Loading...
Ensemble Streamflow Prediction
The ESP method uses the GR6J model run with historic climate data.
River flow forecasts using rainfall forecasts
This method uses the GR6J model run with historic weather analogues.
chance will be
Low
chance will be
Normal
chance will be
High
The graphics above demonstrate the proportions of ensemble members within each of the five Outlook categories. Note: the map and the info boxes above combine the "below normal" with "low" and "above normal" with "high" to create three categories. The ensembles are presented in two ways:
- as the percentage of ensemble members within each category
- as a relative number, indicating how many more, or less ensemble members are in each outlook category compared to the number we would expect
This distinction is important, as the expected probabilities of the five categories are not of equal size. You can adjust the colouring of the points on the map to represent either of these options.
This boxplot shows the distribution of the accumulated flows predicted by the ESP ensemble members. The forecast for the first month is a one-month average forecast; the forecast for the second month is a two-month average forecast, and so on. The last three months of the simulated observations are shown as a line graph at the start of the timeseries. The background colours show the bandings of the historic flows for reference; these also represent accumulated flow bandings, with the exception of the first three "status" months on the plot.
Continuous ranked probability skill scores (CRPSS) are shown at the bottom of this graph. This skill score measures the skill of the ESP method over the hindcast period, and compares it with a simple climatology forecast (a simple forecast based just on the distribution of historic flows). A skill score of 1 represents a perfect forecast, a score of 0 indicates the forecast is no better than a climatology forecast. It is worth noting that when there is low skill, the ESP forecast defaults to a climatology forecast, so despite the low skill, it can still be used as a basic forecast. These scores are calculated on the accumulated forecast (e.g. the score for 6 months is for the flow forecast for the 6-month averaged flows).
Years with the lowest daily flow:
Thresholds:
The spaghetti plot shows daily flows from each ensemble member in more detail. This plot simulates what would happen in the river in question if we received the weather conditions of each of those years today. You can choose to highlight five years for reference (from 1962–present). The default highlighted years are the five historic years with the lowest average flow from the month preceding the start of the forecast. The dashed lines set three thresholds of flow, which you can adjust (according to particular flow thresholds of interest to you, e.g. percentiles relevant to management decisions). The default thresholds are the flows exceeded 80, 90 and 99 percent of the time in the simulated observational record. The stacked plot underneath then represents the percentage of ensemble members that fall below each threshold.
Important note: while this plotting functionality allows the user to examine the probability of reaching particular flows within the forecast horizon, it is important to take account of the uncertainty in modelling which means the model outputs can be biased relative to observations.
Historical Flow Analogues and Persistence
This page shows forecasts produced using two statistical approaches based on historic river flow data.
In each of the time series graphs the bold black line represents the observed flow during the past months. The grey band indicates the normal flow range (the normal band includes 44% of observed flows in each month). The selected historical analogues are shown as thin lines and the trajectories that flows took in the following are also shown. The year each analogue started in is shown in the legend. The forecast is shown as the dashed red line, and in each plot it states whether this has come from the analogues or has been generated on the basis of persistence.
Monthly mean simulated river flows
Simulated for
This map shows the simulated monthly mean flow ranked in terms of simulated historical flow estimates.
These flows are produced by the 1km resolution Grid-to-Grid (G2G) hydrological model (Bell et al., 2009), which is run up to the end of each calendar month using observed rainfall (Met Office NCIC) and MORECS (Hough and Jones, 1997) potential evaporation as input. Flows are ranked relative to monthly mean flows simulated by the same model over that month in the period 1963–2016.
Note that the G2G model provides estimates of natural flows.
Bell V A, Kay A L, Jones R G et al. 2009. Use of soil data in a grid-based hydrological model to estimate spatial variation in changing flood risk across the UK. J. Hydrol., 377(3–4), 335–350.
Hough M, Jones R J A. 1997. The United Kingdom Meteorological Office rainfall and evaporation calculation system: MORECS version 2.0 — an overview. Hydrol. Earth Syst. Sci., 1(2), 227–239.
Current simulated subsurface water storage
Simulated values for
This map shows the estimated total subsurface water storage relative to simulated historical extremes.
These maps are based on Grid-to-Grid (G2G) hydrological model simulated subsurface water storage (water in the soil and groundwater) for the last day of the previous month, expressed as an anomaly from the historical (1963–2016) monthly mean. This anomaly is then scaled relative to the maximum or minimum subsurface storage anomaly simulated for that month over the period 1963–2016.
Rainfall in wet areas (with high positive subsurface wetness) could result in flooding in the coming days/weeks. Areas of negative relative wetness indicate locations which are particularly dry, and little or no rain in these areas could potentially lead to (or prolong) a drought.
Current simulated soil water storage
Simulated values for
This map shows the estimated soil water storage anomaly relative to simulated historical extremes.
These maps are based on Grid-to-Grid (G2G) hydrological model simulated soil water storage simulated for the last day of the previous month, expressed as an anomaly from the historical (1963–2016) monthly mean. This anomaly is then scaled relative to the maximum or minimum soil water storage anomaly simulated for that month over the period 1963–2016.
Soil wetness will often look similar to subsurface wetness (shown on another page), since subsurface storage comprises both soil storage and storage in the saturated zone.
Daily soil moisture conditions at higher spatial resolution are available from the UKCEH soil wetness explorer.
Current simulated subsurface storage anomaly
Simulated values for
This map shows the regional means of the simulated subsurface storage anomalies.
Each month the 1km gridded Grid-to-Grid (G2G) hydrological model is run with observed inputs of precipitation and potential evaporation. These storage deficit maps show the water storage relative to the historical monthly mean (1981–2010).
Negative anomalies (red/pink) represent subsurface storage deficits, while positive anomalies (blue) represent a subsurface water surplus.
Subsurface water storage deficits (mm) can be interpreted as an estimate of the additional rainfall that would be required in future months to overcome dry conditions (i.e. rainfall in addition to what is expected on average).
Will average rainfall overcome any dry conditions?
Estimated based on simulated storage deficits on
These maps show the return period of rainfall required to replenish any current subsurface water deficits.
Note: for most months these maps are white, indicating that there is either no subsurface water storage deficit or unexceptional rainfall will replenish the deficit by the specified month.
These maps show the return period of the rainfall required to overcome the estimated current subsurface water storage deficits estimated by the G2G hydrological model. Colours indicate the return period of accumulated rainfall required to replenish stores over the next one to six months.
These maps do not provide a drought forecast; only whether particularly heavy rainfall would be required to return to normal conditions for the time of year.
These maps are not currently available at 1km resolution.
River flow forecasts using rainfall forecasts
This page shows the distribution of river flow forecasts estimated by the water balance model. Click on any coloured river pixel to show the flow distribution for that pixel.
This page shows the distribution of regional mean river flow forecasts estimated by the water balance model. Click on any coloured region to show the flow distribution for that region on the bar charts.
These monthly-mean river flow forecasts are produced using the hydrological initial conditions from the Grid-to-Grid model and a c.400 member ensemble of rainfall forecasts from the Met Office as inputs to a monthly-resolution water balance hydrological model.
The river flow forecasts are ranked in terms of 54 years of historical flow estimates (1963–2016). The colours on the map indicate each river pixel's ensemble median forecast river flow for the next month.
The river flow forecasts are ranked in terms of 54 years of historical flow estimates (1963-2016). The colours on the map indicate each region's ensemble median forecast river flow for the next month.