TelcoRain output data for TelcoSense platform

TelcoRain rainfall maps – monthly historic outputs This repository contains gridded rainfall products generated using the TelcoRain processing pipeline, which estimates surface rainfall from attenuation measurements of commercial microwave links (CMLs). The dataset is intended for spatial and temporal analysis of precipitation and for comparison with radar-based or gauge-based rainfall products. It contains output .png images (including overall rain strength for each whole frame) for CML-derived rain estimation from all over Czechia in 2025. The CLI application that computes the images is available here: https://github.com/TelcoSense/telcorain and the TelcoSense platform that visualizes the ouputs is available here: htttps://telcosense.cz/rain. In the .zip files, there are images from 2025-01-01 to 2025-12-31 with YYYY-MM-DD_RI.png format, where RI is median rain instensity (mm/h) for a given frame. Dataset contents The repository provides monthly historic outputs, processed independently for each calendar month. For every month, the following products are included: -- outputs_web/ -- Time-resolved rainfall maps representing instantaneous or short-interval rainfall intensity, suitable for visualization and web-based mapping. -- outputs_sum/ -- Hourly accumulated rainfall maps derived by temporal aggregation of the instantaneous fields. All raster outputs are stored as PNG images on a regular spatial grid and are cropped to the target study area. Spatial characteristics -- Data source: Commercial microwave links (CMLs) operated in the Czech Republic -- Spatial interpolation: Inverse Distance Weighting (IDW) -- Grid: Regular Cartesian grid in projected coordinates -- Cropping: Applied using a predefined geographic boundary (GeoJSON mask) The rainfall fields represent area-based rainfall estimates derived from link-center measurements, not point observations. Processing configuration Wet–dry classification Method: Rolling standard deviation–based detection CNN / MLP: Disabled Rolling window Duration: 1 hour Samples: 10 values Wet/dry threshold: 0.8 Baseline estimation: Last 5 dry samples Dry periods handling: Dry values are kept as zero rainfall (dry_as_nan = False) CML filtering Minimum link length: 0.5 km Link exclusion list: Disabled All valid links longer than the minimum threshold are used in the interpolation. Rainfall interpolation Interpolation method: Inverse Distance Weighting (IDW) Power parameter: 2 Maximum number of neighbors: 12 Maximum influence distance: 20 km These parameters control the spatial smoothness of the resulting rainfall fields and limit the influence of distant links. Temporal aggregation Hourly accumulation: Enabled Hourly rainfall sums are computed from the underlying instantaneous rainfall estimates. Rendering and masking Spatial cropping: Enabled All output rasters are masked to the target geographic region before export. Temporal coverage Mode: Historic processing Processing strategy:Each month is processed independently to ensure consistent aggregation windows and manageable data volume. Time resolution Instantaneous maps: native TelcoRain output resolution Accumulated maps: hourly totals Intended use This dataset is suitable for: -- Comparison with radar rainfall products (e.g. weather radar composites) -- Hydrological and climatological analyses -- Evaluation of CML-based rainfall estimation methods -- Visualization and web-mapping applications The data represent model-based estimates derived from telecommunication infrastructure and should be interpreted accordingly.