RECLAIM Model Development Resources

RECLAIM Data This folder contains the datasets used for RECLAIM (Reservoir Estimation of Capacity Loss using AI-based Methods), the globally scalable machine learning framework to predict reservoir sedimentation rates. Folder Structure 1. global_datasets Contains global-scale input datasets used for feature generation: glc_shared_combined.nc — Land cover data (GLC-Share) hwsd2_soil_d1.nc — Soil data (HWSD2) terrain.nc — Terrain/DEM derivatives veg_gain_loss_1960_2019.nc — Vegetation gain and loss from 1960–2019 These datasets provide static environmental and catchment information used to compute features for reservoirs worldwide.Tip: These global datasets can be used to generate reservoir and catchment features using the pyreclaim Python package. 2. model_development_data Contains processed data for model development: 88 predictor features for each reservoir Additional columns from GRILSS or derived features used to compute other features Prepared for training, validation, and testing with an 80/10/10 split This dataset is used to train and validate the RECLAIM machine learning models. Authors & Contact Sanchit Minocha, University of Washington — LinkedIn Faisal Hossain, University of Washington — Website For questions regarding the datasets or RECLAIM workflows, feel free to reach out to the authors. References GitHub Repository: https://github.com/UW-SASWE/RECLAIM Documentation: https://reclaimio.readthedocs.io/en/latest Note: This data supports the feature generation and model training workflows of RECLAIM and is not intended for standalone analysis.