GIS

River Flow Prediction from Historical Data on Google Earth Engine

A Google Earth Engine model that predicts river flow from historical climate and catchment data, for a basin with thin gauge coverage.

Client: Water Resources StakeholderWater Resources & HydrologyDec 2019 – Jan 2020Project lead: M. A Zia

The challenge

The client needed to anticipate river flow for planning, but the basin had sparse and inconsistent gauge records. Decisions about water availability were being made on short, patchy history, and there was no practical way to install and wait on new instrumentation.

Our solution

We rebuilt the picture from satellite and reanalysis archives on Google Earth Engine. The catchment was delineated from elevation data, then decades of precipitation, temperature and snow-cover records were extracted over it and aligned with the gauge discharge that did exist. A regression model trained on that relationship produces flow predictions, and because the whole thing runs on Earth Engine, it re-runs against fresh data without any local processing burden.

Results & impact

  • Flow prediction made possible despite sparse gauge records
  • Decades of climate and catchment history brought into the model
  • Runs entirely on Earth Engine, no local processing infrastructure needed
  • Re-runnable as new seasons of data become available

Technology

Google Earth EngineJavaScript APICHIRPSMODISPythonRegression Modelling

Skills applied

Hydrological ModellingCloud Geospatial AnalysisTime-Series AnalysisWatershed DelineationStatistical ModellingGeography

Let's build something that delivers

Tell us about your project. We'll respond with a clear plan, timeline, and quote, no jargon, no pressure.