Evaluating the impact of urban morphology configurations on the accuracy of urban canopy model temperature simulations with MODIS

Abstract

Simulations of the urban environment contribute to assessments of current and future urban vulnerabilities to extreme heat events. The accuracy of simulations of the urban canopy can be degraded by inaccurate or oversimplified representations of the urban-built environment within models. Using a 10 year (2003–2012) series of offline 1 km simulations over Greater Houston with the High-Resolution Land Data Assimilation System (HRLDAS), this study explores the model accuracy gained by progressively increasing the complexity of the urban morphology representation in an urban canopy model. The fidelity of the simulations is primarily assessed by a spatiotemporally consistent comparison of a newly developed HRLDAS radiative temperature variable with remotely sensed estimates of land surface temperature from the Moderate Resolution Imaging Spectroradiometer. The most accurate urban simulations of radiative temperature are yielded from experiments that (1) explicitly specify the urban fraction in each pixel and (2) include irrigation. The former modification yields a gain in accuracy that is larger than for other changes, such as increasing the number of urban land use types. The latter modification (irrigation) substantially reduces simulated temperature biases and increases model precision compared to model configurations that lack irrigation, presumably because watering of lawns, parks, etc. is a common activity that should be represented in urban canopy models (although it is generally not). Ongoing and future efforts to improve urban canopy model simulations may achieve important gains through better representations of urban morphology, as well as processes that affect near-surface energy partitioning within cities, such as irrigation.