. Poor soil water capacity and frequent natural fires also cause many areas in the middle and south of the US to have sparser vegetation than their climate would suggest.

 

Otherwise, the bioclimate zones fairly faithfully reflect the subtropical-temperate-boreal transitions in the east; semiarid-arid transitions and patches of montane forests in the mountains; transition from drier Mediterranean climates to temperate and boreal rainforests in the west; and patches of wetter and drier tropical climates in the south.

 

Some other notable points:

• There's a large patch of Mediterranean (CAMa)/submediterranean (CMa) zones in Texas and some other patches in Georgia; these seem to correspond to an area of wet springs and falls with dry summers and winters, and though they aren't usually recognized as Mediterranean, they do mostly have an appropriate mix of oak, conifers, and scrub.
• Mediterranean zones also extend rather far inland in the west compared to most classification schemes, and northern Idaho/Montana features the world's largest patch of continental submediterranean (CMb); these areas are a bit inconsistently classified in biome schemes, but generally feature montane pine forests or Mediterranean-like scrub.
• A common complaint I've seen about Koppen is that it classifies New York and Louisiana together into the same subtropical climate, so I'm happy to report New York is oceanic temperate (CDa) here, distinct from both the subtropical deep south and the simpler continental forests of New England.
• Hawaii's famous climate diversity in Koppen is matched here, ranging from oceanic tundra (CFa) at the volcanic peaks to hyperpluvial rainforest (TUrp) where lowland slopes catch rain from the trade winds.
  
• The transition across central Mexico from the tropical forests of Central America to the Sonoran Desert is rather complex and I can't say I've taken the time to match up all the climate zone lines here to biome distribution, but overall I can't see any obvious discrepancies between the variously semiarid zones and the dry forests, savanna, and scrub in reality.
• Patches of steppe in Alaska and northwest Canada are variously identified in biome classifications, though often just included with tundra; these are generally dry regions, but low evapotranspiration and a short growing season means their growth aridity count may depend on the exact timing and amount of a few summer rains and so their extent may vary considerably based on the dataset used.
  
• Nova Scotia and some other patches of Canada's east coast are often recognized as temperate or boreal rainforest, but this is not generally extended as far inland as the pluvial zones here; this may be because the colder inland climate has shorter summers and much of the heavy precipitation arrives as winter snow, so is less of a benefit to vegetation.
• Greenland presents our best example of a large continental glacier directly adjacent to unglaciated land; this map underestimates the extent of the ice sheet a bit, but this reflect issues with the temperature data (I didn't bother to pull up a map of actual ice cover); there's also a couple patches of boreal (CEb) on the south coast, which do appear around Greenland's only forest, which somewhat supports my choice to use the 350 GDD line for tundra.

Europe

Much of Europe is variously surprisingly dry or surprisingly wet, so a lot of the definitions of semiarid and Mediterranean zones have been tuned around ensuring their proper distribution here. In particular the choice to define semiarid zones by growth aridity (GAr) helps to reflect the proper distribution of the Pontic Steppe, and the choice to define Mediterranean zones by growth supply (GrS) helps to ensure that they're widespread in the south but absent from the north.

 

Pluvial zones were also tuned in part by the distribution of temperate rainforests in the north, and the temperate and boreal transitions tuned by their boundaries in central Europe; I hope I'm not coming off as too eurocentric, but the continent has a number of peculiarities in its precipitation patterns that help test the flexibility of the overall system.

• Much of the motivation for including submediterranean (CM) zones was to prevent large stretches of the Italian and Turkish coasts from being classified as moist temperate or subtropical, though submediterranean turns out to be a good fit for the aforementioned regions of North America as well.
  
• Much of central European Russia is covered in patchy woodland or forest sometimes called "forest steppe", which I considered trying to classify as a separate biome from temperate and boreal forests, but ultimately I couldn't find a good definition that extended well to similar transitional forests in North America but not to dissimilar parts of western Europe (the climate has some similarity to Mediterranean with a burst of high evapotranspiration in summer, but again trying to stretch the submediterranean zones to include these areas would have extended them too far elsewhere), and ultimately I decided I shouldn't add too many extra zones to account for these transitional regions.
  
• The choice to define the subtropical-temperate transition based on growth interruption (GInt) rather than winter temperature or GDD count was based in part on the lack of a clearly distinct subtropical biome in western Europe despite its quite long summers and mild winters by those measures, though ultimately there is still a subtropical strip in coastal Spain and France; but it's not clear if the lack of subtropical forests is related more to the current climate or the cooler climates of the past. During the Last Glacial Maximum, Europe's deciduous forests could survive in refuges around the Black Sea, but there were no suitable nearby refuges for subtropical forests, so these areas may be occupied by temperate vegetation now simply for lack of competition; but the definition of subtropical climates by GInt still works best in Asia even neglecting Europe's peculiarities.
• Whatever credit I've gained for dividing New York from Louisiana is probably lost by classifying central Italy as the same temperate (CDa) climate as coastal Scotland due to excluding Koppen's Xxa/Xxb distinction, but again I didn't want to get too granular and it is common for biome classifications to lump most of Europe together into one vast stretch of temperate forest.
  
