try topographic wetness index in place of elevation & distance to stream

[teixeirak]

@mcgregorian1, topographic wetness index is a metric that can be calculated from DEM data, and would likely be a better metric than elevation of distance to stream. If we could get that without too much effort (there's some R package), it would be great to look at that instead.

[teixeirak]

A colleague here at the workshop who's using this gave me this info: "I choose the R package"dynatopmodel" to calculate TWI from DEM data. The function "build_layers" could easily extract 5*5 meter quadrat's TWI with 5m DEM data. However, I have not compare the result with the other method."

[mcgregorian1]

Hi @teixeirak

It took a bit but I managed to get a DEM of our plot, which yields a TWI map looking like this. I'm a little confused by it since it seems to disregard the elevational gradient of the deer exclosure. Also, notice the difference between the TWI when I specify 20 cols x 31 rows (top) and 21 cols x 33 rows (bottom). I've checked the source code for the function but I didn't notice anything that jumped out at me as being the culprit.

I think from first glance I should use this one (21cols x 33 rows), since it includes the whole plot.

For reference, this is what the sample code for the package yields:

[teixeirak]

Thanks for looking into this! Are the blanks in your maps missing values or off scale? (I assume probably the former.) I wonder why they're missing. Perhaps the spatial scale is too course? What was the resolution/ source for the DEM? (See this issue.)

[mcgregorian1]

Ok @ValentineHerr has helped figure this out. The data I was originally using was a local shapefile, and it wasn't the best. Now we have a better map (using a raster from online), which is super cool because the river areas (more accurate than our current shapefile we use) are completely calculated by the function.

With regards to that issue you referenced, I believe they can use this code we made to get the 5mx5m resolution. I put the code here and it's called "topo_wet_index".

[teixeirak]

Awesome! Thanks to both you and @ValentineHerr!

Let's use the TWI in your analysis (in place of elevation and distance to stream).

Could you please put a note on the issue I referenced?

[mcgregorian1]

Ok I can do that.

I tried making viewable .tif files but it seems that they can't be viewable in standard image viewers and be geo-referenced simultaneously. Thus, I made both png and tif files, which are located here. The tif files are normal and can easily be read into ArcGIS.

[teixeirak]

Many thanks! Could you please make sure the readme gives a clear explanation?

[mcgregorian1]

Oops yep, I've just uploaded that, and I'll put the TWI in the model next week

[mcgregorian1]

Adding in TWI makes it come out as top, along with position and height.

Here are the coefficients:

Then, according to what we were talking about earlier with using the top aspects of the biophysical model in addition to the leaf traits, I get this. Interestingly, position does not come out in the full top model

Next step is to address #36

[mcgregorian1]

Krista mentioned that potentially trees that are suffering more are doing so not because they have a high TWI but because of their other traits. A quick plot showing TWI vs resistance values backs this up: