| Description |
The climate of urban areas is influenced by the composition and configuration of different land cover types. Urban forests increase human comfort in urban areas by cooling the environment through evapotranspiration and shade. A tradeoff of urban forests in semiarid and arid climates is that they require large quantities of irrigation water to maintain. This study aimed to quantify the relationship between urban vegetation and land surface temperature (LST). Datasets derived from high-resolution lidar and National Agriculture Imagery Program (NAIP) orthoimagery were used in a random forest algorithm to classify urban vegetation, human-made and natural surfaces at a 1-meter scale, in the Salt Lake Valley of Utah. The resulting classification accuracy was 94%. LST was retrieved from an Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) scene captured on a hot, summer day. Percentages of each land cover class were calculated per ASTER pixel. These composition variables were compared to LST using Pearson’s correlation analysis and were also used to create a multiple linear regression model. Percent deciduous tree cover was the variable most strongly correlated with LST, with a correlation coefficient of -0.55. Irrigated low-stature vegetation was also negatively correlated with LST (-0.33). Residuals from the multiple linear regression model varied over space, and additional dates of ASTER imagery are needed to determine whether these second-order spatial patterns are persistent. |
