Outdoor thermal discomfort analysis in Harare, Zimbabwe in Southern Africa

Abstract

This study investigated the impact of seasonal land cover changes on human thermal comfort outdoors in Harare, Zimbabwe in Southern Africa. Multi-temporal thermal infrared and in situ air temperature data were used to develop simple linear regression model for retrieving air temperature from land surface temperature (r²=0.6897). Season specific simple linear regression models for deriving relative humidity from land surface temperature were also developed (r² greater than 0.78). When tested against in situ observations, the LST-based approach retrieved DI with high accuracy for each sub-season (mean percentage error less than 20%). The findings showed that vegetation fraction was higher (0.60) in the most comfortable post-rainy season than in the most thermally uncomfortable season (0.43), hot season. Outdoor thermal discomfort was high in hot season (mean DI of 31 °C), while the post-rainy season was the most thermally comfortable (mean DI of 19.9 °C). During the hot season, thermal discomfort was higher in densely built-up areas (DI greater than 27 °C) than in the northern areas where low-density residential areas, forests and most well maintained parks are located (DI less than 27 °C). It was concluded that Landsat 8 data detects seasonal land use/cover and thermal discomfort changes with high accuracy.

Keywords:

• Surface temperature
• vegetation fraction
• discomfort index
• heat island
• Landsat 8
• land cover classification
• remote sensing