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

[Charerg]

Looks pretty good. There is just one potential issue that pops up regarding the classification into arid climates: the calculation of the aridity threshold.

Remember that the aridity threshold uses different criteria for choosing which threshold to use than the f/w/s used in the climates, at least if we follow Kottek. As a reminder, climate group w needs 10x the rain in summer compared to winter, but for the "aridity group W", just 2x is enough (or 2/3 of the annual rainfall, to put it differently)! So, you need to also specify which "aridity group" each precipitation combo belongs to. Of course you could use the climate groups instead if you want to simplify things, but then you might end up with many BS climates being pushed into Cf or Df because they use the lower "aridity group F" threshold. And I do recall several areas being a bit too humid in the latest script, so that is definitely a potential issue.

[Azélor]

About the excel spreadsheet of the world. Here's the explanation of the process starting with the precipitations:

Convert the main data from January and July to an hex code
using a table were each value in mm is tied to a hex value.
Each hex value represent a category

Example:

Attachment 104225

Then, we combine the values, one is red and the other is blue.
Convert these to a green value, one for each category. So far this is almost identical to the already existing tutorial.
Precipitations done for the moment.

---

Temperature

Same as with precipitations, we take the base data and give them hex values.
Combine both months.
Then convert temperatures to reduce the number of possibilities.
That's it for the temperatures.

---

Now we need to combine precipitation and temperatures, with an addition.
When that is done, we have the 500 something combinations.
Whit what I've explained in the previous message, we know what output give what climate, we just have to replace the values with the ones in the table and we get something like this once we format it properly
https://en.wikipedia.org/wiki/K%C3%B...h_authors).svg

---

It is still unfinished but so far it's seems to work. It is slow and I'm using the data at 10m resolution. It would be unthinkable for me to increase the resolution. We most of the time if only gets sluggish if there is missing data in the table as Excel tries to find something that does not exist, in 1 million of cells at the same time. Having good table is crucial. right now, each file has it's own table. I find it easier to work that way but I will have to link them to a master table so if I change the master, all the others will change to.

I haven't done extended test but it looks like changing the data in the categories, such as changing the temperature boundaries, works fine. and it's not too difficult as long as the tables include the new values.
It's harder to add new categories as the links between tables need to be added, but I haven't tried it.

For the moment, I will concentrate on finishing the generation of the climate world map based on the criteria we have settled for and we shall see the results.
One of my goal is to allow the modification and testing of different parameters without having to select a bunch of colours in Gimp/Ps

So this is just a first step.
Later, I will create another version that should be more flexible by including all the possibilities. It's an complete/unsimplified version if we can call it that way. Yet, it fuses the redundant temperature combinations such as Hot/cold and Cold/Hot since they are actually the same thing. Also simplify the tundra and polar climates so we only keep one of each. Lastly, it will only keep on Da and one Dd from the deadly cold winter column since they are either impossible or always wet unless then precipitations gets absurdly low. Even if we divided the precipitations by 2, they would still be humid. I also consider fusing together the tow Dc from the severely cold column but it might be affected if we decide to lower the precipitations.

It uses the same principle but because it doesn't uses precipitations categories (actually, I still need to convert the value to green, but they are all unique) , it is easier to use for the sake of testing and seeing the results (I assume).
The only drawback maybe is that is results in a lot more combinations in the end. Not a problem for Excel as long as all possible results are covered in the tables, but a problem if we want to use that in the tutorial.
With the actual number of categories, I get around 1298 combinations?

[Azélor]

Looks pretty good. There is just one potential issue that pops up regarding the classification into arid climates: the calculation of the aridity threshold.

Remember that the aridity threshold uses different criteria for choosing which threshold to use than the f/w/s used in the climates, at least if we follow Kottek. As a reminder, climate group w needs 10x the rain in summer compared to winter, but for the "aridity group W", just 2x is enough (or 2/3 of the annual rainfall, to put it differently)! So, you need to also specify which "aridity group" each precipitation combo belongs to. Of course you could use the climate groups instead if you want to simplify things, but then you might end up with many BS climates being pushed into Cf or Df because they use the lower "aridity group F" threshold. And I do recall several areas being a bit too humid in the latest script, so that is definitely a potential issue.

