A linear dynamical systems approach to streamflow reconstruction reveals history of regime shifts in northern Thailand

Abstract

Key points• The state variable reveals regime-like behavior in the catchment history• The linear dynamic model has higher accuracy than conventional linear regression • The model can generate stochastic replicates of both streamflow and catchment state time series ABSTRACT Catchment dynamics is not often modeled in streamflow reconstruction studies; yet, the streamflow generation process depends on both catchment state and climatic inputs. To explicitly account for this interaction, we contribute a linear dynamic model, in which streamflow is a function of both catchment state (i.e., wet/dry) and paleo-climatic proxies. The model is learned using a novel variant of the Expectation-Maximization algorithm, and it is used with a paleo drought record-the Monsoon Asia Drought Atlas-to reconstruct 406 years of streamflow for the Ping River (northern Thailand). Results for the instrumental period show that the dynamic model has higher accuracy than conventional linear regression; all performance scores increase by 45-497%. Furthermore, the reconstructed trajectory of the state variable provides valuable insights about the catchment history-e.g., regime-like behavior-thereby complementing the information contained in the reconstructed streamflow time series. The proposed technique can replace linear regression, since it only requires information on streamflow and climatic proxies (e.g., tree-rings, drought indices); furthermore, it is capable of readily generating stochastic streamflow replicates. With a marginal increase in computational requirements, the dynamic model brings more desirable features and value to streamflow reconstructions.

Increased Variability of Thailand's Chao Phraya River Peak Season Flow and Its Association With ENSO Variability: Evidence From Tree Ring δ18O

“…D'Arrigo et al () reconstructed Indonesia's Citarum River based on ring width measurements from regional teak sites that exhibited a strong direct correlation between rainfall and September–November streamflow. More recently, Nguyen and Galelli () used the tree ring width‐derived Monsoon Asia Drought Atlas data (Cook et al, ) to reconstruct the annual flow of north Thailand's Ping River. In comparison, our simple linear regression using the CPRB18O as a predictor resulted in a reconstruction model that accounted for 57% of the variation in CPRPF for the period 1956–1999.…”

“…One promising way to extend the instrumental record of streamflow is to utilize regional tree ring records as proxies for streamflow through the connection to hydroclimate (e.g., Stockton & Jacoby, ). Due to their annual resolution, tree rings are easily comparable with observed streamflow data and have been successfully applied in many regions of the world for reconstructing streamflow (e.g., D'Arrigo et al, ; Derose et al, ; Gou et al, ; Lara et al, ; Meko et al, ; Nguyen & Galelli, ; Woodhouse et al, ). Despite the challenges associated with cross‐dating tropical tree rings, a growing body of annually dated tree ring records have emerged from Southeast Asia that make it possible to attempt such a reconstruction for the CPR.…”

Multi‐Proxy, Multi‐Season Streamflow Reconstruction With Mass Balance Adjustment

“…Also, there may be nonlinearities in the streamflow–proxy relationships at the extremes. In future studies, nonlinear reconstruction models (e.g., Nguyen & Galelli, 2018; Torbenson & Stagge, 2021) could be incorporated to address this problem.…”

Streamflow Reconstruction in the Upper Missouri River Basin Using a Novel Bayesian Network Model

“…Nguyen and Galelli () used a linear dynamic systems modeling approach to streamflow reconstruction in northern Thailand. Ho et al () demonstrated another modeling approach that was applied to reconstructing streamflow in the Missouri River Basin using a gridded paleo‐proxy called the Living Blended Drought Atlas (LBDA).…”