Table 1 Hybrid models for forecasting streamflow over different time scales.

Three diverse watersheds

in Germany

AR-RBF, MLP-RF-PR

R² = 0.7–0.84

Granata et al.⁶⁷

Ten watersheds’ data set

gathered from CAMELS, US

EFS-KESVR-EMA, EFS-LSTM-EMA

NSE = 0.73–0.94

Xu et al.⁶⁸

Two USGS stations, US

hybrid EMD-RFR, EMD-Bagging,

EMD-AdaBoost, EMD-ANN

R = 0.97, RMSE = 0.33, MAE = 0.17,

NSE = 0.94

Heddam et al.⁶⁹

Dez River, Iran

SVMD-MLP-PSO

R² = 0.89, RMSE = 13.91, NSE = 0.88

Parsaie et al.⁷⁰

Nile River at the High

Aswan Dam, Egypt

MLP-EO, MLP-HGSO, MLP-NRO

MAE = 1.35, RMSE = 2.35, R = 0.92

Ahmed et al.⁷¹

Gaula barrage site in

Uttarakhand state of India

ANFIS, ANN, WANN

R = 0.99, RMSE = 5.51 (ft³/sec), WI = 0.96,

COE = 0.99

Shukla et al.⁷²

Yuelai New City, China

LightGBM-SSA

NSE more than 0.9, peak flow forecasting

error within 18%

Cui et al.⁷³

Han River, China

VMD-DBN-IPSO

NSE more than 0.8, peak flow forecasting

error within 20%

Xie et al.⁷⁴

Pahang River, Malaysia

ANFIS-FFA, ANFIS

R = 1, RMSE = 0.98, MAE = 0.36, NSE = 1

Yaseen et al.⁷⁵