I thought about it. I have 3 different columns, one for each.
As to how one knows if the climate is dry in the summer, in the winter or neither: there is no formula (yet?)
You need to change these tables manually, that is where the data from the "cd climates" come from:

sheet 37.png

the issue I have with the s,f,w table is that because I'm using 6 categories maybe, following the strict formula doesn't give as a good climate distribution (based on experimentation in the past).
of course, I don't have that issue much with the A climates and could probably automate that part since I already have the formula (and it works), I just need to change the output.

I take back what I said about the A climates

[Charerg]

I thought about it. I have 3 different columns, one for each.
As to how one knows if the climate is dry in the summer, in the winter or neither: there is no formula (yet?)
You need to change these tables manually:

sheet 37.png

the issue I have with the s,f,w table is that because I'm using 6 categories maybe, the formula doesn't give as a good climate distribution (for experimentation in the past).
of course, i don't have that issue much with the A climates and could probably automate that part since I already have the formula (and it works), I just need to change the output.

Good to see that you have it handled! I'm not sure about what you meant with the formula, but if we follow Kottek's definitions, it would be the following for determining which aridity threshold belongs to each prec combo:

• W threshold: at least 2/3 of annual prec in summer (summer ~67% or more)
• S threshold: at least 2/3 in winter (summer ~33% or less)
• F threshold: other cases (summer rainfall 33-67%)

[Azélor]

S: the precipitations in winter must be at least 3 times the precipitations in summer
W: the precipitations in summer must be at least 10 times the precipitations in winter
F : whaterver's left

results, the new of the left and the old on the right:
sheet 38.png

We already know that expanding the s is a bad idea. I don't know about the changes to w. The distribution looked ok last time.
The 2 w missing are 1-2% lower than the target. They need 91%, one is at 89% and the other at 90%. Considering we are using the average, in reality, about half of them would be classified w and the other f.

[Azélor]

I got the results. I have to admit that is was really cool to see the map unfold.
But there are several problems as you can see:

sheet 39.png

The most obvious ones would be:
many w climates are categorized as s in Brazil for example
Af and Am climate look way off

Actually, I think the problem is in the method we use. The north and south end up having different colours for the winter/summer precipitations. They have the same her and so are not categorized correctly.
I guess that is the problem since most of these areas are in the south.

Also:

there are As climates between the desert and the steppe
Steppes are too common in Europe and in Labrador.
in Asia, there are w and s climates right next to each other
A desert in Vietnam

These are the most obvious.
I know there is missing data in a few places. This might explain some weird things like the steppe in Labrador.
I noticed that in Qgis while looking at a 0 in Alaska. That was the temperature and it really popped out since it was surrounded by -18, -19.
lastly, the map uses the 10m resolution, which is the lowest

---

I just noticed a few places in Europe with steppes on your map. Mine is worst but I was not expecting this.
Vietnam is also shown to have steppes, strange.

[Azélor]

Fixing the southern hemisphere problem is easy. I just need to add a formula that will pick the red and blue code seperatly and reconstruct the code by switching these values.

Fixing the other cases will need more thoughts.

[Charerg]

I just noticed a few places in Europe with steppes on your map. Mine is worst but I was not expecting this.
Vietnam is also shown to have steppes, strange.

Probably just a result of having data from 2 months (as well as a discrete number of categories). Those areas in Vietnam are in rain shadow during the peak of the monsoon season. Also, the steppe region in Myanmar is probably genuine since they appear in the Hans Chen Köppen map too (though you have to zoom in to notice it). For the rest, difficult to say, since there aren't any high-res Köppen maps of SE Asia around the web. On a quick search I was able to find this one, which includes the steppes in Myanmar too, but nowhere else. So, I'm inclined to think that the steppes in Vietnam probably appear due to the large data gap from using data of 2 months only.

Similarly, the BS areas In Europe aren't too far off the mark, there are BS climates in Spain, Greece, and even a bit around the Black Sea. Areas like northern Bulgaria are probably "borderline BS" too, since that was a favoured migration route for steppe nomads in the past. Although the BS is a bit too extensive in our case, but again this is probably caused due to data limitations (discrete categories and only two months). It's worth noting that the BS distribution is better with the 8-step map in Europe, so in this case I suspect the discrete number of categories to be the issue.

One problem that also turns up in your map is those anomalous northern s climates, they definitely seem out of place.


