Introduction:Drought is a gradual and pernicious natural catastrophic event with global socioeconomic and environmental consequences. This is a highly perilous climate-related catastrophe that has a substantial effect on both the environment and human existence. The main aim of drought risk analysis is to improve drought management and forecasting techniques. It focuses on various aspects of droughts, such as their size, duration, intensity, and spatial extent. Droughts develop slowly, and their consequences become apparent over a long period of time. To monitor and predict droughts, various drought indices are used to measure the deviation of meteorological variables, such as precipitation, from their long-term averages . In general, drought monitoring relies on several indices, including the Palmer Drought Severity Index (PDSI) , Effective Drought Index (EDI), Reconnaissance Drought Index (RDI), Standardized Precipitation Evapotranspiration Index (SPEI) , Weighted Anomaly Standardized Precipitation Index (WASP) , and Standardized Drought Indices (SDIs) . A comprehensive list of these indices and their descriptions can be found in Zargar et al. . However, the most widely used method for drought monitoring using drought indices is the Standardized Precipitation Index (SPI), primarily because it only uses one parameter (precipitation).The standard precipitation index can be modeled by physical modeling that requires a lot of effort and information, or conceptually by using intelligent models. The results obtained from physical and experimental models due to its non-linear nature, the complexity of this index has many fluctuations. Therefore, nowadays artificial intelligence models are more attention due to their non-linear nature and reduction of time to predict climatology issues.Therefore, the goals of this research are to accurately estimate the standard precipitation index using the support vector regression hybrid model with firefly and gray wolf algorithms.

Materials and methods:In this research, in order to model the standard precipitation index of Borujerd, Durood, Delfan and Seleh plains located in Lorestan province, the support vector regression model with firefly and gray wolf algorithms was used. The precipitation parameter (P) was used as an input and the 12-month standard precipitation index (SPI) was used as the output parameter of the model in the monthly time period of 2002-2022 for the rain gauge stations of Borujerd, Durud, Delfan, and Selesh. The general purpose of intelligent models is to express the relationship between variables whose complexity is difficult to find in nature with high uncertainty. The standard precipitation index is one of the important climatology parameters, whose estimation is of great importance in future time steps. For this purpose, in order to reduce the error and estimate the parameter of the standard precipitation index with high accuracy, using the lowest input parameters, the mentioned method was used, which provides much better performance compared to the approximate methods. The purpose of this research is to understand this natural complexity between meteorological parameters and provide a model for future forecasting, and since the amount of standard precipitation index is more important than other parameters, this parameter was chosen as the target variable. It should be noted that for modeling, 80% of the data for training and the remaining 20% for testing were randomly selected to cover a wide range of data types.



Results and Discussion: In order to model the standard precipitation index of Lorestan plains, the support vector regression model with gray wolf and firefly algorithms was used. In all the investigated hybrid models, the radial basis function kernel has shown the best performance compared to other investigated kernels. Also, according to the evaluation criteria according to Table 3, it can be seen that the support vector-nightworm hybrid regression model at Borujerd station has a correlation coefficient of 0.945, the lowest root mean square of 0.042, the lowest average absolute value of error 0.036, and the highest Nash Sutcliffe coefficient. 0.952, at Durud station the correlation coefficient is 0.955, the lowest root mean square is 0.038, the lowest average absolute error value is 0.030 and the highest Nash Sutcliffe coefficient is 0.961, at Delfan station the correlation coefficient is 0.960, the lowest average root value squared 0.032, the lowest average absolute value of the error 0.028 and the highest Nash Sutcliffe coefficient 0.970 and in the station of the series the correlation coefficient 0.958, the lowest root mean square 0.035, the lowest average absolute value of the error 0.031 and the highest coefficient Nash Sutcliffe 0.968 has shown a better performance in the verification phase.

Conclusion:In general, it is suggested to use the support vector-firefly hybrid regression model as a model with a small error to solve nonlinear problems with large dimensions with a suitable speed in convergence towards an optimal solution. It can also be considered as a new solution in predicting the amount of standard precipitation index in order to make appropriate management decisions to improve water resources and prevent drought.