Wed, Aug 17, 2022: 3:45 PM-4:00 PM
518C
Background/Question/MethodsSatellite-derived land surface temperature (LST) has been widely used for investigating thermal conditions with important ecological implications. However, satellite-derived LSTs had a trade-off between spatial and temporal resolutions and missing observations caused by clouds, while weather modeling was limited because of potential bias and required computationally expensive model calibration. Therefore, it is highly desirable to build an effective method for improving LSTs at high temporal and spatial resolutions. Taking the urban and surrounding areas of Polk County, Iowa, USA, as a test site, this study generated hourly LSTs in 1 km spatial resolution from August 1st to August 20th, 2012, using the weather research and forecasting model (WRF) and the morphing technique (hereafter named as WRFM). First, the WRF simulations were performed with the updated satellite-observed land surface properties. Second, the hourly LSTs at 1 km spatial resolution were estimated using WRFM driven by the gap-filled MODIS LST data. Third, urban hotspots, which experience excessive energy use and lower comfortability, were explored at an hourly scale using WRFM-generated LSTs
Results/ConclusionsBy comparing with satellite observed LSTs, we found that the WRF simulation could well capture diurnal variations of LSTs with correlation coefficient > 0.8 but tended to have a bias on the magnitude and spatial variations with a wide range of RMSE (root mean square error) from 2.4 ℃ to 8 ℃. The utilization of the WRFM framework could remove bias on spatial variations with a reduced RMSE at a maximum of 4.3 ℃ while preserving WRF-simulated temporal variations of LST with a slight change of correlation coefficient (<±0.06). The WRFM-generated LSTs revealed more spatial and temporal details of thermal conditions, such as the diurnal evolution of LSTs, locations, and duration of LST hotspots. With the more resolved LSTs, policymakers can better design and plan urban infrastructures, such as power transmission systems and green spaces, to improve the urban environment's eco-environmental quality and thermal well-being.
Results/ConclusionsBy comparing with satellite observed LSTs, we found that the WRF simulation could well capture diurnal variations of LSTs with correlation coefficient > 0.8 but tended to have a bias on the magnitude and spatial variations with a wide range of RMSE (root mean square error) from 2.4 ℃ to 8 ℃. The utilization of the WRFM framework could remove bias on spatial variations with a reduced RMSE at a maximum of 4.3 ℃ while preserving WRF-simulated temporal variations of LST with a slight change of correlation coefficient (<±0.06). The WRFM-generated LSTs revealed more spatial and temporal details of thermal conditions, such as the diurnal evolution of LSTs, locations, and duration of LST hotspots. With the more resolved LSTs, policymakers can better design and plan urban infrastructures, such as power transmission systems and green spaces, to improve the urban environment's eco-environmental quality and thermal well-being.