Spatio-temporal changes in water demand of urban greenery

Accurate estimation of evapotranspiration (ET) and water demand of urban green spaces (UGS) remain critical, especially in water-limited cities. Measuring ET helps decision‐makers, urban planners and urban water managers formulate strategies and plans for sustainable green cities worldwide. In this study, we used three satellites, WorldView2, Landsat (OLI, TM5 and ETM+), and MODIS to measure the greenness and ET of a 780‐ha public green space, the Adelaide Parklands in Australia. Different satellite‐based vegetation indices (VIs) including the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI) and Enhanced Vegetation Index 2 (EVI2) were assessed. The VI-based ET from these three satellites were estimated. We then validated these remote sensing-based ET with a field-based method of Soil Water Balance (SWB) using Artificial Neural Network (ANN). Inter‐ and intra‐annual changes of VIs and their relevant ET were mapped and analyzed during 2010-2018. Our study, using multi-sensor remote sensing data fusion, systematic methods and machine learning techniques confirmed the suitability and feasibility of remote sensing-based ET as accurate long‐term monitoring mean for ET trends over large UGS. Our techniques rely on public and free-access satellite images, and therefore, can be adapted to other water-limited cities.