A modified PML model considering anthropogenic heat to estimate urban evapotranspiration

Urban evapotranspiration (ET) is the key variable in urban water cycling, and its accurate estimation is important for the efficient urban water resource management. One of the limitations for achieving long-term high accuracy urban ET modeling lies in the inadequate incorporation of anthropogenic factors. In this study, we modify the Penman-Monteith-Leuning (PML) model by incorporating physical properties of urban areas and human activities, named as PML-urban model, to improve the ET estimation accuracy especially in urban regions. The major modifications include the incorporation of underlying surface properties (the ratio of precipitation to potential evapotranspiration (f) and leaf conductance (g_sx)) into the ET's calculation, and the anthropogenic heat flux (AHF) retrieved from night light remote sensing data in the energy balance. The case study in two eddy covariance (EC) towers demonstrates that the PML-urban model outperforms PML model in urban ET estimation compared with the ET observations, with RMSE, Rbias and NSE changing from 1.026, 41.83 %, 0.546 to 0.654, 2.80 %, 0.631 in Tianjin and from 0.793, −22.23 %, 0.546 to 0.654, −16.65 % and 0.504 in Miyun. The consideration of AHF will induce an annual average ET increment of 25.8 mm in the central urban area of Tianjin city, while the expansion of urban impervious surfaces will lead to 230.0 mm of annual average ET reduction, and finally cause the lower annual average ET estimation by PML-urban model (406.2 mm) than the PML model (455.7 mm) during 2001–2018. The modified model in our study provides an efficient tool to give a relatively more accurate ET estimation.