Your browser doesn't support javascript.
loading
Pixel level spatial variability modeling using SHAP reveals the relative importance of factors influencing LST.
Hu, Yuhong; Wu, Chaofan; Meadows, Michael E; Feng, Meili.
Afiliación
  • Hu Y; College of Geography and Environmental Sciences, Zhejiang Normal University, Jinhua, 321004, China.
  • Wu C; College of Geography and Environmental Sciences, Zhejiang Normal University, Jinhua, 321004, China. cfwdh@zjnu.edu.cn.
  • Meadows ME; College of Geography and Environmental Sciences, Zhejiang Normal University, Jinhua, 321004, China.
  • Feng M; Department of Environmental and Geographical Science, University of Cape Town, Cape Town, 7700, South Africa.
Environ Monit Assess ; 195(3): 407, 2023 Feb 16.
Article en En | MEDLINE | ID: mdl-36795252
As an important indicator of the regional thermal environment, land surface temperature (LST) is closely related to community health and regional sustainability in general, and is influenced by multiple factors. Previous studies have paid scant attention to spatial heterogeneity in the relative contribution of factors underlying LST. In this study of Zhejiang Province, we investigated the key factors affecting daytime and nighttime annual mean LST and the spatial distribution of their respective contributions. The eXtreme Gradient Boosting tree (XGBoost) and Shapley Additive exPlanations algorithm (SHAP) approach were used in combination with three sampling strategies (Province-Urban Agglomeration -Gradients within Urban Agglomeration) to detect spatial variation. The results reveal heterogenous LST spatial distribution with lower LST in the southwestern mountainous region and higher temperatures in the urban center. Spatially explicit SHAP maps indicate that latitude and longitude (geographical locations) are the most important factors at the provincial level. In urban agglomerations, factors associated with elevation and nightlight are shown to positively impact daytime LST in lower altitude regions. In the urban centers, EVI and MNDWI are the most notable influencing factors on LST at night. Under different sampling strategies, EVI, MNDWI, NL, and NDBI affect LST more prominently at smaller spatial scales as compared to AOD, latitude and TOP. The SHAP method proposed in this paper offers a useful means for management authorities in addressing LST in a warming climate.
Asunto(s)
Palabras clave
Texto completo
Añadir a Mi BVS
Imprimir
XML
PubMed Links
Buscar en Google

Texto completo: 1 Colección: 01-internacional Base de datos: MEDLINE Asunto principal: Monitoreo del Ambiente / Calor Tipo de estudio: Prognostic_studies Idioma: En Revista: Environ Monit Assess Asunto de la revista: SAUDE AMBIENTAL Año: 2023 Tipo del documento: Article País de afiliación: China Pais de publicación: Países Bajos
Texto completo
Añadir a Mi BVS
Imprimir
XML
PubMed Links
Buscar en Google

Texto completo: 1 Colección: 01-internacional Base de datos: MEDLINE Asunto principal: Monitoreo del Ambiente / Calor Tipo de estudio: Prognostic_studies Idioma: En Revista: Environ Monit Assess Asunto de la revista: SAUDE AMBIENTAL Año: 2023 Tipo del documento: Article País de afiliación: China Pais de publicación: Países Bajos