Land Surface Temperature Variation and Related Factors in Barranquilla, Colombia

Abstract

One of the most important factors affecting human well-being, particularly in metropolitan settings, is temperature variation. Warmer temperatures in summer and cooler temperatures in winter increase concern about electricity and gas consumption. This study compares the effectiveness of the two predictive models while examining the seasonal patterns and trends of land surface temperature (LST) variation in Barranquilla, Colombia, from January 2001 to December 2020. The National Aeronautics and Space Administration (NASA) website was where the data for this study were found. The patterns and trends of LST over the past 20 years were examined using simple linear regression and cubic spline techniques. Use the initial 70% of the data as a training set and the remaining 30% as a testing set. Multiple Linear Regression to examine how time-lag LST and NDVI (Normalized Difference Vegetation Index) impact one another throughout the seasons. After fitting a simple linear regression on seasonally adjusted LST, it was discovered that during the past 20 years, the LST in Barranquilla, Colombia, has increased most in area 4 (about 0.171�C per decade) and declined most in region 6 (about -0.409�C per decade). Both increments were statistically significant with a p-value less than 0.05. While the seasonal patterns for LST and NDVI showed an increment in the opposite direction, Specifically, LST attained its peak around March, while NDVI attained its lowest point during the same time of the year. Moreover, two predictive models (ARIMA, ARIMAX, and MLR) were considered. It was found that when incorporating an exogenous variable (NDVI) into the model (ARIMAX), the performance on the train set slightly improved from considering only its own time-lag variable in the ARIMA model. Additionally, MLR was also considered a baseline model. The performance of MLR measured by the RMSE shows values slightly higher than those of the ARIMA model.Changes in land surface temperature (LST) will be examined in this study.For these data sets, three predictive model techniques were investigated: multiple linear regression, ARIMA, and ARIMAX. Choosing the ARIMA model made obvious, given its purpose was to investigate how its lag factors would affect the current value. An ARIMA (p,d,q) combination of historical values and historical errors, in particular, can be used to represent a non-seasonal time series.

??????????????????????????????????????????????????????????????? ?????????????????????? ???????????????????????????? ????????????????????????????????????????????????????????? ????????????????????????????????????????? ????????????????????????????? ??????????????????????????????????????????????? ?????????????????????????????????????????????????????????????????? (LST) ?????????????????? ??????????????? ?????????????????? ?.?. 2544 ?????????? ?.?. 2563 ??????????????????????????????????????????????????????????????????????????????????? (NASA) ??????????????????? LST ?????? 20 ????????????????????????????????????????????????????????????????????????? Cubic Spline ??????????????????????????????? ????????? 70% ???????????????????? ?????????????????? 30% ??????????????????????????????????????????????????????? ?????? LST ?????? ??? NDVI (Normalized Difference Vegetation Index) ????? LST ? ??????????????? ??????????????????????????????????????? LST ???????????????????? ??????????? 20 ??????????? LST ?????????????????? ??????????????? ????????????????????????????????? 4 (region 4) (?????? 0.171�C ?????????) ???????????????????????????? 4 (region 4) (?????? -0.409�C ?????????) ????????????-????????????????????????????????????? p - value ???????? 0.05 ???????????????????????????????? LST ??? NDVI ?????????????????????????????????? ????????????????? LST ???????????????????????????? ???????? NDVI ?????????????????????????????????? ????????? ?????????????????????????????????????? (ARIMA ??? ARIMAX) ????????????????????????? (NDVI) ????????????? (ARIMAX) ???????????????????????????????????????????????????????????????????????????????????? ARIMA ???????????????????????????????????? MLR ????? ?????????????? MLR ????????? RMSE ????????????????? RMSE ??? ARIMA ???????? ???????????????????????????????????????????????????????????? (LST) ?????????????????? ?????????????????????????????????????????: ?????????????????????? ARIMA ??? ARIMAX ????????????? ARIMA ???????????? ???????????????????????????????????????????????????????????????????????????? ?????????? ARIMA (p,d,q) ??????????????????????????????? ??????????????????????????????????????????????