Research Article

Short-Term Daily Univariate Streamflow Forecasting Using Deep Learning Models

Table 4

Performance comparison of the proposed models for different input time lags in Borkena station.

BorkenaT + 1T + 2T + 3T + 4
RMSEMAEMAPER2TTPE (sec)RMSEMAEMAPER2TTPE (sec)RMSEMAEMAPER2TTPE (sec)RMSEMAEMAPER2TTPE (sec)

MLP1.090.470.850.980.240.890.380.760.990.161.060.350.190.980.120.810.290.340.990.29
GRU1.150.640.840.981.290.850.310.160.990.790.910.350.180.980.991.350.691.410.971.31
S-LSTM1.070.380.290.980.300.870.360.260.990.541.020.640.990.980.811.530.961.720.960.58
Bi-LSTM1.060.390.310.980.880.900.460.760.991.551.120.510.950.981.550.980.440.530.985.3

TTPE(training time per epochs). The bold values indicate the best performance score for each time lag.