Six centuries of Upper Indus Basin streamflow variability and its climatic drivers

Area of Science:

• Hydrology
• Paleoclimatology
• Statistical Modeling

Background:

• Understanding natural streamflow variability in the Upper Indus Basin (UIB) is limited by short instrumental records.
• Reconstructing past streamflow is crucial for assessing long-term water availability and risk.

Purpose of the Study:

• To reconstruct six centuries of streamflow in the UIB using Hierarchical Bayesian Regression (HBR).
• To assess the reliability of HBR for reconstructing streamflow from short and incomplete instrumental records.
• To compare HBR with Principal Components Regression (PCR) for streamflow reconstruction.

Main Methods:

• Employed Hierarchical Bayesian Regression (HBR) with partial pooling for streamflow reconstruction.
• Developed six-century-long (1394-2008 C.E.) streamflow reconstructions at three UIB gauges.
• Validated reconstruction methods using a longer instrumental record at Partab Bridge, comparing HBR, Bayesian Regression (BR), and Principal Components Regression (PCR).

Main Results:

• HBR successfully reconstructed streamflow from short records with missing data.
• PCR and BR yielded nearly identical reconstructions at Partab Bridge, indicating method independence.
• Reconstructions revealed that streamflow in the 1990s exceeded the mean flow of the past six centuries.
• Streamflow sensitivity was identified for winter precipitation and summer temperatures, with warm summers enhancing melt.

Conclusions:

• Projected warming may increase future streamflow due to enhanced snow and glacier melt.
• Long-term water security in the UIB will depend on future snowfall and glacial mass balance changes.
• HBR is a viable method for reconstructing paleohydrological data from limited instrumental records.