Machine Learning-Based Modeling of Air Temperature in the Complex Environment of Yerevan City, Armenia

Machine learning (ML) was used to assess and predict urban air temperature (Tair) considering the complexity of terrain features in Yerevan (Armenia). The estimation performed based on Partial Least-Squares Regression (PLSR) model with a high number (30) input variables. relevant parameters include newly purposed modification spectral index IBI-SAVI, which turned out strongly impact Tair prediction together land surface (LST). Cross-validation analysis predictions across station-centered 1000 m circular area revealed quite correlation (R2Val = 0.77, RMSEVal 1.58) between predicted measured from test set. It concluded remote sensing is an effective tool estimate distribution where dense network weather stations not available. However, further developments will incorporation additional stations, such as precipitation wind speed, well use non-parametric ML techniques.