A multi-objective ensemble approach to hydrological modelling in the UK: an application to historic drought reconstruction

摘要: Abstract. Hydrological models can provide estimates of streamflow pre- and post-observations, which enable greater understanding past hydrological behaviour, and potential futures. In this paper, a new multi-objective calibration method was derived tested for 303 catchments in the UK, and the calibrations were used to reconstruct river flows back 1891, order to much longer view hydrological variability, given the brevity most UK flow records began post-1960. A Latin hypercube sample 500 000 parameterisations GR4J model each catchment evaluated against six evaluation metrics covering all aspects of regime from high, median, low flows. The results top ranking parameterisation (LHS1), also top 500 (LHS500), for each catchment deterministic result whilst also accounting parameter uncertainty. are generally good at capturing observed flows, with some exceptions heavily groundwater-dominated catchments, snowmelt artificially influenced catchments across country. Reconstructed appraised over 30-year moving windows shown simulations early parts record, cases where observations available. To consider the utility reconstructions drought simulation, data the 1975–1976 event explored detail nine case study catchments. The model's performance reproducing events found vary by catchment, as did level uncertainty LHS500. The Standardised Streamflow Index (SSI) assess model simulations' ability simulate extreme events. peaks troughs of the SSI time series well represented despite slight over- or underestimations magnitudes, while accumulated deficits extracted verified that overall very at simulating drought events. This paper provides three key contributions: (1) a robust multi-objective calibration framework calibrating models for use both general hydrology; (2) model the 303 catchments that could be further research, operational applications such forecasting; (3)  ∼ 125  years spatially temporally consistent reconstructed data that will allow comprehensive quantitative assessments drought events, long-term analyses variability have not been previously possible, thus enabling water resource managers to better plan build more resilient systems the future.