Novel Deep Learning Transformer Model for Short to Sub‐Seasonal Streamflow Forecast
Abstract Accurate short‐to‐subseasonal streamflow forecasts are becoming crucial for effective water management in an increasingly variable climate. However, streamflow forecast remains challenging over extended lead times, uncertainty in meteorological inputs, and increased frequency and variabilit...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2025-07-01
|
| Series: | Geophysical Research Letters |
| Online Access: | https://doi.org/10.1029/2025GL116707 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850075319741972480 |
|---|---|
| author | Anukesh Krishnankutty Ambika Kshitij Tayal Vimal Mishra Dan Lu |
| author_facet | Anukesh Krishnankutty Ambika Kshitij Tayal Vimal Mishra Dan Lu |
| author_sort | Anukesh Krishnankutty Ambika |
| collection | DOAJ |
| description | Abstract Accurate short‐to‐subseasonal streamflow forecasts are becoming crucial for effective water management in an increasingly variable climate. However, streamflow forecast remains challenging over extended lead times, uncertainty in meteorological inputs, and increased frequency and variability in extreme weather and climate events. We implemented a Future Time Series Transformer (FutureTST) model for streamflow forecasting that separately integrates past meteorological and streamflow data while incorporating future weather conditions. FutureTST achieves a mean Nash‐Sutcliffe Efficiency (NSE) of 0.82 to 0.67 for 1‐ to 30‐day streamflow forecasts. Incorporating upstream streamflow information improved forecast accuracy by up to 10%. During real‐time forecast, FutureTST maintains higher forecast skills of 9.03 for 1‐day and 5.74 for 14‐day forecasts. In contrast, calibrated process‐based hydrological model forecasts become unreliable beyond a 4‐day lead time. Our findings demonstrate the potential of FutureTST as a reliable streamflow forecasting tool that offers a valuable addition to operational flood monitoring systems and climate‐resilient decision‐making. |
| format | Article |
| id | doaj-art-c005b0fd30ea4d16968a579c875fb591 |
| institution | DOAJ |
| issn | 0094-8276 1944-8007 |
| language | English |
| publishDate | 2025-07-01 |
| publisher | Wiley |
| record_format | Article |
| series | Geophysical Research Letters |
| spelling | doaj-art-c005b0fd30ea4d16968a579c875fb5912025-08-20T02:46:20ZengWileyGeophysical Research Letters0094-82761944-80072025-07-015214n/an/a10.1029/2025GL116707Novel Deep Learning Transformer Model for Short to Sub‐Seasonal Streamflow ForecastAnukesh Krishnankutty Ambika0Kshitij Tayal1Vimal Mishra2Dan Lu3Computational Sciences and Engineering Division Oak Ridge National Laboratory Oak Ridge TN USAComputational Sciences and Engineering Division Oak Ridge National Laboratory Oak Ridge TN USAEarth Science Indian Institute of Technology Gandhinagar Gujarat IndiaComputational Sciences and Engineering Division Oak Ridge National Laboratory Oak Ridge TN USAAbstract Accurate short‐to‐subseasonal streamflow forecasts are becoming crucial for effective water management in an increasingly variable climate. However, streamflow forecast remains challenging over extended lead times, uncertainty in meteorological inputs, and increased frequency and variability in extreme weather and climate events. We implemented a Future Time Series Transformer (FutureTST) model for streamflow forecasting that separately integrates past meteorological and streamflow data while incorporating future weather conditions. FutureTST achieves a mean Nash‐Sutcliffe Efficiency (NSE) of 0.82 to 0.67 for 1‐ to 30‐day streamflow forecasts. Incorporating upstream streamflow information improved forecast accuracy by up to 10%. During real‐time forecast, FutureTST maintains higher forecast skills of 9.03 for 1‐day and 5.74 for 14‐day forecasts. In contrast, calibrated process‐based hydrological model forecasts become unreliable beyond a 4‐day lead time. Our findings demonstrate the potential of FutureTST as a reliable streamflow forecasting tool that offers a valuable addition to operational flood monitoring systems and climate‐resilient decision‐making.https://doi.org/10.1029/2025GL116707 |
| spellingShingle | Anukesh Krishnankutty Ambika Kshitij Tayal Vimal Mishra Dan Lu Novel Deep Learning Transformer Model for Short to Sub‐Seasonal Streamflow Forecast Geophysical Research Letters |
| title | Novel Deep Learning Transformer Model for Short to Sub‐Seasonal Streamflow Forecast |
| title_full | Novel Deep Learning Transformer Model for Short to Sub‐Seasonal Streamflow Forecast |
| title_fullStr | Novel Deep Learning Transformer Model for Short to Sub‐Seasonal Streamflow Forecast |
| title_full_unstemmed | Novel Deep Learning Transformer Model for Short to Sub‐Seasonal Streamflow Forecast |
| title_short | Novel Deep Learning Transformer Model for Short to Sub‐Seasonal Streamflow Forecast |
| title_sort | novel deep learning transformer model for short to sub seasonal streamflow forecast |
| url | https://doi.org/10.1029/2025GL116707 |
| work_keys_str_mv | AT anukeshkrishnankuttyambika noveldeeplearningtransformermodelforshorttosubseasonalstreamflowforecast AT kshitijtayal noveldeeplearningtransformermodelforshorttosubseasonalstreamflowforecast AT vimalmishra noveldeeplearningtransformermodelforshorttosubseasonalstreamflowforecast AT danlu noveldeeplearningtransformermodelforshorttosubseasonalstreamflowforecast |