• Norway and parts of the British isles have substantially drier summers than winters in terms of precipitation, but still feature the typical vegetation of temperate and boreal rainforests because summer precipitation is still high relative to evaporation even if it's lower than in winter; ensuring proper classification of these regions helped tune Mediterranean and pluvial zones.
• Classificiation of Iceland's coastal biomes varies significantly; here they're classified largely as oceanic boreal (CEa), which is an okay fit but perhaps implies somewhat more tree cover than in reality; much of it might fit into the category of "taiga savanna" which I was tempted to split into a separate zone but ultimately decided was too granular, and again I chose a lower tundra GDD threshold with somewhat more of a mind to ExoPlaSim data, which tends to underestimate polar summer temperatures.

South America

As mentioned, the Americas tend to be wetter than their eastern counterparts, and the tropical aridity transitions were tuned in large part to make sure they were suitably widespread in Africa while not being too widespread here. There's still probably a bit too much tropical forest (Tf) to moist savanna (Ts), but it's an acceptable compromise, distinguishing the dry strip across central Brazil from the rainforests of the coast and central Amazon, and even distinguishing the moister Cerrado savanna from the neighboring dry forests.

 

To the west and south, the Andes is one of the largest areas of near-equatorial highlands, which are often challenging to classify in systems tuned based on the heavily seasonal polar climates. The drier southern parts are easy enough, with the largest patch of cold true desert (Ahc) outside of Asia corresponding to the Atacama plateau; to the north, the mountains have a complex temperate mix which roughly correspond to the cloud forests of the Amazon, and boreal (CE) climates take on a different but sensible role here, corresponding not so much to conifers but to "elfin" forests of stunted trees adapted to slow but steady growth.

• The exact division of tropical rainforest from seasonal forest varies between sources, and many don't distinguish them at all; I've gone with a somewhat conservative extent of true rainforest (Tr), though with a mind to the additional patches in central America.
• The distinction of hyperpluvial rainforests (Trp) was motivated in part by the presence of tropical heath forests with soil too poor for full rainforest growth, but in reality these areas only account for a small portion of this zone, concentrated along the banks of major rivers; mostly this zone just gives a bit more texture to rainforest, indicating the absolutely wettest areas. 
• The subtropical biomes in the southeast are often a bit tricky to pin down, with overlapping gradients in latitude, elevation, and aridity, and they're generally sparser than their high precipitation would imply; the boundary of forest (CTf) is probably too far south, but you still get an overall sense of the divisions between the lusher campos, sparser pampas, and drier chaco, with an additional wetter patch for the yungas forest in the foothills of the Andes.
• In the southwest, Mediterranean zones extend perhaps a bit too far east and south and not far enough north, but the mix of temperate and boreal forests in southern Chile does a good job of showing the Valdivian and Magellanic forests, which feature a fairly unique conifer and deciduous mix, the latter corresponding neatly to one of the world's few areas of oceanic boreal (CEa).

Africa

Africa always provides a bit of a challenge for climate and biome classification, as the competing influences of average and seasonal climate, herbivores, fires, human activity, and terrain create a complex patchwork of semiarid climates, hence the occasional tendency to lump most of the continent together into a single "tropical savanna" category. I can't show all these biome types with my scheme—and as mentioned the definitions are tuned as a compromise between here and South America, so the extent of forests may be a tad underestimated in places—but it does a decent job of showing that there's some variation in these areas, distinguishing the dry Sahel and East African rift valley from the wetter coastal forests.

 

Southern Africa also gives a good case study in the transition from tropical to subtropical climates, and part of my motivation for extending the moist savanna category into subtropical (CTs) climates was to properly represent this area; without it, the scheme would imply that tropical savanna (Ts) in the north transitioned to subtropical forest, when in fact there's a fairly consistent gradient to sparser woodlands as you head south towards the Kalahari desert, with this subtropical region mostly corresponding to the Miombo woodlands.

• Vegetation in the Sahel extends surprisingly far north into the Sahara, and is underrepresented by the spread of semidesert (Ada) here; but lowering the aridity threshold any further would have overrepresented the extent of semidesert in Arabia (and no other parameters for measuring aridity I tested performed better).
• The Fynbos region on the tip of South Africa provided a good reference point for tuning the Mediterranean (CAM) definition, as much of it isn't actually all that much drier in summer but still has fairly clear Mediterranean vegetation.
• Madagascar is rather climatically diverse for its size, and this is another spot where the moist (Ts)/dry (TA) savanna distinction helps highlight the difference between the dry scrub and woodlands in the south and thicker forests in the north.
• The East Africa rift highlands are an eclectic mix of climates as usual across different schemes; the tendency for Mediterranean climates to appear here so close to the equator always struck me as odd, but on close investigation it makes some sense: these areas do have a subtle twice-yearly seasonal cycle with warm periods close to the equinoxes, and heavy rains tend to come at about the same time, but carried by very consistent winds from the east; so areas to the west of highlands are rainshadowed during the hottest periods and instead receive their heaviest rains somewhat earlier or later.

