Efficacy of the Gaussian fitting model in predicting diurnal hourly air temperature based on Landsat-8 OLI TIR-1 channel

Abstract Near surface air temperature ( $$\:{T}_{a}$$ ) represents a vital climatic parameter essential for comprehending and modeling complex surface phenomena, as well as for elucidating the dynamics of hydrological processes. This study integrated $$\:{T}_{a}$$ data acquired using smartphone weather application and portable anemometer, and land surface temperature ( $$\:{T}_{s}$$ ) to predict diurnal near-surface hourly air temperature for data-scarce area by fitting these data into the Gaussian function method. The spatial continuity of $$\:{T}_{a}$$ was computed from the air temperature data collected using the smartphone weather app. The linear scaling approach was applied to minimize bias in the $$\:{T}_{a}$$ predicted using smartphone weather app. Thermal infrared (TIR-1) band of Landsat-8/9 Operational Land Imager (OLI) was used to derive $$\:{T}_{s}$$ . The computed diurnal hourly $$\:{T}_{a}$$ revealed a polynomial relationship with time. Comparative analysis results of the interpolated and measured hourly $$\:{T}_{a}$$ revealed the measured hourly $$\:{T}_{a}\:$$ being often higher than the predicted hourly $$\:{T}_{a}\:$$ across the surveyed hours. The linear scaling bias correction technique corrected the interpolated hourly $$\:{T}_{a}$$ to the values closer to the measured values. Moreover, $$\:{T}_{s}$$ was found to be generally higher than both measured and interpolated $$\:{T}_{a}$$ , though $$\:{T}_{s}$$ lower than $$\:{T}_{a}$$ was noted in the vicinity of water bodies, which points to the inability of the interpolated $$\:{T}_{a}$$ to account for the influence of water. The linear regression analysis results of the relationship between the measured $$\:{T}_{a}$$ and remote sensing-based $$\:\:{T}_{s}$$ was established, with r 2 of 0.69. The diurnal hourly $$\:{T}_{a}$$ pattern extrapolated by the Gaussian function also revealed a polynomial relationship between the simulated hourly $$\:{T}_{a}$$ and time. Analysis of the root mean square error (RMSE) and observation standard deviation ratio (RSR) values revealed good performance of the Gaussian fitting technique in simulating diurnal hourly $$\:{T}_{a}$$ in the study area, with RMSE and RSR values of 1.41 and 0.55 respectively. Most importantly, the simulated $$\:{T}_{a}$$ based on $$\:{T}_{s}$$ accounted for the influence of land use, such as water bodies. The results of this study reveals that the Gaussian fitting model can predict diurnal hourly $$\:{T}_{a}$$ based on $$\:{T}_{s}\:$$ data derived from Landsat-8 OLI TIR-1 channel. Ultimately, this research highlighted the continuous role of geospatial technology in addressing environmental and climate change issues.