Table 1 Daily and hourly summary table.
From: Artificial intelligence for streamflow prediction in river basins: a use case in Mar Menor
Papers | Region | Algorithms | Features | Horizon | Best model | Metrics |
|---|---|---|---|---|---|---|
Kelantan River, Malasya | EELM, ELM, SRV | River flow of correlated lag times, rainfall, water lever and other physical parameters | Daily | EELM | NSE = 0.799 (EELM) WI = 0.938 (EELM) RMSE = 314.16 (EELM) MAE = 100.42 (EELM) | |
Tajan River, Iran and Hongcheon RIver, South Korea | MSTree, MARS, EEMD | Daily rainfall and runoff data | Daily (1,2,3,4) | EEMD-MSTree | NSE = 0.855 (EEMD-MSTree) RMSE = 8.677 (EEMD-MSTree) WI = 0.960 (EEMD-MSTree) | |
Fujang River, China | OP-ELM, MARS, ANFIS-PSO, MSTree | Streamflow and precipitation | Daily | OP-ELM | RMSE = 155 (OP-ELM) NSE = 0.751 (OP-ELM) | |
North Folk, Chehalis, Carson and Sacramento Rivers | SVR, ANN-BP, ELM | Precipitation, max and min temperature and its lags | Monthly and daily | SVR | RMSE = 41.86 (SVR, daily, Sacramento) R = 0.58 (SVR) NSE = 0.33 (SVR) | |
Alexander Stream in Israel, Koksilah River in British Columbia, Canada Upper Boe River | ANN, SVR, WANN, WSVR | Daily streamflow and precipitation, min and max temperature and their lags, high and low frequency components from wavelet | Daily (1,2,3) | WANN | FSE = 3.854 (WANN, 1 day, Mediterranean) R2 = 0.965 (SVR) NSE = 0.681 (WANN) | |
Coruh River Basin, Turkey | DAR, SM, DAR-SM, DAR-SM-ANN | Drainage areas, streamflow in donor basins, | Daily | DAR-SM-ANN | NSE = 0.991 (DAR-SM-ANN) RSR = 0.089 (DAR-SM-ANN) R2 = 0.991 (DAR-SM-ANN) | |
River Arno, Italy | ANN | Rainfall and hydrometric data | Hourly (1–12 h) | ANN | RMSE = 0.4 (ANN, 6h) | |
Shenzen, China | ANN, LSTM, ANFIS, SOBEK | Meteorological data, hourly streamflow, river profiles | Hourly | LSTM | NSE = 0.976 (LSTM) R2 = 0.977 (LSTM) RMSE = 2.109 (SOBER) | |
Kelantan and Muda Rivers, Malasya | LSTM, CNN, ConvLSTM | Hourly streamflow and rainfall | Hourly (1,3,6) | ConvLSTM | NSE = 0.995 (ConvLSTM) | |
Audun Basin, China | DIFF-FFNN-LSTM, MLR, AR, ARIMA, LSTM, FFNN and combinations | Hourly streamflow and precipitation | Hourly | DIFF-FFNN-LSTM | RMSE = 9.31 (DIFF-FFNN-LSTM) MAE = 3.63 (DIFF-FFNN-LSTM) NSE = 0.99 (DIFF-FFNN-LSTM) | |
State of Iowa, USA | NRM, GRU, Ridge Regression, RF, LSTM | Previous rainfall, streamflow and evapotranspiration + forecasted rainfall and evaporation | Hourly (up to 120 h) | NRM | NSE \(\sim\)0.7 (NRM) r \(\sim\)0.9 (NRM) | |
Johnson County, Iowa | ConvBIGRU, ConvGRU | Past streamflow and precipitation (36 lags) | Hourly (up to 36 h) | ConvGRU | NSE: 1–0.82 (ConvGRU) |
