Evaluating the Effect of Roughness Length Index on Modelling the Maximum Intensity of Urban Heat Islands Using Remote Sensing and Geospatial Information Systems (Case study: District 22, Tehran)

Introduction: In recent year, modeling and identifying spatial distribution patterns of urban heat islands phenomenon with the aim of planning to face the effects of this phenomenon and predicting the provision of infrastructure needed to provide better thermal comfort to citizens has increased. Oke's model is one of the prominent models in this field that simulates the maximum intensity of the heat island based on the urban canyon’s aspect ratio index. The dependence of the Oke's model on the climatic and physical conditions of cities requires that this model be before being used in any urban area to be modified if needed. Considering the effects of aerodynamic resistance of urban canyons (roughness length) on the maximum intensity of the heat island, considering the index of this factor in the localization process of the model can affect the accuracy of the results. In this study, it has been tried to localize Oke's model in an area of the 22nd district of Tehran, and to investigate the effect of roughness length in this process. Preparation of temperature data of urban canyons is one of the important challenges in the modeling process. Research shows that the air temperature in the central and suburban areas at night is close to the land surface temperature (LST), and canyons ’s LST can be used as a convenient approximation of the air temperature. Therefore, in this study, it was tried to solve the problem of preparing temperature data by using satellite thermal sensors and using an appropriate LST retrieval algorithm. Calculating the geometric and aerodynamic strength indices of the canyons in the modeling process is complex and time-consuming due to the need to perform various spatial processing, and therefore, it is another challenge in this field. Geospatial information systems (GIS) with the ability to store the topological relationships of geographical features and analyze them can facilitate the calculation of these indicators. Therefore, in this study, geospatial information systems have been used.

Materials and Methods: In this study, in order to prepare the required temperature data, ASTER sensor data and meteorological data of the nearest meteorological station to the study area during the period of 2016 to 2022 were used. These data were processed in MATLAB software using a separate window algorithm (SWA) and the LST and the maximum intensity of surface heat islands in the study area were calculated. Then, canyons ’s aspect ratio and roughness length indicators and also, their simulated maximum intensity of heat island (based on Oke’s model) were calculated by processing digital maps in the ModelBuilder program in the ArcGIS software environment. After dividing the study area into training and check areas, localization of the maximum heat island intensity model was performed in two different cases. In the first case, the coefficients of the local model of the training area were calculated by considering the aspect ratio index. For this purpose, the canyons were classified into 11 different classes based on their aspect ratio index and their simulated and measured maximum intensity heat island were calculated and through regression analysis of these two sets of data, the localized Oke’s model was obtained. In the second case, the canyons of the training area were classified into two separate classes based on their roughness length index. Then, the first and second class were classified into 8 and 3 separate groups based on their aspect ratio. By calculating the simulated and measured maximum heat island intensity of each group and using regression analysis, the localized Oke’s model was determined for each of the two mentioned classes.



Results and discussion: by validating the obtained models in the check area, the values of R^2, ρ, RMSE and MAE obtained from regression in the first case were 0.53, 0.73, 1.18 ± and 0.98, respectively, and in the second case, 0.80, 0.89, 1.05 ± and 0.87, respectively. Comparison of these results shows that the inclusion of the aerodynamic resistance index in the process of modeling the maximum intensity of the heat island, while increasing the correlation coefficients and the regression detection coefficient, has increased the accuracy of the results obtained from the local model and improved the model.