Reconstructed Daily Air Temperature for 2023 for Tunka Mountain Depression (Buryatia, Russia)

This dataset accompanies the study “Machine Learning-Based Reconstruction of Air Temperature in Mountainous Regions of Eastern Siberia Using Remote Sensing and In-Situ Data” (2026) and contains the primary data used for model development, validation, and the resulting high-resolution temperature products for the Tunkinskaya depression, Buryatia, Russia. The dataset includes: 1. In-situ observations from 34 microclimate monitoring stations operated by the V.B. Sochava Institute of Geography SB RAS across the Tunkinskaya depression for the year 2023. For each station, the file provides geographic coordinates (latitude, longitude), and daily mean air temperature calculated as the arithmetic mean of hourly measurements. Data format: MS Excell 2. The dataset contains gridded temperature products for the entire study area at 30 m spatial resolution, derived from the machine learning model (Bagged Trees ensemble) trained on the in-situ observations and remote sensing predictors. These include daily mean air temperature maps for each day of 2023. Data format: Matlab 3. Monthly mean temperature maps aggregated from daily values (12 GeoTIFF files), and an annual mean temperature map (1 GeoTIFF file). All grids are aligned to the ALOS World 3D – 30m (AW3D30) digital elevation model and cover the Tunkinskaya depression within the southwestern Baikal region. The dataset enables reproduction of the temperature reconstruction, validation of the machine learning approach, and supports further applications in frost risk assessment, hydrological modeling, and climate monitoring in data-sparse mountain regions. Users are kindly requested to cite the associated research article when using these data.