O clima e as cidades: avaliação do índice de correlação entre temperaturas medidas por sensores termais do satélite Landsat-8 e sensores de superfície na cidade de Patos de Minas - MG

Abstract

Investigations aimed at understanding the urban climate have come to rely, especially in recent decades, on different instruments for its diagnosis, with emphasis on geotechnological tools added to a set of mathematical and computational techniques for the treatment of geographic information . In this group, the use of thermal images aimed at mapping the climateit has acquired a special prominence, mainly due to its rapid ability to diagnose parameters of soil surface temperatures after treatment. This technique has enabled the development in recent years of several studies on urban climate in different geographical conditions, which have started not to use surface data collection. This scenario opened the possibility of inquiries about the actual index of representativeness of the temperature close to the surface of the soil, especially considering the particularities of the local geography and also the seasons.In this context, this study aimed to evaluate the temperature correlation index indicated by the thermal bands of the Landsat 8 satellite with temperatures collected by surface sensors in different seasons of the year in the city of Patos de Minas - MG. Therefore , n the study was carried out l evantamento and analysis of scientific works on the subject, where it was possible to identify the pattern of scientific publications on the subject in recent years. In the next stage, the historical contextualization and survey of the main physico-social characteristics of the city of study, Patos de Minas-MG, was performed, proving the feasibility of studies in it. The third step was a technical roadmap for the construction of low-cost equipment for data collection in the field and a description of the methodology adopted for the production of cartographic materials, both at ambient and surface temperatures. Field data were collected for one year, between August 2018 and July 2019, respecting the time of the satellite's passage, making future correlations viable. In the case of surface temperature maps, Landsat 8 images , band 10, of the thermal infrared channel / TIRS were used, treated in the Qgis software , using the satellite sensor calibration constants, obtained on the USGS6 website, for the conversion of the numbers radiant energy. With the data in hand, it was performed in the Minitab ® statistical program , Pearson's correlation method, which identified some response patterns. As a result, this study points out that during extreme seasons, summer and winter, surface temperatures, indicated by band 10 of Landsat-8, demonstrate a low index of correction with field data, which points to the need for refinement or even a more accurate treatment of the data, because it does not represent the actual surface temperature. An aggravating factor is that in the summer months, due to the high level of cloud cover, there is an impracticability of image processing, and it is not possible to perform the surface temperature mapping. On the other hand, the months of October, April and May, which correspond to the transition periods of the seasons, autumn and spring, presented a favorable correlation and a high level of significance. This study concludes, using Pearson's linear correlation method, that the adoption of satellite climate analysis methods does not present precise answers in terms of urban temperature in medium-sized cities, being necessary, depending on the purpose of the study, to field data collection to fine- tune the information generated by the software.