Asia

Asia is a very large and diverse continent, with a number of biomes either unique to the continent or best represented here. The extent of the East Asian monsoon isn't quite as clear here as it is in Koppen, but various pluvial zones mark the areas where it has the largest impact, highlighting the coastal forests of south Asia, the orographic rains against the edges of the Himalayas, and the highland temperate rainforests of Japan.

And yes, the Caspian Sea is not properly cut out here, it's a common quirk of climate datasets that treat all lake area as land and the Caspian as a giant lake.

• The tropical-cold and subtropical-temperate boundaries were tuned largely based on east Asia, based both on the transition of forest types in China and the presence of small evergrowth regions in Japan and South Korea.
• As in south America, the extent of hyperpluvial rainforest (Trp) in Indonesia gives some sense of regions with some stunted tropical heath forests or peat swamps, but is far from a perfect match; again it more indicates regions more likely to contain such biomes rather than their actual extent. 
• Eastern Siberia features the world's largest region of frigid winters, and both here and in Koppen this area is marked out in recognition of its unique needle-leaved but deciduous larch forests. The exact extent of this region is a bit tricky to pin down and may reflect patterns of forest recovery after the Last Glacial Maximum as much as exact winter severity; I've gone with a somewhat conservative extent here, which is still far larger than Dxd zones in Koppen, probably due to underestimated winter temperatures in the older datasets used to originally define it.
• The exact line between boreal forest and tundra on the Tibetan plateau is often not well mapped out, but suffice it to say the that the patches of forest in the plateau interior predicted here are not borne out; this may represent weaknesses in the definitions for the boreal-tundra line or perhaps the moist-steppe line when applied to highland plateaus, or these areas could be unsuitable for forests for reasons outside of climate.
• The deserts of central Asia provided good reference for tuning the aridity definitions to ensure they worked as well here as in the main subtropical deserts, and the addition of a separate semidesert (Ad) zone was motivated in part to better represent these areas.
• Iran and Pakistan are trickier; they're rather inconsistently classified in biome schemes as they're quite dry in any average measure but the mountainous terrain provides some cover and shade for shrubs that makes them at least a tad more hospitable than the large patches of hyperarid desert (Aha) may imply (and there's the Indus river, but at this resolution none of the major desert rivers seem to provide enough evaporation to shift the needle much). There are some genuinely barren regions here, but at any rate it was tricky to find any measure that would extend semidesert (Ad) as far as I wanted here and in the Sahara without extending it too far in Arabia and the Gobi, so I settled on this compromise.
• Central and western India are also a bit less diverse than I might have liked (outside the coastal monsoon forests, which were a major motivating factor for including pluvial zones as a separate category rather than trying to lump monsoon forests together with the moderately seasonal fringes of tropical rainforests as in Koppen); the distribution of forest (Tf), moist savanna (Xs), dry savanna (XA), and semidesert (Ada) gives a decent sense of the dominant gradient to drier conditions, but somewhat misses the diversity of dry forests, savanna, and thorny scrub here.
• The large patch of Mediterranean (CAM) climates in central Asia is a tad odd for a climate usually associated with coastlines, but represents the combined influence of strongly shifting seasonal wind patterns over Eurasia and rainshadowing from various mountains, and corresponds to a stretch of conifer woodlands thicker than nearby non-Mediterranean steppe (CAb). The stretches of continental Mediterranean (CAMb) on the north side of the Central Asian desert have less correspondence with any real biome distinction; I considered adding some additional minimum aridity requirement to Mediterranean zones to remove them, but that would've also removed a lot of more uncontroversial Mediterranean elsewhere so I'm willing to accept them as a regional quirk.

Australia

Australia is a decidedly semiarid continent, with few areas either very dry or very wet. Compared to the other subtropical continents, the mix of vegetation is a bit more distinct and perhaps less influenced by herbivory but moreso by seasonal fires. Still, the scheme gives a fairly good sense of the major transitions from the coastal forests to the sparse but rarely completely barren outback.

• Mediterranean (CAMa) is fairly extensive here compared to some schemes; while the southwest is usually counted as a Mediterranean biome, classification of the southeast is more variable. The zone is still appropriate to the scrubby woodland common there, but the Mediterranean definition was tuned in part to avoid including too much of the wetter east coast.
• Though north Australia has few pluvial zones, the stretch of moist savanna (Ts) and patches of forest (Tf) in the northeast still gives a good sense of the influence of the Australian monsoon, forming generally lusher woodlands here.
• New Zealand and Tasmania are rather diverse mixes of subtropical forest (CTf), submediterranean (CMa), and temperate to boreal forest or rainforest, which is a decent representation of their fairly varied vegetation; there's a reason Lord of the Rings could be filmed largely just in New Zealand.
• As typically recognized in biome schemes, the New Guinean highlands shade from tropical into subtropical and temperate, and even feature a few patches of the rare oceanic tundra (CFa), reflecting the cold but aseasonal mountain climate.

Modelled Earth

With that tour complete, let's see how our baseline ExoPlaSim model compares.

