Stochastic Model for Hydrologic Drought Prediction under Climate Change Scenarios

To evaluate climate change impacts on hydrological drought, a downscaling method using the K-nearest neighbor search algorithm is used to identify time periods having similar properties in future GCM output with respect to historical data. This information is used to derive joint distribution of streamflows at two adjacent months. As the GCM output varies, the joint distribution evolves from time to time as well as the conditional distribution of streamflow at current month given that of previous month. The downscaled streamflow series can be sequentially sampled from the evolving conditional distribution. The projected streamflow sequence is further converted to standardized streamflow index (SSI) sequence, which is essential for assessing future drought status. Using model generated ensemble of streamflow sequences, a SSI series ensemble is obtained, which facilitates quantifying the uncertainty involved in the projected drought status. The proposed model will be applied in the Brazos River basin located in Texas for different GCMs and scenarios. From this research, we want to explore the prospective of reliability of GCMs for drought prediction by incorporating climate change scenarios.