Using multispectral and thermal UAV data to infer the influence of contrasting urban green space on evapotranspiration and heat fluxes

As global urbanization has become more dominant in recent years so have the negative consequences of dense, artificial, urban environments on their inhabitants. The urban heat island is one such phenomena, describing an increase of air temperature in the inner city in comparison to the periphery, with negative effects for human health. Urban green spaces are crucial to mitigating this heat stress due to their higher levels of evapotranspiration (ET) and shading. To maximize cooling potential, the individual contribution of typically heterogenous urban green space to ET and heat fluxes needs to be better understood. Higher ET rates of urban vegetation need to be balanced against on-site water availability to plan the most efficient green spaces. Moreover, trees and shrubs possess the additional benefit of shading the surface below.
Traditional remote sensing methods have focused on the use of satellite data with multi-meter spatial resolution as a cost-effective way to observe and analyze the large spatial extent of cities. However, the individual vegetation compositions of urban green spaces cannot be resolved through these systems, making it hard to evaluate their superimposed spectral signal. Unmanned Aerial Vehicles (UAVs) record data with very high spatial resolution and also allow multiple flights per day to cover the temporal and spatial urban green space heterogeneity.
Estimation of vegetation indices, land surface temperature (LST) or ET can reveal the high spatial heterogeneity of urban vegetation patches and help to better understand the spatial and temporal patterns of ET and urban cooling at a plot scale.
In our study, we assessed multiple remote sensing-based ET modelling techniques for thermal and multispectral UAV remote sensing data and validated them against in-situ measurements. Data has been recorded at a monthly interval from April to October at an urban research garden in Berlin, Germany consisting of representative urban vegetation types. An inference method was tested with different vegetation indices to estimate ET for three different green space classes (trees, shrubs, grass). In situ measurements for sap flow, soil moisture, leaf area index and meteorological conditions were used to compare and validate the UAV-based ET, LST, and greenness estimates.
Results showed a significant difference for ET and surface cooling between green space classes throughout the year, with trees and shrubs showing consistently lower temperatures then grassland. The influence of shadow on cooling potential (ET and LST) also became apparent.
The findings of our study provide further insights into the influence of different urban green spaces on ET and cooling potential and are valuable for upscaling approaches to the city scale. This knowledge can further support valuable decision-making for planning and managing urban green spaces to mitigate heat risks and optimize urban water supply.