Speaker
Description
The monitoring methods, prediction drought and precipitation distribution, the possibilities of their application in Georgia are used in proposed work. Standardized Precipitation Index (SPI) was selected as the research parameter because it better expresses the pronounced drought trends than the Standardized Precipitation Evapotranspiration Index (SPEI). Using prediction model and algorithm, drought-vulnerable areas in Kakheti region were identified. By analyzing the station data on the territory of Georgia and satellite sources, it was determined that using the regression method of machine learning, it is sufficient to evaluate the data of 1960-2000 period for learning and the data of 2001-2022 period for training. SPI-3 three-month standardized precipitation index was selected as training object. R-instat software was used to calculate Pearson's correlation and other statistical parameters. During the creation of the model process various learning algorithms were trained (SVM, Regression Trees, Linear regression Models). The best result was shown by Bagged Trees. The training time of Bagged Trees Optimized Algorithm was recorded as 326.21 sec, prediction speed ~ 7900obs/sec, RMSE - 0.5046, R2-0.64, MSE-0.25466, MAE-0.38065, training process minimum leaf size 19, and 40 iterations are assigned for optimization. CHIRPS satellite data were taken for next generation of the model. The missed stations data that were used during the training period were filled with CHIRPS satellite data. For prediction, it was necessary to calculate a linear regression equation for each station. In the first case of forecast scenario, the amount of precipitation was determined from 0 cm to 10 cm. Gurjaani was highlighted, where forecast showed SPI value from -0.008 to -0.901, and Kvareli, SPI value from -0.002 to -0.138. Use of the presented ML model and algorithm for the analysis of precipitation distribution, drought monitoring and prediction is appropriate both in Kakheti and other regions too, in conditions of proper observation database.