Its greater sensitivity to fundamental climate parameters has perhaps made my scheme more sensitive to ExoPlaSim's faults, most of which are symptomatic of the same key issues that the resolution is too low to model some local climate dynamics and the model lacks proper ocean current modelling, which generally makes the tropics too hot and poles too cold. Thus, there's too much boreal (CE) and tundra (CF) in Europe—though they're actually fairly well distributed in North America and Asia—and there are hot (H) and extraseasonal (E) zones here not seen in the real data; there are some areas on Earth that regularly reach above 50 °C, but they're mostly in deserts; much of India isn't too far off from edging into supertropical (HT), but I don't know why they're appearing in the Americas here. The hot and extraseasonal ocean spots are probably just interpolation quirks, though, where the more variable temperatures of areas modeled as land are here being interpreted as ocean climates.

 

The model is generally a bit too dry in many areas, but not consistently enough that I'd want to try to compensate by adjusting the aridity thresholds, and it's still seemingly biased towards Mediterranean climates even with the GrS threshold adjusted down (though my scheme for estimating land-equivalent evaporation for interpolated maps is a bit inconsistent in how well it agrees with nearby Mediterranean climates along coastlines). The model also isn't terribly good at replicating pluvial zones at this resolution, though the ones it does produce at least tend to be in the right areas (from the tests I've seen, T85 resolution (256 x 128 cells) seems a bit better about this; I wanted to do a T85 run of baseline Earth for this post, but my computer is struggling with it a bit so I didn't want to wait for it; I may add it in later).

 

And of course, despite the poles being too cold, there's too little ice cover, though that seems to be heavily dependent on the initial conditions of the model, as I've managed to get more realistic ice cover in the past by setting some initial global cover or keeping the model cold for a few decades to start out with, but I haven't been doing this for any of my explorations because it tends to take far longer for these models to equilibrate, and this seems to make the higher mid-latitudes even colder and drier than they should be.

 

Still, in broad terms it's a pretty decent fit; if you had no data on Earth's biomes, this really wouldn't be a bad guess, with most areas not more than one gradation in temperature or aridity away from their true climate.

(This chart may require a bit of explanation: it shows how many "steps" removed the Pasta bioclimate zones modeled by ExoPlaSim are from those on real Earth, in terms of zone transitions due to aridity for the wet-dry axis and GDD or temperature extremes for the hot-cold axis. So, for example, TUf is 1 step drier than TUr (TUr-TUf), 3 steps wetter than Ada, (Ada-TUA, TUA-TUs, and TUs-TUf), 2 steps hotter than CDa, (CDa-CTf and CTf-TUf), and 2 steps hotter and 2 steps wetter than CAb (CAb-CAa and CAa-TUA, then TUA-TUs and TUs-TUf). For simplicity's sake, pluvial zones are treated as 1 step wetter than their nonpluvial counterparts but otherwise equivalent with them for comparison with any other zones (e.g. TUfp is 1 step wetter than TUf, but TUfp and TUf are equally 1 step wetter than TUs or TUsp and 2 steps wetter than TUA); CAM is 1 step wetter than CA but equivalent to them compared to other zones; CM is 1 step drier than other other moist C zones but otherwise equivalent to them; CI is 1 step colder than CG and 2 steps colder than CF; CEc is 1 step colder than CEb but otherwise equivalent to it; GDD and temp extreme transitions are summed for thermal differences (e.g. CDa is 2 steps hotter than CEb); and I didn't count past 3 steps removed in either axis.)

 

The general pattern here is that ExoPlaSim predictions are often slightly wrong but rarely totally wrong, which is almost a bit of a happy medium from a worldbuilding perspective, in that you can be confident that the model is pointing you in the right direction—even as you go into more exotic cases where your intuitions based on Earth's climate patterns largely go out the window—but you're always justified in fudging the results a bit if you want to.

Climate Re-Explorations

With that baseline established, let's now go over the results of our past explorations and see what aspects this new system helps highlight and how it might better inform how we picture these worlds.

Temperature

This is the first set of explorations, adjusting CO₂ levels to shift Earth's average temperature.

Hotter climates

This set of course gives us one of our best examples of the utility of the new hot climate class (H), though it's notable how little the hot area expands in the first few cases: relative heating is strongly concentrated towards the higher latitudes, though despite long, warm summers these regions remain stubbornly temperate (CD) rather than subtropical (CT), as cooler temperatures and low light in winter still interrupt growth.

 

The big switch comes with the jump to 40 °C: with the poles fully thawed, additional heating is now more uniform, and so most of the former tropics shift over into supertropical (HT) or swelter (HD). Much of the higher latitudes, meanwhile, swap to quasitropical (TQ), as winters are now warm enough to avoid winter frosts completely, but again growth is still hindered by low light. But the hottest summers come in the mid-latitudes, with longer summer days and not too much cloud cover, creating a somewhat more complex gradient of climates between equator and poles.

The general sense here is a substantial switch in the types of seasonal stresses the climate places on life: at high latitudes, temperature variation is generally tolerable but there may still be cycles in activity related to lower light and long nights in winter; whereas at lower latitudes, much of life will probably need greater adaptations to heat stress, or find ways to avoid the worst stretches of summer through migration, dormancy, or annual life cycles. I also imagine the hot oceans (Oh) will have significant implications for ocean chemistry and life, but I'm not too familiar with the likely considerations there.

