Efficiency assessment of micro-climate change in land cover distinguishing ‎compared to some ‎supervised classification techniques for an arid urban ‎environment

Subject Areas : Natural resources and environmental management

NAJMAE SATARI ¹ , Malihe Erfani ^{2 *} , Fatemeh Jahanishakib ³

1 - Masters student, Faculty of Natural Resources, University of Zabol, Iran‎
2 - Assistant Professor, Department of Environmental Sciences, Faculty of Natural Resources, ‎University of Zabol‎, Zabol, Iran‎
3 - Assistant Professor,, Faculty of natural resources and environment, university of Birjand, Birjand, Iran

Received: 2022-02-04 Accepted : 2022-03-15 Published : 2023-05-22

Keywords: Feature extraction, Arid and semi-arid areas, Hard classification, Micro-climate, remote sensing,

Abstract :

The distinction between barren and build-up areas is one of the most important issues in land use/land cover mapping ‎in arid and semi-arid climates. In this regard, many researchers have tried to increase the accuracy of classification ‎using different methods that, some of which are complex and time-consuming. Therefore, the present study conducted ‎aimed to apply micro-climate change through the implementation of Local Climate Zoning (LCZ) algorithm in land ‎use identification with emphasis on the separation of build-up areas in one of the arid cities of Iran, and the efficiency ‎of the method by investigation the classification accuracy was compared with various supervised methods including ‎maximum likelihood, minimum distance, Fisher, KNN, fuzzy, artificial neural network and support vector machine. ‎The study area is Zahedan city, which has a very significant growth of build-up areas in recent decades. For this ‎purpose, four periods of Landsat satellite images year 2020 were used. Training samples were extracted from Google ‎Earth and the validation of the classification results was performed using 218 random points. The accuracy results ‎showed that the use of LCZ algorithm with overall accuracy and kappa coefficient of 96.33% and 0.95, respectively is ‎the highest and then the support vector machine and Fisher methods with overall accuracy of 86.61 and 83.03 and ‎kappa coefficient of 0.82 and 0.75, respectively. Therefore, for land use / land cover studies, the LCZ method that ‎considers the micro-climate, is proposed.‎