How much electricity will Nepal need from India this winter?

From January to April 2027 Nepal will need about GWh of electricity from India. The models alone expect GWh; the rest comes from the August 2026 flood, which took MW off the grid.

River flow, three big basinsNormal flow (2010–2022)Import from India, with flood lossesImport, models only
Every dry season the rivers fall and imports rise. Dashed river line: GloFAS forecast, band = ensemble quartiles. Import band: P10–P90.
Import, Jan–Apr 2027
GWh

+% on the models-only GWh

River flow, Jan–Apr 2027
% vs normal

GloFAS forecast, three big basins

Capacity lost in the Aug 2026 flood
MW

MW destroyed, the rest cut off

Demand, Nov 2026–Apr 2027
GWh

within 1% of the GDP-growth path

Demand keeps growing, winter included

Thin line: NEA monthly reports. Blue: forecast with P10–P90 band.

The story in four points

  1. Demand is not the problem: supply is. Monsoon demand is higher than winter demand, but in the monsoon Nepal exports. In winter it imports, because the rivers are low (Analysis).
  2. Rivers drive imports. Domestic generation follows river flow (correlation 0.82), and imports move the other way (−0.77). GloFAS forecasts this winter's rivers 13–30% below normal from January (Forecast).
  3. The August 2026 flood changed the picture. About 510 MW of hydro and solar was destroyed or cut off, and Kulekhani, the only storage reservoir, is running down. No model can learn this from history, so it is added explicitly from a sourced register (Map, Forecast).
  4. Simple models win here, and we show why. Ridge regression beats every baseline for daily demand (2.5% error next day, 5–8% at 30 days). LightGBM was tested and lost, which is expected on two winters of data (Models, Method).