High-Resolution Global Water Temperature Modeling

Abstract

The temperature of river water plays a crucial role in many physical, chemical, and aquatic ecological processes. Despite the importance of having detailed information on this environmental variable at locally relevant scales (≤50 km), high-resolution simulations of water temperature on a large scale are currently lacking. We have developed the dynamical 1-D water energy routing model (DynWat), that solves both the energy and water balance, to simulate river temperatures for the period 1960–2014 at a nominal 10-km and 50-km resolution. The DynWat model accounts for surface water abstraction, reservoirs, riverine flooding, and formation of ice, enabling a realistic representation of the water temperature. We present a novel 10-km water temperature data set at the global scale for all major rivers, lakes, and reservoirs. Validated results against 358 stations worldwide indicate a decrease in the simulated root-mean-square error (0.2 °C) and bias (0.7 °C), going from 50- to 10-km simulations. We find an average global increase in water temperature of 0.16 °C per decade between 1960 and 2014, with more rapid warming toward 2014. Results show increasing trends for the annual daily maxima in the Northern Hemisphere (0.62 °C per decade) and the annual daily minima in the Southern Hemisphere (0.45 °C per decade) for 1960–2014. The high-resolution modeling framework not only improves the model performance, it also positively impacts the relevance of the simulations for regional-scale studies and impact assessments in a region without observations. The resulting global water temperature data set could help to improve the accuracy of decision-support systems that depend on water temperature estimates.