Meteorological Ensemble Forecasts Research Articles

Improving the accuracy of multimodel short-to-medium-range precipitation and streamflow forecasts over the Upper Bhima river basin, India

Accurate precipitation forecasting with sufficient lead time is a prerequisite for developing a robust flood warning system (FWS), which is very challenging, particularly in developing countries like India. This study evaluates the utility of the TIGGE multimodel ensemble meteorological forecasts over the Upper Bhima river basin and investigated the hydrological utility of the TIGGE forecasts through a calibrated hydrological (VIC-RAPID) model followed by the postprocessing of streamflow through Bayesian model average (BMA) approach. Results show that the quality of the meteorological forecasts of precipitation, and simulated streamflow deteriorated with increasing lead time, which can be ameliorated with a suitable bias-correction technique. The BMA-based postprocessing further improved the streamflow simulations, especially in case of extreme events, which highlighted its efficacy in flood forecasting. From the results of the study, it is recommended that a compound system of improved precipitation prediction, calibrated VIC-RAPID model and postprocessing of streamflows in an integrated manner would facilitate a reliable FWS for operational purposes.

Subseasonal to seasonal streamflow forecasting in a semiarid watershed

AbstractOperational streamflow forecasting is critically important to managers of river basins that supply water, hydropower, and flood protection. While seasonal water supply forecasts (WSFs) are important for long‐term water resources planning operations, shorter term (e.g., 1–5 weeks) streamflow forecasts are critical for balancing water conservation with flood risk during wet periods. In this study, we designed a streamflow forecasting system with the water resources group at the Salt River Project (SRP), a provider of water and power to millions of customers in central Arizona (AZ), to provide streamflow forecasts for a diverse and operationally important set of watersheds in AZ. The forecast system uses machine learning to make seasonal WSFs, a rainfall–runoff model driven by ensemble meteorological forecasts to make 35‐day streamflow forecasts, and an innovative approach to improve the WSFs based on the 35‐day streamflow forecasts. This model integration allows for an assessment of the impact of different meteorological forecasts on WSFs, helping SRP to balance water conservation goals with shorter term flood risks. In addition, seasonal WSFs are improved in the early winter when they incorporate the 35‐day streamflow predictions. Furthermore, these improvements are larger than when they incorporate 7‐day streamflow predictions, demonstrating the value of using subseasonal to seasonal (S2S, >1–2 weeks) forecasts to improve seasonal WSFs in these watersheds.

A Multi-Temporal-Scale Modulation Mechanism for the Postprocessing of Precipitation Ensemble Forecasts: Benefits for Streamflow Forecasting

Abstract In the postprocessing of ensemble forecasts of weather variables, it is standard practice to first calibrate the forecasts in a univariate setting, before reconstructing multivariate ensembles that have a correct covariability in space, time, and across variables, via so-called “reordering” methods. Within this framework though, postprocessors cannot fully extract the skill of the raw forecast that may exist at larger scales. A multi-temporal-scale modulation mechanism for precipitation is here presented, which aims at improving the forecasts over different accumulation periods, and which can be coupled with any univariate calibration and multivariate reordering techniques. The idea, originally known under the term “canonical events,” has been implemented for more than a decade in the Meteorological Ensemble Forecast Processor (MEFP), a component of the U.S. National Weather Service’s (NWS) Hydrologic Ensemble Forecast Service (HEFS), although users were left with material in the gray literature. This paper proposes a formal description of the mechanism and studies its intrinsic connection with the multivariate reordering process. The verification of modulated and unmodulated forecasts, when coupled with two popular methods for reordering, the Schaake shuffle and ensemble copula coupling (ECC), is performed on 11 Californian basins, on both precipitation and streamflow. Results demonstrate the clear benefit of the multi-temporal-scale modulation, in particular on multiday total streamflow. However, the relative gain depends on the method used for reordering, with more benefits expected when this latter method is not able to reconstruct an adequate temporal structure on the calibrated precipitation forecasts.

An ensemble data assimilation modeling system for operational outdoor microalgae growth forecasting.

