Abstract: Objective　This study investigates the heat island effect in Tianjin City, China. Firstly, based on Landsat 8 remote sensing image data from 2013, 2015, 2018, and 2020, the land surface temperature (LST) is retrieved using the atmospheric correction method, and grade classification is performed via the mean-standard deviation approach. Subsequently, the relationship between LST and influencing factors in Tianjin is analyzed using the geographical detector model. The primary objective of this study is to investigate the temporal and spatial evolution of LST and the dynamic changes of the urban heat island (UHI) effect in Tianjin from 2013 to 2020. Additionally, the study aims to identify the dominant factors influencing LST in Tianjin and elucidate the interaction mechanisms among these factors. The findings are expected to provide scientific support and a theoretical foundation for optimizing the thermal environment in Tianjin and formulating sustainable urban development strategies, ultimately serving to improve the ecological environment and effectively mitigate the UHI effect during the urbanization process. Methods　We employ the atmospheric correction method to derive LST in Tianjin, apply the mean-standard deviation method to categorize the temperature retrieval results into different grades, and incorporate the geographical detector model to investigate the relationship between LST and influencing factors in Tianjin in this study. The specific steps are as follows: Firstly, the atmospheric correction method is used to retrieve the LST data of Tianjin from 2013 to 2020. Following the preprocessing of remote sensing images, LST retrieval results are obtained via calculating the land surface emissivity, radiance values, and applying Planck's formula. Subsequently, the retrieved LST results undergo normalization to mitigate errors arising from temporal differences among multi-temporal remote sensing images. After normalization, the retrieved LST values are segmented and classified into different temperature grades using the mean-standard deviation method. Finally, based on the geographical detector model, two approaches, namely single-factor detection and interaction detection, are employed to analyze the relationship between LST in Tianjin and influencing factors (e.g., land use, Digital Elevation Model (DEM), and slope), in order to elucidate the explanatory power of each factor and their interaction intensity. Results and Discussion　Based on Landsat remote sensing data, this study retrieves LST in Tianjin and investigates its temporal and spatial variations using the mean-standard deviation method and the urban heat island ratio index (UHIRI). Additionally, the geographical detector model is employed to quantitatively investigate the relationship between LST and its influencing factors. The results indicate that: (1) From 2013 to 2020, the proportion of UHI areas in Tianjin demonstrated a downward trend, and the UHIRI also exhibited an overall decreasing trend, indicating a gradual improvement in the UHI effect in the region. (2) Throughout the study period, land use consistently emerged as the dominant factor influencing LST in Tianjin, with an average q-value exceeding 0.62. And the normalized difference negetation index (NDVI), normalized difference built-up index (NDBI), and modified normalized difference water index (MNDWI) were identified as secondary influencing factors. While slope exhibited the least impact, serving as a minor contributing factor. (3) The combined influence of multiple factors on LST was more pronounced than that of individual factors. Land use, when interacting with other variables, consistently exerted the most substantial effect on LST, followed by NDVI, NDBI, and MNDWI. Furthermore, the interaction detection results of these four factors demonstrated significantly greater explanatory power compared to any single factor, underscoring the presence of interrelationships among the various factors influencing LST. Conclusions　This study comprehensively utilized the mean-standard deviation method, UHIRI, and the geographical detector model to explore the relationship between LST and its influencing factors in Tianjin. Based on the findings in this work and the specific conditions of Tianjin, the following recommendations are proposed: Mitigate and reduce the UHI effect by increasing vegetation coverage in built-up areas, protecting wetland parks to limit the expansion of impervious surface areas, and establishing ecological green corridors and other thermal buffer zones between the two primary heat island regions (the central urban area and Binhai New Area) to prevent their convergence into a more intense heat island. Additionally, mathematical prediction models can be implemented to forecast and analyze the future temporal and spatial evolution of UHIs, thereby providing more precise data and references for UHI research.

Key words: remote sensing images, heat island effect, atmospheric correction method, geodetector, Tianjin City