The inverse correlation between NDVI and LST is widely known for its long time series. However, when more specific statistical tests were performed, subtle differences in the correlation behavior over time are more clearly observed. In this work, regression analyses were performed between NDVI and LST at intervals of approximately 10 years, quantifying this relationship for an area of transition from vegetation to urban occupation from 1985 to 2018. The removal of vegetation cover (reduction of 51% to 7% in grassland and 14.4% to 0.6% in forest) to occupy impermeable surfaces ( increase of 31% to 91% in urban areas) caused an average LST increase of 4.18 °C when compared to the first and last decades of the historical series. Temporal analysis allowed us to verify the increase in temperature in the four seasons. The largest difference was 6.36 °C between the first and last decade of autumn, 4.40 °C in spring, 4.09 °C in summer, and 2.41 °C in winter. The results also show that LST has a negative correlation with NDVI, especially in urban areas, with an increase in this correlation during the period (1989: R = −0.55; 1999: R = −0.58; 2008: R = −0.59; 2018: R = −0.76). Our study results will help policymakers understand the dynamics of temperature increases by adding scientifically relevant information on the sustainable organization of the urban environment.

O infravermelho termal (TIR - ThermalInfraRed) é uma porção do espectro eletromagnético com várias aplicações no Sensoriamento Remoto, tais como: geologia, climatologia, análises de processos biológicos, análises geofísicos, avaliação de desastres e detecção de mudanças, entre outras. No TIR a emissão de radiação dos alvos é dominante, comparado com a reflexão, e esta radiação é uma função de duas variáveis, a emissividade e a temperatura do alvo. Para estudos no TIR é necessário estimar com precisão a temperatura e/ou a emissividade a partir da radiação medida, e isto é um problema devido a relação não linear existente entre estas variáveis e a radiação medida. Por isso, nos últimos 40 anos vários pesquisadores têm desenvolvido métodos visando minimizar este problema, porém, todos estes métodos possuem restrições em suas aplicações. Assim, este trabalho tem como objetivo revisar os principais métodos propostosna literaturafacilitando a sua compreensão, reprodução, além de criar umasínteseque permitaráo leitor a escolha do método mais adequado a determinadas situações

The land surface emissivity (LSE) is an important property in the remote characterization of targets. The estimation of LSE involves the application of an indeterminate function of several variables in radiance data contaminated by a complex modeling atmosphere. In this work, four methods of temperature-emissivity retrieval were evaluated from the radiance data of Advanced Spaceborne Thermal Emission and Reflection Radiometer in a coastal dune field of a sub-tropical region: Normalization Emissivity Method (NEM), Reference Channel Method (RCM), Temperature Emissivity Separation (TES) and Alpha Residuals Method (α-Residual). In order to compare the emissivity curves with a minimal error of initial hypothesis, an average brightness temperature of the dune field was estimated and used to select the signature of pure quartz from the Spectral Library of the Geologic Remote Sensing Laboratory both at 302.15 K. In this article, it was assumed that the atmospheric contribution was corrected with parameterization of physical variables in the Moderate Resolution Atmospheric Transmission (MODTRAN) codes, despite the well-known restrictions for humid atmospheres. As a result, the NEM and RCM methods presented the best approximation when compared to the emissivity reference curve. With regard to the average temperature, NEM and RCM presented the best approximation, both with 302.37 K. TES presented 303.25 K. The TES method obtained a value close to the reference value with a difference of 1.12 K, while for NEM and RCM the difference was 0.1 K. α-Residual method did not generate temperature values, since it does not consider this variable in the estimation of the alpha spectrum. It is dependent on the radiance values and, since it has a very similar geometry related to the RCM and NEM, it can reinforce that the atmospheric effects have an important contribution, at least, in these three methods. The methods tend to overestimate the emissivity due to a complex parameterization of physical variables for atmospheric modeling, in addition to the heterogeneity of the target. In contrast, a good performance was obtained in the individualization of the dune field, satisfactorily preserving the geometry of the spectral curve of the quartz.