Development of a high resolution grid-based river flow model for use with regional climate model output

Abstract

A grid-based approach to river flow modelling has been developed for regional assessments of the impact of environmental change on hydrologically sensitive systems. The approach also provides a means of assessing, and providing feedback on, the hydrological performance of the land-surface component of a regional climate model (RCM). When combined with information on the evolution of climate, the model can give estimates of the impact of future climate change on river flows and flooding. The high-resolution flow routing and runoff-production model is designed for use with RCM-derived rainfall and potential evaporation (PE), although other sources of gridded rainfall and PE can be employed. Called the Grid-to-Grid Model, or G2G, it can be configured on grids of different resolution and coverage (a 1 km grid over the UK is used here). The model can simulate flow on an area-wide basis as well as providing estimates of fluvial discharges for input to shelfsea and ocean models. Configuration of the flow routing model on a relatively high resolution 1 km grid allows modelled river flows to be compared with gauged observations for a variety of catchments across the UK. Modelled flows are also compared with those obtained from a catchment-based model, a parameter-generalised form of the Probability-Distributed Model (PDM) developed for assessing flood frequency. Using RCM re-analysis rainfall and PE as input, the G2G model performs well compared with measured flows at a daily time-step, particularly for high relief catchments. It performs less well for low-relief and groundwater-dominated regions because the dominant model control on runoff production is topography.

How spatially coherent and statistically robust are temporal changes in extreme precipitation in the contiguous USA?

“…Probability distributions computed by fitting data from these three windows to the mixed-exponential distribution also indicate greatest divergence for large events that have probabilities ≤3 × 10 −3 (and at these sites magnitudes in excess of approximately 40 mm; see Figure 8(a) and (c)) with consistently higher accumulation amounts for these extreme events in the later time segments. While these intense events may be subject to large measurement uncertainties (Groisman and Legates, 1994), they are of critical importance in dictating flood frequency and changes therein (Bell et al, 2007) and hence are of particular interest. The implied extension of the upper tail of the probability distribution of precipitation accumulations is consistent with results of other analyses presented herein and prior research that shows extreme precipitation events of 1-7 day duration increased by approximately 3% per decade between 1931 and 1996 over much of the upper Midwest (Kunkel et al, 1999).…”

Use of a grid‐based hydrological model and regional climate model outputs to assess changing flood risk

“…The G2G (Bell et al, 2007) has been configured to represent spatial variability in catchment response and to make full use of the gridded spatially-distributed precipitation data obtained from climate models or observations. The model can be configured for use at (almost) any spatial resolution, with the temporal resolution determined by stability criteria.…”

“…The Grid-to-Grid model A grid-based approach to river flow modelling (Bell et al, 2007) has been developed for use in regional assessments of the impact of environmental change on hydrologically sensitive systems. The overall procedure provides a grid-based methodology that uses a grid-togrid hydrological model to translate RCM-derived meteorological variables, such as precipitation and potential 1659 evaporation, into estimates of river flow and fluvial discharges to the sea.…”

“…The model has been assessed using raingauge rainfall data and the ERA-driven RCM rainfall data (1979 to 1993) used by Kay et al (2006a). This assessment, presented in Bell et al (2007), indicates that when the G2G is used as an area-wide model, with one set of model parameters for all catchments, it performs best on catchments where the hydrological response is dominated by topography rather than soil/geology effects on baseflow. For catchments where some calibration was undertaken, the G2G model performance compares reasonably well to an alternative catchment-based model of lumped form, the parameter-generalized PDM (see Section 2.2), when flows are assessed at a daily timestep.…”

“…The Grid-to-Grid hydrological model used here (Bell et al, 2007) couples a simple grid-based runoff-production scheme with a grid-based flow routing component to produce river flow on a regular grid. The routing scheme is also designed so that it can be linked to a land-surface model, such as the Met Office's MOSES scheme (Cox et al, 1999), thereby providing a direct link between the atmosphere, land surface, river flow and (potentially) shelf-sea models.…”