Table 2 Performance comparison with closely related water quality forecasting models.
From: Time series forecasting of chlorophyll-a concentrations in the Chesapeake Bay
Study | Method(s) | Time Window | Location | RMSE (mg/m3) |
|---|---|---|---|---|
Eze et al.20 | EEMD-LSTM | 6–24 h | Scottish salmon farm (marine coast) | 0.0356 (6 h), 0.0417 (24 h) |
Lin et al.24 | WA-LSTM, WA-BPNN | 2 weeks | Neuse River Estuary, USA (estuarine) | 1.27–12.05 (WA-LSTM), 1.14–8.45 (WA-BPNN) across 10 sites |
Barzegar et al.23 | LSTM, CNN-LSTM | 15 min | Small Prespa Lake, Greece (freshwater) | 0.614 (LSTM), 0.579 (CNN-LSTM) |
Yussof et al.18 | LSTM, CNN | 8 days | West Coast of Sabah, Malaysia (marine coast) | 3402.142 (LSTM), 4361.724 (CNN) |
Zheng et al.22 | LSTM, EFDC-LSTM | 1–3 h | BeiYun River, China (freshwater) | 0.94 (LSTM), 1.86 (EFDC-LSTM) |
Cen et al.19 | LSTM | 5 days | East China Sea (marine coast) | 0.0223–4.6475 across 4 regions |
Gupta et al. (our study) | LSTM | 1 week | Chesapeake Bay (estuarine) | 0.121–0.199 across 3 regions |
