I have been analyzing our SLAMM out put data two ways: 1) using the SLAMM output data tables (and converting ha to ac), and 2) using GIS (calculating ac and ha by converting pixel counts in attribute tables based on pixel size). The GIS analysis is necessary for post-processing some information, like which lands are already conserved, or tracking specific changes in SLAMM classes (e.g. how much area converted from RFM to TF).
I've noticed a significant difference in acreages between the two methods. For example, at time 0 there is a difference of 5 acres in my study area (of about 7,000 acres). But looking at total tidal wetland acreage by 2100, one simulation has a difference of 85 acres. I find these differences even when I'm not aggregating classes (i.e. when comparing the acreages of a single class, like RFM).
Have you encountered this before? Do you have any thoughts on what might be causing these differences?
I would like to continue to use both data sets, as the one from SLAMM reports on all the time steps (which would take a long time to do in GIS).
As long as the sum of all land cover types are equal to the study area in both cases (otherwise there is a problem ...), I think that this discrepancy is because the numerical calculation reported in the tables accounts also of partial conversion of a cell, e.g. 30% IFM and 70% RFM, while the raster only show the majority of the land cover in a cell, e.g. 100% RFM.
Considering all uncertainties, I would not worry too much about these differences in the sense that your conclusions should be pretty much the same whether you use one number or the other. Obviously an uncertainty analysis would be more robust to achieve confidence on the predictions of land parcels that are at the edge of different land cover types ....
In any case, when possible, I would use the numbers of the output data tables.
Hope it helps.
Thanks Marco, its very helpful.
I can get the total numbers pretty close if I use more decimals on my calculation converting pixel count to acres. And I think the bigger differences I was finding have to to with an offset of GIS layers used in a raster calculation- I'll fix that and see if it gets better. But either way its good for me to understand how SLAMM calculates area vs GIS, so thanks!