Estimation the evapotranspiration of urban parks with field based and remotely sensed datasets

Abstract

Climate data and remote sensing images are used in this study to estimate the evapotranspiration of urban landscape vegetation. The study was conducted on a historic public park in Vrgac (Serbia), which is an important Serbian national heritage site. After comparing recordings from 14 weather stations in the region with recordings from the City Park, the weather station in Vrgac was chosen. The daily averaged values of climatic data for March, June, July and October of each year between 1949 and 2016, were used to compute the daily reference evapotranspiration ET] (mm/day), according to the FAO-56 Penman-Monteith equation. A landscape coefficient was estimated through field monitoring based on the Water Use Classification of Landscape Species (WUCOLS) principles. Also, thermal infrared images from Landsat 8 and the Normalized Difference Vegetation Index (NDVI) from the QuickBird Satellite data are used as the remote sensing inputs to model daily evapotranspiration. This research explored the relationship between urban vegetation ET1 computed by the FAO-56 Penman Monteith method with the required meteorological data and by means of the remote sensing. The analysis revealed the significant correlation between the average daily evapotranspiration estimates from 0.98 to 0.99 by the Nadaraya-Watson kernel regression and the weak Spearman's Rank correlation (p ranges from -0.006 to 0.384) leading to the conclusion that the Nadaraya-Watson kernel regression is more suitable for evaluating urban landscape water requirements.