| Autor: |
Rios, Manuel, Pahlow, Markus, Srinivasan, M. S., Singh, Shailesh Kumar |
| Předmět: |
|
| Zdroj: |
Journal of Hydrology (00221708); 2023, Vol. 62 Issue 1, p19-34, 16p |
| Abstrakt: |
Groundwater recharge rates and volumes underpin water resource allocation decisions. Drainage lysimeters ('lysimeters') provide a direct, continuous measurement of drainage below the root zone. However, owing to their pointscale nature and limited depth, such measurements are seldom used to estimate potential groundwater recharge and subsequent allocation decisions. A combination of Artificial Neural Network (ANN) and uncertainty estimation techniques were applied to upscale lysimeter-based drainage measurements to the regional scale to estimate unconfined aquifer groundwater recharge. We trained ANN models using drainage data from three lysimeter sites across the Canterbury plains of New Zealand. The best predictive model, in terms of accuracy and parsimony, provided coefficient of determination (R2) values of 0.65, 0.74 and 0.86 at the three lysimeter sites, with a relative model uncertainty of 6%. The model was then implemented within a geographic information system (GIS) to predict the spatial variability of land surface recharge to support groundwater allocation at the regional scale. The GISintegrated ANN model provides a computationally inexpensive method of upscaling pointscale drainage data into regionalscale recharge estimates. In a next step these estimates can be refined by implementing additional data, such as spatially distributed soil moisture measurements, into the ANN. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
