Computational Hydrology · Stochastic Modelling · GeoAI for Water
Shalini Balaram
I build stochastic and machine-learning models of how water moves, fails, and recovers — from a single river gauge to the monsoon over a continent.
Research
Three scales of water
My work runs along one thread — uncertainty in water — at three scales. Simulate: stochastic models that generate plausible streamflow and rainfall when the record is too short to trust. Recover: how droughts and reservoir storage fall fast and return slowly. Predict: GeoAI that learns where drought propagates across a continent of catchments.
Selected work
- Published Environmental Research Communications, 7(2), 021011. A novel multi-step methodology for stochastic simulation of streamflow time series using PcStream clustering.
- Under Review Environmental Research Letters Reservoir Storage Anomalies Often Recover More Slowly Than They Develop Across Global Records.
- Under Review Climate Dynamics (Springer) Hysteresis in Hydroclimatic Drought Under CO₂ Removal: Atmosphere–Land Recovery Decoupling in the Indian Monsoon Region.
Field notes
Writing on method
Essays on when machine learning is the wrong tool, how to choose a distribution, where to find free hydrology data, and automating the work that should be automated — grounded in real papers and real code.