So I see this comment come up a lot. I have done a ridiculous amount of experimentation with FT3, mostly trying to come up with different methods of getting legitimate deserts onto my world. I've learned all sorts of things about axial tilt, as well as all the stuff that affects climate.

Before I continue on with a more general comment, I want to put up here one thing I came up with which I found intriguing. In doing the Israh tutorial, I noticed that the author often uses quite ridiculous numbers to achieve their effects, so I thought, why not try that with precipitation. Taking the earth data map provided with FT3, I focused on trying to get Africa to look like it does on the real world (because my world has a similar land mass). I settled on putting the precipitation amount at 0, and the random figure at 1000. This produced a reasonably similar Africa, but dried the Amazon rainforest out. What are you going to do?

Anyway, the point is, using 0 precipitation and 1000 random will get you small deserts, ranging from decent deserts to mere strips of aridity. If you then play around with the setting, you'll be able to get those deserts to shrink or grow. One thing that's great about using this setting is you're also more likely to get evergreen rainforest, which is also generally lacking in the climate model. It does reduce temperate grassland quite a bit, unfortunately. One thing I did like about it is it put a desert in a place that really surprised me, actually along the equator. When I looked at it more closely, I realized that it was actually a reasonable place for a desert, since there was a lot of mountains in the region that would cast rainshadow. I liked that because the deserts weren't just in all the obvious places, just as they aren't on earth (Somalia/Ethiopia, for instance).

The other difficulty with producing deserts is of course temperature, and I came up with an interesting idea based on axial tilt. From my research, I determined that January is 7% hotter than July, so I altered the sun setting in temperature to 1.035 and 0.965 to create a temperature map for both months. It doesn't sound like a lot, but it had a profound effect on the world's temperatures (5 C at the equator). I then varied the random category for precipitation to 750 in January and 1280 (I think) in July, reasoning that there would be more rain during the cold time of the year. It then occurred to me that for January you should really use the July map in the northern hemisphere and the January map in the southern.

Then I had kind of a brilliant idea: why not create a dry map for the deserts of your world? You could maybe have 40 precipitation with the usual variation. Wherever this resulted in desert, you could use the values generated on the dry map and transfer them to a master map. You could do the same with the ITCZ , ocean currents, whatever.

One final thing: maybe its just for my map, but I've actually found that the FT3 climate model is fairly accurate if you just look at which areas it predicts will be dry or wet, cold or warm. The trouble is that the range of values in both cases is too narrow. For instance, my world's temperature ranges from about -10 at the poles to around 29 on land at the equator. It should really be something more like - 58 to 34 (for regions where there is human habitation).