A Methodological Framework for Harmonized Comparison of Model-Based and Satellite-Derived Snow Cover Products in Data-Sparse Mountain Regions of Kyrgyzstan.

In the glacier-melt-dominated regions of Kyrgyzstan, accurate cryosphere monitoring is essential for Central Asian water resource forecasting. However, the lack of consistent in situ observations necessitates the integration of diverse remote sensing datasets and physically based models which often vary in their underlying assumptions and resolutions. This study presents a transparent, reproducible framework for the comparative evaluation of heterogeneous snow products in complex terrain, applied to SnowMapper (a NWP-driven physical model) and GlacierMapper (a MODIS-based NDSI product) for the period 2000–2024.

The framework employs spatial harmonization via nearest-neighbor resampling to a common independent grid and temporal alignment across differing calendar conventions. To address variable incompatibility, Snow Water Equivalent (SWE) outputs from SnowMapper are transformed into binary snow/no-snow classifications using literature-derived thresholds. Sensitivity analyses reveal that product agreement is significantly influenced by these methodological transformations. Evaluation using complementary spatiotemporal diagnostics such as fractional snow cover area, balanced accuracy, Cohen’s kappa and snow depletion curves (SDCs) identifies periods of systematic divergence across decadal and seasonal timescales. Results demonstrate that apparent product discrepancies arise not only from physical inconsistencies but also from methodological treatment. This standardized intercomparison approach is transferable across sensors and regions enhancing the reliability of snow-product assessments in data-scarce mountain environments.