We employ regional analytic techniques to estimate suspended sediment load in watersheds that lack statistics due to their size and the absence of sediment measuring stations. In order to model the suspended sediment load for the Poonel watershed in Gilan and the Kowter watershed in West Azerbaijan province, this research used gene expression planning (GEP) methods, an adaptive network-based fuzzy inference system (ANFIS), support vector regression (SVR), and autoregressive integrated moving average (ARIMA). Finally, a comparison was made between these methods. For this purpose, information was gathered from 1979 to 2016 from the hydrological stations at Poonel and Kowter, as well as data on the flow rate, sediment discharge, precipitation height, minimum, maximum, and average temperatures of two synoptic stations at Bandar Anzali and Mahabad. Any precipitation that occurred on a day with a mean temperature below zero was classified as snowy precipitation in order to identify the kind of regime. We took into consideration the rainfall regime of the Poonel watershed (0.02% snowfall) and the snow regime of the Mahabad watershed (10% snowfall). The objective was to simulate the flow rate of sediment through five inputs: precipitation height, minimum temperature, maximum temperature, average temperature, and flow rate. The findings indicated that the SVR with polynomial kernel was the best model for the Kowter basin (snow regime), with a root mean square error of 0.49 and a coefficient of explanation of 0.74; similarly, the Poonel basin (rain regime) exhibited an optimal model for the SVR with polynomial kernel, with a root mean square error of 0.42 and a coefficient of explanation of 0.75. After all 14 models were ranked, it was discovered that the models associated with the Poonel basin performed better; these models had ranks ranging from 1 to 5. This indicates that model performance is affected by a weighted precipitation regime.