Premium
Nonstationarity driven by multidecadal change in catchment groundwater storage: A test of modifications to a common rainfall–run‐off model
Author(s) -
Grigg Andrew H.,
Hughes Justin D.
Publication year - 2018
Publication title -
hydrological processes
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.222
H-Index - 161
eISSN - 1099-1085
pISSN - 0885-6087
DOI - 10.1002/hyp.13282
Subject(s) - streamflow , environmental science , drainage basin , hydrology (agriculture) , groundwater , catchment hydrology , climate change , water storage , hydrological modelling , vegetation (pathology) , climatology , geology , geography , medicine , oceanography , cartography , geotechnical engineering , pathology , geomorphology , inlet
Abstract Many parts of the world are experiencing a drying climatic trend with significant implications for water resources, especially where there is interaction between surface water and groundwater. Where drying leads to a shift in a catchment's hydrological regime, hydrological models may display bias in predictions and therefore degradation in their predictive capacity. We present two modifications to GR4J, a commonly used rainfall–run‐off model, that (a) allow the sensitivity of storage dependant evaporation and streamflow production to differ significantly, effectively mimicking “catchment memory” and (b) account for temporal variation in catchment vegetation cover, for example, due to land use change. We test the original and two modified models on a forested headwater catchment in the south‐west of Australia. In this catchment, a data set of more than 21,000 individual groundwater records collected over a period of more than 40 years chronicles a significant decline in catchment groundwater storage, resulting in a nonstationary streamflow regime. These longer‐term changes were overlaid with a shorter‐term hydrological response to clearing for bauxite mining and subsequent revegetation. The model structural modifications led to substantially improved streamflow prediction when compared with the original model, related to improved ability in the revised models' production store to match changes in the observed groundwater storage within the catchment. The performance of all three models and hence the apparent effectiveness of the modified models was dependent on the hydrological conditions in the calibration period. Best model fits were obtained when the calibration period encompassed both wetter and drier conditions in the hydrological record. The modified models presented here may reduce prediction bias for catchments in other parts of the world where long‐term shifts in hydrologic regime are being experienced and improve prediction for catchments with significant changes in vegetation cover.