Simulation-based validation of sampling designs and prediction models for environmental monitoring

I just finally published my first two articles with Stan! Both model the latent ground truth of an environmental variable in the context of high measurement/sampling variability. Then they do what I’m calling simulation-based validation: draw posterior simulations of the ground truth from the model and use them to validate things.

In one article, we cross-validated previously published prediction models of bacteria levels in Lake Michigan which previous articles did ignoring the large measurement errors.

In another article we validated sampling designs for estimating soil organic carbon stocks in an agricultural field.

You can find pre-print versions of the articles on my website. If anyone has any questions/feedback on this work or is working on similar things and wants to chat/collaborate, please reply here or email me. Thanks so much to everyone (@rtrangucci @bgoodri and @James_Savage come to mind) who answered my questions both on this forum and off it over the years!