 

For the very hottest cases, a broad equatorial swathe is too hot for growth, potentially implying a wasteland inhabited by nothing more complex than extremophile microbes. But that's based on a fairly vague estimate of maximum growing temperatures; use more optimistic GDD limits and you'll see a more diverse range of climates emerge at the equator, with a notable proliferation of pluvial zones in the generally wet climate.

In the more habitable higher latitudes, one notable oddity here is a lack of semiarid zones (HA) between moist and semidesert (Adh) zones; because the growing season is such a short period at the coldest and dimmest time of year, when potential evapotranspiration is lowest, many areas have much higher values for growth aridity than annual aridity, so few are both above the 0.2 Ar threshold for HA but below the 0.5 GAr threshold. I'm not sure I'm totally happy with that result, but I'll leave it for now; there's a decent chance life in torrid climates (HXb) must have growth and activity strongly restricted to that short growth season, and will tend to face intense water stress due to high evapotranspiration rates regardless of soil moisture levels, so the balance of soil water may really be less important. What has worked a bit better is the distribution of parameditteranean (HM) climates; I was initially afraid these would be too common, because a dry-growth climate would imply dry winters and wet summers here, and that's usually the default seasonal precipitation cycle, but it seems that some pattern of more severe winter or spring drought is still needed to reach the necessary GrS threshold.

Colder climates

For these, I'll reiterate that ExoPlaSim has some issues modelling glacial cover and they'd likely be much more extensive in many of these cases. If anything, this is clearer in the Pasta scheme because it highlights large areas with subfreezing temperatures where ice has failed to accumulate. One of these days I may do another run at these climates, starting with a cold case and then warming up from there, which should better reflect this.

 

That issue aside, the clearest difference here is that tropical climates (TU) are much more persistent to cooler temperatures in the Pasta scheme; even though average temperatures at the equator are declining, nights still don't get particularly cold, so frosts remain unlikely.

 

There's also various nuances in the different aridity schemes: in the first few cases, the Pasta scheme perhaps better highlights how the remaining hospitable regions in Europe and central Asia are more likely to be steppe or Mediterranean scrub rather than boreal forest; and in the later cases the proliferation of equatorial Mediterranean climates in Koppen is more modest with Pasta, as many of these areas still have mild thermal seasons and so can still grow mostly in the wet season and are more likely to have some variety of savanna or steppe. The spread of semidesert (Ad) also helps indicate how milder summers may make some areas a bit moister.

 

In these last couple cases, the Pasta scheme is also better equipped to show how low average temperatures will have a different impact on the still largely aseasonal equator compared to the strongly seasonal polar regions of actual Earth. Tropical climates (TU) do finally almost disappear, but there are still large patches of subtropical (CT) or quasitropical (TQ) forest. Tundra is much less widespread than in Koppen and there are substantial areas of milder oceanic tundra (CFa); temperatures are low but steady, allowing for slow-growing grasslands or boreal forests.

Obliquity

Our second exploration, altering Earth's obliquity or axial tilt.

Lower Tilt

These cases give us some insight into how well the two systems handle seasonless climates. Again, the Pasta scheme is better prepared to handle cold, aseasonal climates, whereas Koppen defines them by summer temperature thresholds tuned to the highly seasonal climates of our polar regions. Where Koppen shows a wide tundra strip, the Pasta scheme indicates a broader region likely capable of at least slow tree growth, as well as a strip of oceanic tundra (CFa) which might support a thicker cover of slow-growing grasses and shrubs—or perhaps some form of succulent-like plants capable of tolerating nightly frosts—compared to Earth's tundra where growth can only occur in short summer bursts.

(The hot parch appears to be a single-pixel interpolation artifact; there's some isolated mountains in Venezuela that seem to cause the temperature adjustment algorithm a lot of confusion.)

 

There's also more variety shown here in semiarid climates; A strict reading of the Koppen algorithm essentially requires rainforest to transition directly to arid steppe in sufficiently hot climates, but here we see a relatively steep but still not instantaneous gradient, with many areas receiving steady but not quite sufficient rain, such that vegetation likely either has to adapt to capture what water it can from occasional storms or root deep enough to reach more reliable deep soil water sources.

Both Koppen and Pasta still have to be read with the context of the particular world in mind. Most of Earth's biomes are defined in some way by their seasonality: temperate regions are as a rule alternately hot and cold, and semiarid regions are alternately wet and dry, so life in these areas has largely adapted to take best advantage of the more clement parts of these cycles rather than the overall average conditions. I've tried to make my scheme less dependent on assumed seasonality so that we can better see what variation still exists in aseasonal climates, but the meaning of the different zones in terms of what biomes we expect to see is a bit different; rather than being able to concentrate growth in a single burst during the wet season, life in semiarid regions here has to be more adapted to conservative or episodic growth taking advantage of sporadic rain; and similarly temperate life likely consists of steadily growing evergrowths rather than deciduous trees.

High Tilt

Here we get our first good chance to look at extraseasonal zones (E), though there's a number of other notable differences between schemes. Even in the earliest cases, there's a clear increase in semiarid zones for the Pasta scheme; as the thermal seasons become stronger, this also causes greater shifts in rain patterns, and so most areas alternate between droughts and seasonal bursts of rain rather than remaining either consistently wet or dry.

