Very cool project. I’m curious: after manually tuning parameters to match observations, where do you draw the line between a first-principles simulation and a calibrated model? It sounds like every climate model eventually needs some amount of parameter fitting.