Code for seasonal crop yield forecasting in Ghana, Kenya, and Zimbabwe using LPJmL forced with SEAS5

Software used to produce the results in "Process-based seasonal crop yield forecasting for early warning in Sub-Saharan Africa". This archive contains the full processing chain for a seasonal crop yield forecasting system in which bias-corrected and downscaled SEAS5 seasonal climate forecasts drive the LPJmL crop model. It covers the download and preparation of climate and land-surface inputs, the bias correction and downscaling of SEAS5, the configuration and execution of LPJmL, and the evaluation of forecast skill against an ERA5-Land-driven reference simulation. Hindcasts span 1994 to 2023 for all major crops in Ghana, Kenya, and Zimbabwe. The archive bundles six repositories, each snapshotted at the commit listed below: The LPJmL model version used: lpjml-wur-ssa (e4362fdf)pycoupler-wur (c0be21a): Pycoupler scripts used to configure and run the modellpjml-prepare (6697913): preparation of input data for LPJmLclimate-data (3deac24): Scripts to download and process ERA5-Land and SEAS5 from the Copernicus Climate Data Storebcca-seasonal-forecast (341feba): Bias correction and downscaling of SEAS5 climate variables using the BCCA methodlpjml-analysis (73603a0): Validation of the LPJmL historical reference run and assessment of seasonal hindcast skill Each repository contains a README with instructions on dependencies, input data, and expected outputs. Development versions are hosted at git.wur.nl. The LPJmL model is developed at the Potsdam Institute for Climate Impact Research. The associated datasets are available at Seasonal crop yield hindcas… (seasonal crop yield hindcasts), Historical crop yield simul… (historical reference simulation), and Seasonal crop yield forecas… (forecast verification metrics).