 

At higher obliquities, we're used to seeing a confused mix of semiarid and continental Koppen climates in the polar landmasses, but the more robust accounting of potential evapotranspiration in the Pasta scheme makes it far less forgiving to these areas, so it's all marked as semidesert (Ad). Seasonal rains will support some life, and there are few hyperarid deserts (Ah) anywhere on the globe, but the intensely hot summers will bake the soil dry, so we're unlikely to see much in the way of forests. But there are still a fair few Mediterranean climates (XAM) on the coasts, where the summers are milder and the fall rains a bit heavier.

In the wetter climates of the lower latitudes, much as we saw with the colder temperature set, equatorial tropical climates (TU) are much more persistent here than in Koppen, but they do eventually disappear at the higher obliquities—though a few patches of quasitropical (TQ) survive even at 90° tilt in the islands of the southern ocean. Where they give way, the equatorial continents tend to become dry and cold, but coasts and islands show a temperate (CD)/boreal (CE) mix. Given the rapidly shifting light conditions and mild winters, I suspect that evergreens would tend to be favored across both climate categories.

The monsoon belt we saw in Koppen appears here as a strip of climates moister than the polar and equatorial semideserts, though there are also some pluvial areas where mountains help enhance the seasonal rainfall further. The wetter of these areas may support these worlds' only large forests, while the rest might have a mix of grasslands, savanna, and woodlands, potentially favoring migratory animals, especially in continents like Africa that span both wet belts (and for now don't have any ice in the way).

And of course, we have some extreaseasonal climates (E), though at first they're not too diverse because they appear mostly in the polar deserts. Where they do appear in wetter areas, we can perhaps presume some alternative biome of life adapted for quick bursts of spring and fall activity between periods of summer and winter dormancy, but I'll leave any more speculation about what that looks like as an exercise for the reader. It is worth noting that, like with hot climates, we have that issue of semidesert (Ad) shading directly into moist climates because GAr is so much higher than Ar, but again I'll let that stand for now, because if life has abundant water through the growing period but remains dormant otherwise, there may indeed be less of a distinct niche for life adapted to moderate but dry growing conditions.

All these trends continue into the 90 tilt cases, but one notable pattern made a bit more obvious by the Pasta scheme is that there appears to be an optimal range of temperatures and ice cover to clear the polar deserts: Too hot and the intense summer heat bakes these areas dry (and somewhat suppresses rainfall until late fall), but too cold and global precipitation is too low to keep them wet despite lower evapotranspiration; somewhere in the middle, near the limits of stability for the ice belt, summers are still severe but counterbalanced enough by heavy spring and fall rains to potentially allow for some type of thicker vegetation cover should some life be able to adapt to these conditions. 

A couple other interesting notes: first, there are some notable gaps in the ice belt where snowfall was never great enough to build up glaciers despite the low temperature; give it a few thousand years for the ice to build up and spread out from the snowier areas and these may disappear, but then again the high levels of sunlight near the equinoxes might hold it back in some areas—maybe I'll throw this is in an ice sheet simulator at some point. Second, these maps actually show a few patches of hyperseasonal pulse (Efb), where spring and fall growing periods are too short for tree growth, a condition I hadn't been sure would be possible (because it implies such a rapid jump from below 5 to over 50 °C) but seems to occasionally turn up near the poles.

Altogether, this is one of the primary cases I wanted my classification system to better represent, and I think overall you can see that it's a bit more robust in reflecting how these greater temperature extremes affect both growth patterns and aridity, and better distinguishes the different climate regions of this world which might foster radically different types of life and societies.

Day Length

Our third exploration, and also a somewhat tricky case for handling sampling periods.

Shorter Days

The first couple cases here don't look too different between schemes, but the increased hot (H) and extraseasonal (E) climates help emphasize how the weakened equator-pole circulation isn't just making the polar regions colder, but also makes the equatorial regions hotter, especially around the edges of the expanded deserts. 

 

The Pasta scheme also seems to predict notably fewer mediterranean zones, perhaps as a reflection that it's a bit more flexible about how it counts growing and non-growing periods rather than having to split the year into a summer and winter half like Koppen, and so better suits these areas with fairly mild seasonal temperature variation.

 

The final case with 3-hour days is about as much as a confused mess in both schemes, with the Pasta system mostly serving to give more variation to that confusion. In the tropics, cloud cover is so thick and consistent in some areas that it reduces sunlight on the surface enough to potentially inhibit photosynthetic production, creating the quasitropical (TQ) patches, which then shade to eutropical (TU) and hot (H) where the cloud cover is thinner.

 

At higher latitudes, there's a very steep temperature gradient at the edges of the tropics which shifts north and south with the seasons: areas passed over by this gradient have extreme temperature swings and so show patches of extraseasonal  climates, (E) while areas that manage to stay just clear of that gradient have much milder winters and show more quasitropical climates (TQ). At higher latitudes, much as with the high-obliquity cases, the Pasta scheme's more reliable accounting of PET is less forgiving to regions with intense seasonal variation and marks large variously arid regions, though as with Koppen there are few true deserts (Ah).

