Datasets associated with “RTSEvo v1.0: A Retrogressive Thaw Slump Evolution Model” submited to Geoscientific Model Development

This dataset provides the input data and model simulation results for the three experiments presented in our manuscript, "RTSEvo v1.0: A Retrogressive Thaw Slump Evolution Model," submitted to Geoscientific Model Development. The data supports the calibration and validation of a novel dynamic model for simulating the spatiotemporal evolution of retrogressive thaw slumps (RTS).The data is organized into three folders corresponding to the experiments described in the paper:Experiment 1 – Areal demand forecasting<b>Purpose:</b> To test the Holt's linear trend model for forecasting total regional RTS area.<b>Contents:</b>Observed total annual RTS area for the Beiluhe Basin (2016–2020).Forecasted total RTS area for 2021 and 2022.Experiment 2 – Base occurrence probability mapping<b>Purpose:</b> To evaluate the machine learning models (Random Forest and Logistic Regression) used to map the baseline probability of RTS occurrence.<b>Contents:</b>Training and testing datasets from 2016–2020.Input feature layers (e.g., topography, climate, geology, permafrost).Predicted RTS occurrence probability maps for 2020, 2021, and 2022.Experiment 3 – Full evolution model simulation<b>Purpose:</b> To test the performance of the full, calibrated RTSEvo model in simulating RTS distributions for subsequent years.<b>Contents:</b>Model calibration data using the 2020 RTS map as a reference.Final simulated RTS distribution maps for 2021 and 2022 from both the LR-EM and RF-EM versions of the model.Corresponding model evaluation metrics (FoM, Kappa, F1 Score, Moran’s I).All spatial datasets are provided in GeoTIFF (.tif) format at a 10-meter spatial resolution. The coordinate reference system for all layers is WGS 84 / UTM Zone 45N.<br>