Projection of urban land surface temperature: An inter- and intra-annual modeling approach

Abstract

An easily applied and efficiently exploited approach was proposed for the prediction of inter- and intra-annual land surface temperature (LST) in a city's central zone. Chengdu is chosen as the study area to demonstrate the applicability of the novel approach. Built-up areas were extracted using Landsat images in 2008, 2013, and 2019 and projected for 2030 using the cellular automata-Markov chain model, indicating that built-up areas grew nearly 109% during 2008–2019 in Chengdu and will continue to grow up to 2030. Employing the multiple linear regression, the LST of built-up areas in 2019 was predicted using the normalized difference built-up index (NDBI) and the mean LST of its surrounding pixels (LST-R3) in 2013 (R2 = 0.981). The model's intra-annual application for 2019 showed a mean LST deviation of 0.93 °C. According to the model's inter-annual application, the projected LST of the built-up areas in 2030 will increase to 36.5 ± 2.0 °C. The results can contribute to achieving a thermally comfortable urban environment by providing insights into the impact of urban expansion on the urban heat island (UHI) phenomenon under the limited availability of continuous LST data.