The Köppen–Geiger climate classification made simpler (I hope so)

[Charerg]

The result

Attachment 104269 vsAttachment 104270 posted earlier

The version posted earlier still had that problem were increased precipitations resulted in increased aridity. The new version does not but it has 25 thresholds instead of 3.
Now I just need to find how to make this work without creating a bunch of complicated things.

I'm not sure how good an idea it is to use the gradual threshold. Sure, the threshold will be more gradual, but is that necessarily a good thing? A lot of areas will become "more humid" as a result of this, since like 80% of the planet receives more than 2/3 of the total rainfall in summer. And I seem to recall we had the opposite problem of many such regions being classified as too humid to begin with?

Personally, I'm ok with the standard criteria used with the Climate Classification, after all we're not seeking to create a new or updated version of Köppen, so I'm reluctant to change the criteria (especially since it makes generated maps difficult to compare to real ones).


Also, in your latest temperature sheet, you shouldn't merge the lower right corner of Ca with the rest (because it has less than 18 °C annual mean, so Bk instead of Bh).

[Azélor]

Yes I changed it after but did not update the last picture.

So the final count is 552 combinations.

Some areas become more arid, other more wet. As I said, the logic sounds right to me but I don't know if it will make the map better.
Don't forget that we use categories and Koppen used continuous data. We need to consider that.

result : multiple thershold.png

I also just tried the improved 3 thresholds method but I must have done something wrong. It looks horrible.
I tried to fix the errors but ended up creating new ones. There is now a large desert in Siberia. I really have no idea what I have done wrong.
Also, there is a steppe in Alaska and I don't think it is an error. After all, the maps you linked showed steppes in Alaska.

I think I've found what was causing the problem.

[Azélor]

now the results:

3 threshold.png

Do you want the files?
I can send you the multiples threshold version, the 3 threshold version or both.
They are 150mb each uncompressed.

[Azélor]

I also did a quick test (updated the file mostly)
to include a 8 categories precipitation model. Right now, it has 1022 combinations.

Having more categories makes it harder to combine stuff together. As the model is supposed to be more accurate, simplification would go into the other direction.

[Charerg]

now the results:

3 threshold.png

Do you want the files?
I can send you the multiples threshold version, the 3 threshold version or both.
They are 150mb each uncompressed.

Yeah I'd like to update the script to use the 3 threshold version since that's the "standard criteria". And it does seem to overall work better, apart from that somewhat anomalous Dsa in Kazakhstan.

I also did a quick test (updated the file mostly)
to include a 8 categories precipitation model. Right now, it has 1022 combinations.

Having more categories makes it harder to combine stuff together. As the model is supposed to be more accurate, simplification would go into the other direction.

I think 1022 is still manageable, provided that the final climate table can be compiled in Excel, it's possible to copy-paste the final hex codes into the appropriate climates in the script without too much trouble.

[Azélor]

Yea, that area is odd. Depending on the exact precipitation, it could be that while they are almost identical, they use different thresholds.

That is a strange thing, the resulting code of these Dsa is not in the database. I must have made a mistake while converting the temperature values (simplification). But the temperature should have been converted to another value, and that value is a Dsa climate too. After verification, the data looks right. They are supposed to be humid but barely. The precipitation is 135mm, threshold 100mm. Seems fine just by looking at the numbers.
I admit that I haven't checked the surroundings.
The exact data is for the cell BDG256:
Temp: Jan -12 / Jul: 26 (Da)
Prec: Jan: 12mm / Jul: 7 mm

Reducing precipitations in any way result in a steppe.
Increasing them in summer and it also become a steppe because of the changing threshold.
In creasing precipitations in winter makes it wetter.

It is possible to have all the hex codes, the corresponding climates leters and resulting colour scheme on a single page. I called it final simple (I know I'm not good at finding names).
Paste the result right next to it. Deleted all the N/A having codes with only 4 characters. Delete the remaining duplicates. Put them in alphabetical order and your done. The lettering is akward sometimes but it should be hard to figure it out.
Ex: "Ca s" instead of Csa

[Charerg]

Yea, that area is odd. Depending on the exact precipitation, it could be that while they are almost identical, they use different thresholds.

That is a strange thing, the resulting code of these Dsa is not in the database. I must have made a mistake while converting the temperature values (simplification). But the temperature should have been converted to another value, and that value is a Dsa climate too. After verification, the data looks right. They are supposed to be humid but barely. The precipitation is 135mm, threshold 100mm. Seems fine just by looking at the numbers.
I admit that I haven't checked the surroundings.
The exact data is for the cell BDG256:
Temp: Jan -12 / Jul: 26 (Da)
Prec: Jan: 12mm / Jul: 7 mm

Reducing precipitations in any way result in a steppe.
Increasing them in summer and it also become a steppe because of the changing threshold.
In creasing precipitations in winter makes it wetter.

It is possible to have all the hex codes, the corresponding climates leters and resulting colour scheme on a single page. I called it final simple (I know I'm not good at finding names).
Paste the result right next to it. Deleted all the N/A having codes with only 4 characters. Delete the remaining duplicates. Put them in alphabetical order and your done. The lettering is akward sometimes but it should be hard to figure it out.
Ex: "Ca s" instead of Csa

Shouldn't the threshold be 140 mm?

If you calculate the avg. between Jan -12 and Jul 26 you end up with T_ann 7 °C, so the threshold should be 20*T_ann=140 mm.

[Azélor]

In theory yes, and it would then be a steppe but we use categories.

Therefore the average temperatures are -16 and 26: so 5 for the year.
So the result is 100ml or actually 80mm because the two categories are put together.

[Charerg]

In theory yes, and it would then be a steppe but we use categories.

Therefore the average temperatures are -16 and 26: so 5 for the year.
So the result is 100ml or actually 80mm because the two categories are put together.

So I guess the problem is too broad prec categories (and perhaps the winter temp averages are too low?). Since (12+7)*6 should result in 114 mm annual prec. But instead we end up with (17.5+5)*6 = 135 mm. Okay, that didn't make a difference. So probably we should use higher averages for the winter temperatures. We're using -4 for Cold, right? That's probably ok, but maybe Very Cold should use -12 average to prevent it from dropping the annual mean temperatures into unrealistically low values?

[Azélor]

If it increase the aridity there, it will also increase it somewhere else.
-12 will be too low for some places.

Edit: I mean too high. Since -12 is higher than -16 for example. Oups