Hi @asuleo -
Unfortunately, the NDR model does not really include a measure of retention on each pixel. We usually evaluate retention based on comparing the current conditions with a scenario of land use/land cover and/or management change. The difference in export between current and scenario provides an estimate of the amount of nutrient retained.
Thanks for posting your log file, but I’m very confused by it. Some of the entries seem to be related to NDR (like calc_n and calc_p), but others are related to the Annual Water Yield model (like pawc and seasonality constant, along with the final KeyError message). Please try attaching the actual .txt log file that is created in your Workspace, which should only have information for one model run.
~ Stacie