Skip to main content
U.S. flag

An official website of the United States government

Decadal Prediction and Stochastic Simulation of Hydroclimate over Monsoonal Asia

Funding Program Area(s)
Project Type
University Grant
Project Term
to
Project Team

Principal Investigator

Collaborative Institutional Lead

Our goals are (1) to develop and test statistical methods of decadal prediction of river flows over monsoonal Asia; (2) to further develop methods of stochastic Monte Carlo simulation in order to merge these decadal predictions with GCM climate change projections; (3) to quantify the full probability density function (PDF) of the uncertainties in these merged predictions; (4) these simulations will be tested primarily using hydrologic models for two major reservoir systems over Asia, namely that of the Bhakra Beas Management Board (BBMB) that runs one of the largest reservoir operations in India (in Punjab), and that for the Yangtze River Three Gorges Dam reservoir. (5) Decadal variability and predictability will be approached by further developing tree-ring based reconstructions of stream flow for Asian river basins in order to obtain multi-centennial time series. These long series will enable much better identification of natural modes of low-frequency climate variability across monsoonal Asia, and allow the performance of candidate prediction schemes to be tested retrospectively.

The proposed prediction methodologies will harness and further refine data-adaptive spectral methods and empirical model reductions developed by M. Ghil's group at UCLA, together with retrospective testing and forecast tailoring methods evolved at IRI. The stochastic simulation and downscaling models will advance and combine hidden Markov models and stochastic hydrologic simulation models developed at UCI and CWC, respectively, creating multi-level models for daily joint distributions of spatially correlated rainfall on temperature as well as streamflow, conditioned on the annually (or seasonally) resolved low-frequency modes. The latter will provide a physical basis to merge information from proxy data records and GCM simulations, predictions, and projections, addressing potential issues of nonstationarity. A stochastic, multi-purpose reservoir optimization model in the yield model framework will be developed and applied at each of the sites, based on the resulting stochastic simulations and predictions. Trade-off analyses across uses and across time will be explicit in these scenarios and finite-period optimization will be used to create new quantitative adaptive decision making.

The proposed work will deliver urgently needed methodologies to enable more effective planning and adaptation to near term climate change and variability to 2050. While these methods will be tested using the primary reservoir management case studies in India and China, as well in Indonesia, they will be broadly and geographically applicable. All results and software will be made easily accessible to the public via wiki pages and other means. The project will train two graduate students (CWC and UCI), and build technical and scientific capacity at the collaborating institutions in India, China, and Indonesia.