I am currently learning to use the annual water yield model and collecting data.
For the plant available water content (pawc) data, I used the ISRIC with all different depths. I am not sure about how to calculate the weighted average value among all the rasters. Could someone help me give it a check?
So, my understanding of the instruction is with those 7 rasters
following the formula of the weighted arithmetic mean,
My calculation would be like, for example at 5cm depth,
Also, if this logic is correct, then the raster at 0 cm will be excluded from the calculation. It is necessary to add it in? If so, how?
Thank you for helping!
Hi @urgendalai -
I haven’t used these ISRIC data, but I’m pretty sure that calculating PAWC from different depth information would go something like this:
1/ For each depth layer, calculate
((pixel value) * ((depth of layer) / (total depth)))
to get the weighted PAWC value for that layer.
2/ Add together the weighted layer values calculated in #1.
Each layer represents a range of depths, so the layer s_WWP_M_sl1_0cm.tif represents the depth 0-5cm, so you’d use a layer depth value of 5. The next layer down is from 5-15cm, so you’d use a layer depth value of 10, etc. I’m not sure what the total depth is here, it seems like it should be greater than 200cm, but taking a quick look at the ISRIC documentation, they list the range information for the first 6, but not the last one (200cm). So I’d look for documentation on these layers to see if there’s a particular depth defined for the bottom layer (likely not, since every location will have a different maximum depth, often much less than 200cm, often much more), and if not, you could use a soil depth layer to find the maximum depth.
Thank you very much for your reply, @swolny. Upon checking with their Q&A section, I think the 2017 version (not the newest one you read) did define standard soil depth at 7 numeric values instead of intervals. Seems they derived six interval standards from those 7 values in the new version.
However, the PAWC data has not been converted to the new standard. The total depth is 200cm.
It still remains the soil surface (0cm) that I do not know how to deal with. Should I just give it a very small value such as 0.001 as the weight to preserve the data in the calculation?
Good question. If it really is just a very thin slice of the soil surface, honestly I’d be tempted not to include it in the calculation at all, since that’s not likely to contribute much, if anything, to the overall water content. Using a very small value for the weight seems fine too, again that won’t contribute much to the result. Sorry I don’t have time to dig trough the ISRIC documentation right now, but maybe there’s somewhere (in a more detailed user guide?) that they explain the 0cm layer a little better.
@swolny Sorry to bother you again but I would confirm again about the total depth in the calculation. Shouldn’t it be the sum of all depth? Which is (5+15+30+60+100+200)=430? The maximum depth is 200 cm but I think it is not the total depth in the weighted calculation.
Also, thank you for the advice on the 0 cm layer. I did both calculations with and without it and the results are not affected so I agree with you to simply exclude it.
For the layer depths, we’re calculating the proportion of each particular layer’s depth relative to the total soil depth. If I’m understanding the file naming correctly, each layer’s depth would be as follows:
sl1_0cm: 0cm depth (or some tiny number, as noted previously)
sl2_5cm: 5cm depth (the layer itself goes from the surface to 5cm belowground)
sl3_15cm: 10cm depth (5cm to 15cm belowground)
sl4_30cm: 15cm depth (15cm to 30cm belowground)
sl5_60cm: 30cm depth (30cm to 60cm belowground)
sl6_100cm: 40cm depth (60cm to 100cm belowground)
sl7_200cm: 100cm depth (100cm to 200cm belowground)
These layer depths add up to 200cm, which you confirmed is the maximum soil depth.
So the weight for layer sl2_5cm would be 5/200, the weight for sl5_60cm would be 30/200 etc.
Thank you very much for your help @swolny !
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.