Comparison of Thermal Infrared-Derived Maps of Irrigated and Non-Irrigated Vegetation in Urban and Non-Urban Areas of Southern California

Understanding the distribution of irrigated and non-irrigated vegetation in rapidly expanding urban areas experiencing climate-induced changes in water availability (e.g., Los Angeles) is necessary for developing sustainable water use practices and accurately accounting for carbon exchange and water balance changes. However, pre-existing maps of irrigated vegetation are largely limited to non-urban agricultural regions and are too coarse to resolve heterogeneous and diverse urban landscapes. Previous research suggests that irrigation has a strong cooling effect on vegetation, affecting productivity and uptake of carbon, especially in semi-arid environments. The recent launch (July 2018) of the moderate-resolution (70-100 m, 3-5 day revisit) ECOsystem Spaceborne Thermal Radiometer on Space Station (ECOSTRESS) offers an opportunity to test this hypothesis using retrieved land surface temperature (LST) data in complex, heterogeneous urban environments. In this study, we demonstrate the feasibility of classifying irrigated versus non-irrigated vegetation in mixed pixel-resolution LST imagery using a vegetation-fraction weighted LST across urban and non-urban areas of southern California. We leverage 30 m sharpened afternoon summertime ECOSTRESS LST, apply very high-resolution (0.6-10 m) vegetation fraction weighting, and produce a map of irrigated versus non-irrigated vegetation in Los Angeles with 98% accuracy. We compare this classification to a range of other classifications using different combinations of sensors and offer a preliminary accuracy and uncertainty assessment. This approach verifies that ECOSTRESS LST data provides an accurate map of irrigated urban vegetation in the southern California Air Basin (SoCAB) useful for studying land use fragmentation and has the potential to reduce uncertainties in regional carbon and hydrological cycle models.