Open brownag opened 1 year ago
Thanks. I'll look into this and work on a fix before the next SSURGO refresh.
This will be fixed in the FY24 products.
The soil_depth grid is now modified by the depth to first restrictive feature.
You mean depth to any restriction? In any proportion of the mapunit? How are components with NULL restriction handled in the weighted averaging? The help dialog for the property does not specify
It appears that the area I highlighted in OP for North Dakota does not have restrictions in the major components.
https://casoilresource.lawr.ucdavis.edu/gmap/?loc=47.23230,-99.76105,z16
https://casoilresource.lawr.ucdavis.edu/soil_web/list_components.php?mukey=2564403
In this mapunit there is a minor with natric in the very shallow depth class. It seems that its not reasonable to have ISSR800 report very shallow if only 2% of a mapunit has a natric. I imagine that the lack of restrictions on the majors is causing the weighted average to be skewed to the small proportion of the MU/ISSR800 cell that has a restriction. I can sortof see an argument for portraying a "most limiting" condition, but in general these areas are not anywhere near that limited when 98% of the MU has no restriction.
soilDB::SDA_query(
"SELECT comppct_r, compname, resdept_r, reskind, muacres FROM mapunit
INNER JOIN component ON component.mukey = mapunit.mukey
LEFT JOIN corestrictions ON component.cokey = corestrictions.cokey
WHERE mapunit.mukey = 2564403"
)
#> single result set, returning a data.frame
#> comppct_r compname resdept_r reskind muacres
#> 1 2 Tonka NA <NA> 88968
#> 2 1 Lehr NA <NA> 88968
#> 3 45 Williams NA <NA> 88968
#> 4 5 Zahill NA <NA> 88968
#> 5 3 Zahl NA <NA> 88968
#> 6 36 Bowbells NA <NA> 88968
#> 7 5 Hamerly NA <NA> 88968
#> 8 2 Noonan 20 Natric 88968
#> 9 1 Parnell NA <NA> 88968
https://casoilresource.lawr.ucdavis.edu/soil-properties/?prop=resdept&lat=47.2624&lon=-99.4125&z=9
Good points, thanks for checking this. I had only addressed the problems with "soil depth" as modified by depth to shallowest restriction (any kind).
The current aggregation strategy doesn't correctly interpret the absence of restrictive feature as some "deeper" depth, thus estimates only reflect minor components with restrictions (not correct).
Given the many ways one could interpret restrictive features (e.g. which matter) and what to return, I'm going to remove the restrictive feature depth for now. I'll replace it with a new suite of maps (depth to root limiting, etc.) that use a different aggregation strategy.
I also noticed that now there appears to be a mismatch between the color/class and the actual raster value e.g.
Ah thanks. This is a database issue. I'll get that sorted out.
By sometime next week, there will be a better description of the "soil depth" grid / map, and the "depth to restrictive feature" map will go away. At some point a more carefully defined version will re-appear.
A few issues with soil depth and restrictions appear to be present. Andrew Conlin spotted these while we were discussing some topics related to regional concepts for restrictions and depth class population.
I don't see an obvious cause in the SSURGO data. In both cases there appears to be a mismatch in the soil depth vs. the restriction depth reported.
Examples:
Very shallow restrictive layer across much of north/east North Dakota (does not match very deep soil depth)
https://casoilresource.lawr.ucdavis.edu/soil-properties/?prop=resdept&lat=47.2624&lon=-99.4125&z=9
Very deep soil depth in metavolcanic foothills (does not match shallow/mod deep restrictive layer)
https://casoilresource.lawr.ucdavis.edu/soil-properties/?prop=soil_depth&lat=37.9940&lon=-120.7205&z=9
resdepb_r
of 200cm, they also have a lithic restriction underneath a paralithic restriction.