Longer Days

First off, a few procedural notes here: as explained in the original post, the way ExoPlaSim collates temperature data means it's not always great at recording temperature extremes for days 5 times longer than Earth's or greater, so I made an extra set of outputs for these runs with shorter 120-hour output periods. To read climate zones from these outputs, we can either read them directly with month lengths adjusted accordingly for the shorter output periods, or we can bin them together into regular-length months. But when I tried these two methods I found a surprising difference in the results even though I wrote koppenpasta to preserve the temperature extremes when binning months together.

 

This seems to come down to subtleties in how the temperature data is interpreted: ExoPlaSim samples the maximum and minimum temperatures from surface temperature, but for bioclimate classification we're more interested in near-surface air temperature (because that's what will most directly affect vegetation), so to account for this I find the difference between average surface temperature (ts in the ExoPlaSim data) and near-surface air temperature (tas) in each month and adjust the temperature extremes accordingly. The issue is that it looks like with month-long days there can be a fairly substantial difference between ts and tas, by as much as 30 °C hotter or cooler, larger because tas lags behind ts, rising later in the day and falling later at night (whether this is down to cloud cover, air circulation, or whatever else, I'm not sure). When each 120-hour input period is treated separately, the temperature extremes are adjusted to account for these large differences; when they're binned together into full-length months, these differences largely cancel out, so the extremes aren't adjusted much. These temperature extremes are then used for estimating potential evapotranspiration, so a greater maximum temperature causes greater estimated PET, resulting in a lower aridity index and so more arid zones.

 

At any rate, I've decided that directly working with the shorter inputs is probably the better representation here.

 

In broad strokes the two schemes are largely in agreement, though for once equatorial tropical climates disappear quicker in the Pasta scheme, reflecting increasing daily temperature range. In detail, there's a curious mix of climates like quasitropical (TQ) reflecting very low temperature variation and climates like extracontinental (ED) reflecting much greater temperature variation, often not far from each other, which perhaps reflects a regime where daily temperature variation is a much greater factor in determining the total temperature range, and is in turn more strongly influenced by distance from the sea compared to seasonal temperature variation.

 

The Pasta scheme also perhaps emphasizes a bit better the contrast in aridity between the wet tropics (though pluvial zones are rare) and the high-latitude deserts, much of which are hyperarid (Ah) despite lower average temperatures.

 

Curiously enough the area of extraseasonal climates peaks in the 10x case and declines for the 30x case, probably because Coriolis forces are now weak enough to allow for formation of a dayside cloud formation that moderates daytime temperatures. The classification of the moist forest belt in both cases is still largely determined by the lowest nighttime temperatures rather than seasonal patterns: a few remaining patches of tropical (TU) shade to subtropical (CT) and then directly to continental temperate (CDb) due to cold winters rather than growth interruption, with very few areas of oceanic temperate (CDa), and boreal (CE) also largely restricted to a few highlands. The general picture is that seasons are too gentle to pose much of an issue for vegetation (aside from perhaps at higher latitudes that are too dry anyway) and the main challenge is surviving the nightly chills.

 

The pattern of peculiarly dry coasts is retained, perhaps even made clearer by the greater range of semiarid climates—though the Mediterranean climates seen in Koppen don't appear as much here in the very moderately seasonal climate.

Eccentricity

Our fourth and most recent exploration.

Another good look at extreme seasonality. The differences are pretty much immediately obvious; whereas the Koppen map for 0.3 eccentricity looks surprisingly like that of real Earth at a glance, the Pasta scheme clearly shows that seasons here are far more extreme, with a supertropical (HT) strip of hot but brief summers across the low latitudes and then belts of extraseasonal (E) climates at mid latitudes where any vegetation would need to tolerate both winter frosts and intense summer heat stress; only near the edges of the polar tundra do cool forests appear. Seas, too, might be a bit harder hit by intense summer heat along the equator. On the other hand, this greater seasonal variation does seem to reduce the area of hyperarid deserts (Ah), thanks perhaps to the globally wet autumn.

At 0.4 eccentricity, Earthlike climates have practically disappeared, save for thin polar strips—though even these areas are largely dominated by pluvial steppe (Xaxp), a rare climate on Earth which here seems to reflect summer droughts and wetter (or snowier) falls and winters; usually we might expect that to be classified as Mediterranean, but here precipitation rates don't actually decline much in summer, the drought is driven by much higher evapotranspiration from the intense heat; so while these areas may resemble Mediterranean biomes in some ways, there may not be as much reliance on deep soil water or streams (they may perhaps more resemble the forest steppe of European Russia, which is wetter overall but has a similar pattern of higher evapotranspiration but steady rains through summer).

 

At even higher eccentricity, the Pasta scheme seems to get a bit simpler than Koppen, because what Koppen classifies as a range of continental climates, Pasta's more robust accounting of evapotranspiration again sees mostly as semidesert or steppe. Still, there are some patches of wetter climates where the early and late summer rains might encourage some greater density of life if it can survive or avoid the summer heat, which by this point reaches the boiling point in some areas. Even outside these zones, true desert (Ah) is rare, with most areas drying out in summer but getting enough rain thereafter to likely support some ephemeral vegetation, and there are even larger regions of the steady-precipitation, variable-evapotranspiration pluvial steppe (Xaxp), though also some proper Mediterranean (XAM) where the precipitation is a little more variable and percontinental rainforest (CEcp) where summer heat is lower but winters are bitterly cold.

Rather than just a moderately rearranged set of Earthlike climates, the picture we get is a more radically different world: an equatorial strip of wet climates that must adapt to a flash of summer heat; vast arid to semiarid regions where life may be more ephemeral but has opportunities to fluorish in early summer and the long, moister fall; and vast subpolar woodlands where the seasons are still extreme but more tolerable by sufficiently hardy vegetation.

Tidal-Locked Earth

To test out how well this new system works with tidal-locked worlds, I did a couple quick, 20-year runs with Earth tidal-locked to the sun and all other conditions the same as baseline; I'll do a proper exploration of tidal-locked worlds eventually, with models run fully to balance, this is just a short preview to get a sense of how Koppen-Geiger and the Pasta system compare in these conditions. For one, I set the SSP at 0° latitude and longitude, in the Atlantic just off the coast of Africa.

For the other, the SSP is shifted to 180° longitude, in the middle of the Pacific.

 

These worlds are too cold for us to see any warm dark climates (the TF, TG, and HG appear to be single-pixel interpolation artifacts, and the patchiness of nightside ice cover probably just reflects the short runtimes), and even on the dayside most cooler areas are desert, but there are some patches of quasitropical (TQ), aseasonal oceanic temperate (CDa), and oceanic tundra (CFa).

 

Pasta's aridity scheme allows for smoother transitions from rainforest to desert here than Koppen, but if anything that just emphasizes how stark they still tend to be; most areas either have plenty of rain or none, and the storm region around the SSP is clearly marked by a blotch of hyperpluvial rainforest (TUrp). I sometimes hear concerns that heavy cloud cover could inhibit growth in this area, but at least in this case that doesn't seem to be borne out; average light levels here are a tad lower than in the nearby deserts, but don't reach that ~200 W/m² threshold where we'd expect it to begin slowing photosynthesis. Given that the top-of-atmosphere irradiance is over 1,000 W/m² for much of this area, clouds would need to block over 80% of the light to reach that level, and they're just not quite that consistent.

 

I've mentioned before that some research indicates that shifts in the carbon-silicate cycle could cause substantial shifts in CO₂ levels and global temperature in relation to land distribution near the SSP, but I haven't attempted to model that here; both these cases were run with 300 ppm CO₂. As is, both cases end up with fairly similar temperatures, with the atlantic SSP case 3 °C hotter on average than the pacific SSP, which probably speaks to how much the global albedo is controlled by the dayside cloud formation rather than surface characteristics.

Teacup Ae

To wrap up, here's a quick preview of a recent model run I did with some WIP terrain for Teacup Ae (yes I'm still working on that, it's been slowgoing) at T85 resolution.

 

One clear trend is that the short Teacup years (about 1/3 of Earth's) give us short growing seasons and so large stretches of tundra. We could perhaps compensate with somewhat increased photosynthetic productivity due to Teacup Ae's higher CO₂ levels, but that seems to result in temperate (CD) climates largely disappearing and subtropical (CT) shading directly into boreal (CE) or tundra (CF), which I'm not sure I quite like. Of course, ExoPlaSim represents a fair bit of Earth's CD as CE or CF, so we can perhaps just feel justified in imagining there's a bit more forest cover than implied here, at least on north Hutton (the middle upper landmass). But we could also accept that the oceanic tundra (CFa) corresponds to vast subpolar grasslands, somewhat like the mammoth steppe of Pleistocene Earth.

All in all, the Pasta scheme isn't without its quirks in some of the more extreme cases and there's still an extent to which you need to interpret it with the context of the particular conditions of a world in mind without being able to read it "blind" in all cases; and it is not free of certain assumptions about how alien life might adapt to these different planets, largely because there's really just no way to map worlds in this way without such assumptions. But compared to Koppen-Geiger, in addition to being a bit more reliable at reflecting true conditions of aridity, seasonality, and temperature extremes in the more exotic cases, this new scheme also tends to give us a better sense of the range of climates present on a single world, dividing areas with vastly different seasonal patterns or extreme conditions that Koppen lumps together.

 

In the future, I'll probably switch to using those 3-map plates as my default for climate explorations: uninterpolated maps for both Pasta and Koppen-Geiger at the model resolution and the interpolated Pasta map with a full key. Now that we've worked out this scheme I should have another of those ready fairly soon (combining obliquity with eccentricity for hemispherically asymmetric seasonal patterns), though I also want to take a look at some of the public climate data I've explored in the past; based on some preliminary testing, I'm fairly confident that I can estimate most of the necessary parameters from standard ROCKE-3D output parameters.

For today, I'll end up with a couple quick updates: first, I've now opened a discord server for my supporters on Patreon; for now it's patron-exclusive, so you should be able to join if you sign up for at least the minimum rate (around $1 USD I believe) and link your discord account to Patreon.

Second, I recently had a short story of mine, Momentum Exchange, published in the May/June 2025 issue of Analog Magazine. It's a little scifi story set on a planet with an unusual orbit, check it out if you like.

Buy me a cup of tea (on Patreon; and join the Discord!)