ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Download
Citation
Articles | Volume IV-2/W5
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., IV-2/W5, 513–517, 2019
https://doi.org/10.5194/isprs-annals-IV-2-W5-513-2019
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., IV-2/W5, 513–517, 2019
https://doi.org/10.5194/isprs-annals-IV-2-W5-513-2019

  29 May 2019

29 May 2019

RETRIEVAL OF LOCAL CLIMATE ZONES AND THEIR LINKAGES TO LAND SURFACE TEMPERATURE USING REMOTE SENSING IN HARARE METROPOLITAN CITY, ZIMBABWE

T. D. Mushore1,2 T. D. Mushore
  • 1Physics Department, University of Zimbabwe, P.O. Box MP167, Mount Pleasant, Harare, Zimbabwe
  • 2Discipline of Geography, School of Agricultural, Earth and Environmental Sciences, University of KwaZulu-Natal, P/Bag X01, Scottsville, Pietermaritzburg 3209, South Africa

Keywords: Data Scarcity, Satellite Data, Temperature Variations, Climate, Urban Land Use, Spatial Tools, Urban Heat Island

Abstract. This study sought to determine Local Climate Zones (LCZs) in Harare metropolitan City, using Landsat 8 multi-spectral and multi-temporal data. The World Urban Database and Access Portal Tool (WUDAPT) and Support Vector Machine classifiers were applied. Training datasets were extracted from Google Earth as prescribed by the WUDAPT procedure. Before image classification, we tested the separability of the LCZs, using the Transformed Divergence Separability Index (TDSI) based on the digitized training datasets and Landsat 8 data. Derived LCZs were then linked with Landsat 8 derived Land Surface Temperature (LST) for the cool and hot seasons. TDSI values greater 1.9 were obtained indicating that LCZs were highly separable. Comparatively, the WUDAPT method produced more accurate LCZs results (Overall accuracy = 95.69%) than the SVM classifier (Overall accuracy = 89.86%) based on seasonal Landsat 8 data. However, SVM derived accuracies were within the acceptable range of at least 80% (overall accuracy) in literature. Further, LST was observed to be high in LCZs with high built-up density and low vegetation proportion, when compared to other zones. Due to high proportion of vegetation, sparsely built areas were at least 1 °C cooler. Although LCZs are usually linked at 2 m air temperature, they also strongly explain LST distribution. This work provides insight into the importance of mapping LCZs in third world countries where such information remains scarce.