Edit:
Oh, and a bit more ranting about the difference between the climate and aridity categories:

S: the precipitations in winter must be at least 3 times the precipitations in summer
W: the precipitations in summer must be at least 10 times the precipitations in winter
F : whaterver's left

Those are the definitions of the climate categories (s, w, f). But the aridity categories (S, W, F) use a different definition! So, a climate that is in the "climate group f" could belong to any of the aridity categories, because the criteria are not the same.

Here's an example:

Summer rainfall 100 mm
Winter rainfall 30 mm

This would be classified as an f climate (since ~77% of the annual prec falls in summer), however it would still use aridity threshold for the "aridity group W", because more than 2/3 of the rain fall in summer!

So, the threshold would be calculated as follows:

Annual mean temp 15 °C
Prec threshold 20*T_ann+280 = 580 mm


On the other hand, if the precipitation distribution was different:

Summer rainfall 70 mm
Winter rainfall 60 mm
Mean temp 15 °C

Again, an f climate, but this time the rainfall is spread more evenly between the seasons, and it belongs to the "aridity group F":

Prec threshold 20*T_ann+140 = 440 mm


And as a final example, the following:

Summer rainfall 40 mm
Winter rainfall 90 mm
Mean temp 15 °C

Another f climate, but now over 2/3 fall in the winter, so this would belong to the "aridity group S":

Prec threshold 20*T_ann = 300 mm


Note that all of the above examples were classified as f climates, but the prec thresholds used were very different.

[Azélor]

I'm just using the numbers from the paper you keep talking about. I don't know where you see 2 sets of different numbers.

Take Csa for example:

for the precipitations, winter must be have at least 3 times the precipitations of the summer
If the difference is smaller, then it's not a Cs climate.

(since ~77% of the annual prec falls in summer)

That would clearly fall in the w climates, like Cwa, according to the definition. There is a rainy season and a dry one. If it was Cfa, it would be spread more evenly.
You talk like I'm missing something really obvious.

Maybe it's this? https://en.wikipedia.org/wiki/Aridity_index

if rainfall occurs mainly in the hot season.

What does mainly means to you? I associated that with the w climates.

So if: the precipitations in summer are at least 10 times the precipitations in winter.
the climate is rated w because rainfall occurs mainly in the hot season, that's how I defined it.

[Charerg]

I'm just using the numbers from the paper you keep talking about. I don't know where you see 2 sets of different numbers.

That would clearly fall in the w climates, like Cwa, according to the definition. There is a rainy season and a dry one. If it was Cfa, it would be spread more evenly.
You talk like I'm missing something really obvious.

Maybe it's this? https://en.wikipedia.org/wiki/Aridity_index

What does mainly means to you? I associated that with the w climates.

So if: the precipitations in summer are at least 10 times the precipitations in winter.
the climate is rated w because rainfall occurs mainly in the hot season, that's how I defined it.

That's the thing, the aridity threshold is not synonymous with w or s climates. And the Kottek et al. paper does indeed define when to use which threshold, check page 4 of the pdf:

The annual mean near-surface (2 m) temperature is denoted by T_ann and the monthly mean temperatures of the warmest and coldest months by T_max and T_min, respectively. P_ann is the accumulated annual precipitation and P_min is the precipitation of the driest month. Additionally P_smin, P_smax, P_wmin and P_wmax are defined as the lowest and highest monthly precipitation values for the summer and winter half-years on the hemisphere considered. All temperatures are given in °C, monthly precipitations in mm/month and P_ann in mm/year.

In addition to these temperature and precipitation values a dryness threshold P_th in mm is introduced for the arid climates (B), which depends on {T_ann}, the absolute measure of the annual mean temperature in °C, and on the annual cycle of precipitation:

P_th=
2*{T_ann} if at least 2/3 of the annual precipitation occurs in winter,
2*{T_ann}+28 if at least 2/3 of the annual precipitation occurs in summer,
2*{T_ann}+14 otherwise.

Note that since BS is defined as P_ann<10*P_th and BW as P_ann<5*P_th, you should multiply the above with 10 to arrive with the final threshold. So you end up with 20*T_ann+x formula.

Btw, the wikipedia page for the aridity index has the exact same formula (although in cm instead of mm, so it's already been multiplied by 10). Although you're right in that wikipedia does not define exactly what the "mainly in summer/winter" means. Although the B climate class page in wikipedia does provide that information (in percentages).