This repository contains - Meteorological and Snow Cover Data and Scripts for Machine Learning–Based Gap-Filling of Satellite Snow Products Using Time-Lapse Photography and Meteorological Data

In mountainous regions, satellite-based snow products often serve as the primary data source, but their use is limited by atmospheric, topographic, and sensor-related constraints that result in frequent data gaps. Here, we integrated time-lapse photography, satellite observations, and ground-based meteorological measurements to reconstruct snow cover seasonal profiles in satellite products. A time-lapse camera trap installed at the Skalnaté Pleso Observatory in the High Tatra Mountains (Slovakia) provided daily observations over four winter seasons (2021/22–2024/25). Camera trap snow cover (CT_SC), derived using an automated image-processing workflow, was used both for validation and as a training dataset in a hybrid physical–machine learning framework based on XGBoost for gap-filling Sentinel and MODIS snow products. The model combined meteorological forcing (air temperature, precipitation, global solar radiation) with physically based constraints, including the relationship between snow depth and CT_SC, to represent the dynamics of snow accumulation and melt.