Possible discrepancies in Crop Production results

Hello, I am using InVEST in my PhD. work. I am not sure about my results from Crop Production model. Some results in “results table” are extremely different. Sometimes observed production is higher like a percentile production (e.g. oilseednes in my table). Percentiles would have show intensification level, observed production is production without intensification? Further for example alfalfa has very small observed production but percentiles are extremely high. Could you explain this results table? Is it correct? Thank you. result_table_2016.csv (33.4 KB)

Martin, Slovakia

1 Like

Hi @Martin, offhand, I’m not sure what’s going on here. Could you attach your logfile and maybe link to your data inputs so that we can take a look?

Thanks,
James

Hi @jdouglass, thank you for your response. This is my raster 2016_5_5.tif (585.6 KB) , landcover to crop table landcover_to_crop_table1.csv (277 Bytes)
and logfile (I dont know add third link)

Martin

Logfile InVEST-Crop-Production-Percentile-Model-log-2020-02-05–09_41_29.txt (85.4 KB)

Thanks for these, Martin! I don’t see anything clearly wrong with how the model is functioning, so I’ve reached out to the model’s science lead to see if he could chime in on this. More soon!

2 Likes

Hi Martin, I took a look as well and I don’t see anything that looks incorrect. But to answer your question, the percentile yield can differ substantially from observed yield due to how the percentile climate bins were calculated. Basically, if you have a specific high-yielding pixel in a climate bin that typically has lower yield and you choose to model the intensification level via choosing a percentile, it is quite possible that these high-yield pixels will be brought down (though probably this is offset overall by increases elsewhere). I’m not sure if I’ve answered your question, but I think it would also be worth noting that the InVEST crop model is best used for large-scale assessment and prioritization. For detailed, small-extent studies, we recommend using crop models more refined for that location.

1 Like

Hi Andrew, thank you very much for your response and explain. Probably my interest area is too small for model. I will try another evaluation method and I will compare with InVEST crop production.
Martin

@Martin if you think of it, we’d love for you to share the comparison once you’ve tried another evaluation method!

@dcdenu4 Yes of course. When I try another method and compare it, I will send the results here.