Copernicus DEM GLO-30 (AWS 2021 release): a measured digital surface model with 1-arc-second spacing, approximately 29 × 31 m in Telangana, and EGM2008 heights in metres. Trees and buildings can affect elevations. Adaptive meshes use provider overviews at wide scales and native samples in close-up; no finer detail is invented. Data are packed to 0.1 m increments for transfer, not a claim of 0.1 m accuracy. Pre-generated terrain tiles are served as static files with source metadata. A darker or sharper scene is not evidence of better measurement accuracy.

Produced using Copernicus WorldDEM-30 © DLR e.V. 2010-2014 and © Airbus Defence and Space GmbH 2014-2018 provided under COPERNICUS by the European Union and ESA; all rights reserved

Mapzen / Tilezen Terrain Tiles, sourced from SRTM and other public DEMs. Zoom 9 Terrarium tiles are bilinearly sampled to a 512 × 512 grid. Heights are metres; display exaggeration never changes routing.

geoBoundaries / DataMeet India state boundary, CC BY 2.5 IN; simplified cartographic outline. Geometry is not an official survey boundary.

Natural Earth river centerlines are generalized at 1:10 million scale (not 10-metre accuracy). Thin drainage lines are inferred from this DEM, not surveyed rivers. They can differ.

Natural routing uses steepest D8 descent and stops at a local depression. Fill & spill uses priority-flood depression filling and a deterministic receiver graph. Flat cells route toward their flood parent. Paths can climb raw terrain where a depression has been filled.

Added volume is placed at the first natural sink and fills connected DEM cells to a shared level, stopping at the first lower neighbour beyond a saddle, or the conditioned spill elevation, whichever comes first. Stored + excess = added volume. Excess is reported, not propagated through subsequent reservoirs. The conditioned route is a potential downstream path independent of volume. No timing, flood depth forecast, dam operation, soil absorption, or river discharge is modeled. Edge exits mean the study area ends, not that water stops.

NWIC National Water Data Portal / CWC, hourly telemetry water-level datasets for Godavari and Krishna (2026–2030). Latest valid Telangana station record is selected; NaN, negative values, and values above 1,500 m are excluded by an inland plausibility screen. Other sensor errors can remain; readings are not independently validated. Provider timestamps are interpreted as IST. Resource IDs and fetch code are included with this project.

Readings come from a published JSON file. Reload checks that file for a newer published snapshot; it does not query CWC directly. Optional checks run every 15 minutes while the page is visible. The publisher updates the JSON using the data preparation script and republishes the static site. Readings older than 24 hours are marked stale; future timestamps are flagged. Failed reloads retain the last valid snapshot. A newly published file can contain old observations.