Is anyone aware of sensitivity analyses that have compared soil carbon estimates based on land cover-based lookup tables (InVEST carbon) as opposed to using modeled soil carbon data, e.g., from SoilGrids 250 m and combining those soil carbon data with vegetation carbon modeled using InVEST?
I’m wondering whether there’s any evidence in favor of using lookup tables for vegetation and soil carbon combined (i.e., using InVEST for the whole carbon storage analysis) vs. using InVEST for vegetation carbon storage (aboveground + belowground + woody debris) and combining that sum with raster data from SoilGrids or local soil carbon storage datasets.
I suppose a big part of the answer would depend on how much we’d expect soil carbon storage to change as land use/cover does (i.e., since the soil carbon spatial datasets are essentially static). Thanks to anyone able to shed some light on this issue!
I also have the same question… I believe that the soil carbon content is clearly the most important pool when we assess the Carbon stocks in the landscape. Supporting the facility to obtain soil data which, in general, also provides the Soil Organic Carbon content, we could use this data to integrate the LULC maps and evaluate the whole system on the landscape, bein more important when we have a great contrast in the soil types with the same use/cover.
Do you InVEST developers or users had tried to join these two maps to an specific assessment? Is there any study or research which we can take as example or just information?
Please keep me posted on this discussion. I have observed that there is a lot of variability in the literature on above ground, below ground, and litter carbon biomass-- as well as on soil carbon stocks–and obviously the models are quite sensitive to that. Published values of carbon stocks in cropland vs prairie and wetlands in the Midwest, for example, show much higher values for grass and croplands than the carbon_pools sample csv file. Likewise, there is considerable variability based on forest types. Also, the sample carbon pools seem inaccurate for prairie wetlands, lakes, and woody wetlands. Why assume zero C stocks in lake /wetland sediments, for example?
I haven’t worked with carbon modeling much, but did want to make a note about the sample data carbon pools. They were created for a project in the Willamette Valley, Oregon, so I wouldn’t expect the values to necessarily correspond well anywhere else. Also, those data were made so long ago that we don’t have a record of where the values came from, they may have just been made up or altered for sample purposes (which is why we always advise people not to use the sample data for their own projects.) Last thing is that very often the hardest pool to find data for is soil (in most projects I’ve done at NatCap we end up using global soil maps that are very coarse and imprecise), so it’s common for people to not include it in their analysis.
On the topic of adding soil map data to the output of InVEST - sure, why not? The Carbon model is so simple that it makes a lot of sense to just use it for aboveground/belowground/litter, then add the resulting raster to a soil map that is not based on land cover. Just make sure the units are the same.