Abstract
Trees often grow annual rings. These rings vary year to year depending on the annual climatic conditions. Thus, tree rings are proxies through which we can infer about past climate, such as changes in the water cycle. Specifically, we can use tree rings to reconstruct history of river discharge, or streamflow, hundreds of years back in time. These long time series augment the short instrumental records and can help us better manage our water resources. While this idea was conceived decades ago, challenges in reconstruction methodologies have limited the use of streamflow reconstruction in water management. Most reconstructions rely on linear models that do not account for catchment dynamics. They are also annual, while water management decisions are made at finer time steps. My research tackles these two challenges. First, I develop a new reconstruction method based on linear dynamical systems (LDS) to account for catchment dynamics. I use the LDS method to reconstruct eight centuries of streamflow history at 62 river reaches in 16 countries across the Asian monsoon region. The reconstructions reveal that streamflow in Monsoon Asia is spatially coherent, owing to common influences by the oceans. Second, I produce seasonal and monthly reconstructions, made possible by optimally combining proxies that have different seasonal sensitivities. The monthly reconstructions of the Ping and Nan Rivers are then used to stress-test the two largest reservoirs in Thailand, proving the importance of long streamflow records containing a broad variety of droughts and pluvials. In synthesis, this work advances streamflow reconstruction methodology, provides insights about long term hydroclimatic variability of Monsoon Asia, and demonstrates the practical value of streamflow reconstruction.