Microalgae have received increasing attention as a potential feedstock for biofuel or biobased products. Forecasting the microalgae growth is beneficial for managers in planning pond operations and harvesting decisions. This study proposed a biomass forecasting system comprised of the Huesemann Algae Biomass Growth Model (BGM), the Modular Aquatic Simulation System in Two Dimensions (MASS2), ensemble data assimilation (DA), and numerical weather prediction Global Ensemble Forecast System (GEFS) ensemble meteorological forecasts. The novelty of this study is to seek the use of ensemble DA to improve both BGM and MASS2 model initial conditions with the assimilation of biomass and water temperature measurements and consequently improve short-term biomass forecasting skills. This study introduces the theory behind the proposed integrated biomass forecasting system, with an application undertaken in pseudo-real-time in three outdoor ponds cultured with Chlorella sorokiniana in Delhi, California, United States. Results from all three case studies demonstrate that the biomass forecasting system improved the short-term (i.e., 7-day) biomass forecasting skills by about 60% on average, comparing to forecasts without using the ensemble DA method. Given the satisfactory performances achieved in this study, it is probable that the integrated BGM-MASS2-DA forecasting system can be used operationally to inform managers in making pond operation and harvesting planning decisions.

Operational hydrometeorological forecasting on the Rhône River in France: moving toward a seamless probabilistic approach

Compagnie Nationale du Rhône (CNR) has operated the Rhône River since 1934 according to three core missions – hydropower generation, inland navigation and irrigation – using 19 run of-river multi-purpose hydropower schemes. To ensure hydraulic safety and optimise hydro power production and commercialisation, CNR developed a range of hydrometeorological forecasting tools, currently run every day in real time by CNR forecasters. The first is an hourly deterministic tool, giving forecasts up to the next 4 days. This tool does not allow quantification of uncertainties, which is a key issue to improve forecasts and take better operational decisions. At the same time, national meteorological centres provide ensemble meteorological forecasts, and thus information about uncertainty. Therefore, CNR is developing a chain of ensemble forecasting tools based on ensemble meteorological forecasts and various post-processing methods. These tools provide hourly/daily hydrological forecasts up to the next 4/12 days. The operational use of such a probabilistic forecasting chain is intended to facilitate decision-making processes. All these tools have been designed to leave large room for human expertise. This paper presents the tools described above, but also their interaction with human forecasters as well as the first attempts to link them together to add consistency between the different forecast types.

Calibrated ensemble forecasts of the height of new snow using quantile regression forests and ensemble model output statistics

Abstract. Height of new snow (HN) forecasts help to prevent critical failures of infrastructures in mountain areas, e.g. transport networks and ski resorts. The French national meteorological service, Météo-France, operates a probabilistic forecasting system based on ensemble meteorological forecasts and a detailed snowpack model to provide ensembles of HN forecasts. These forecasts are, however, biased and underdispersed. As for many weather variables, post-processing methods can be used to alleviate these drawbacks and obtain meaningful 1 to 4 d HN forecasts. In this paper, we compare the skill of two post-processing methods. The first approach is an ensemble model output statistics (EMOS) method, which can be described as a nonhomogeneous regression with a censored shifted Gamma distribution. The second approach is based on quantile regression forests, using different meteorological and snow predictors. Both approaches are evaluated using a 22 year reforecast. Thanks to a larger number of predictors, the quantile regression forest is shown to be a powerful alternative to EMOS for the post-processing of HN ensemble forecasts. The gain of performance is large in all situations but is particularly marked when raw forecasts completely miss the snow event. This type of situation happens when the rain–snow transition elevation is overestimated by the raw forecasts (rain instead of snow in the raw forecasts) or when there is no precipitation in the forecast. In that case, quantile regression forests improve the predictions using the other weather predictors (wind, temperature, and specific humidity).

On the number of bins in a rank histogram

AbstractRank histograms are popular tools for assessing the reliability of meteorological ensemble forecast systems. A reliable forecast system leads to a uniform rank histogram, and deviations from uniformity can indicate miscalibrations. However, the ability to identify such deviations by visual inspection of rank histogram plots crucially depends on the number of bins chosen for the histogram. If too few bins are chosen, the rank histogram is likely to miss miscalibrations; if too many are chosen, even perfectly calibrated forecast systems can yield rank histograms that do not appear uniform. In this paper we address this trade‐off and propose a method for choosing the number of bins for a rank histogram. The goal of our method is to select a number of bins such that the intuitive decision whether a histogram is uniform or not is as close as possible to a formal statistical test. Our results indicate that it is often appropriate to choose fewer bins than the usual choice of ensemble size plus one, especially when the number of observations available for verification is small.

Ensemble hydrological forecasts for reservoir management of the Shipshaw River catchment using limited data

Many hydropower companies continue to rely on expert judgment to manage the operations of their reservoirs. Decision-support systems, composed of a hydrological forecasting system and a reservoir model, can ensure that reservoir operation objectives are attained more effectively than by relying solely on expert judgment. In this study, a simple ensemble inflow forecasting system coupled with a reservoir model is developed and the proposed model-based operational water management decisions are compared with those based on expert judgment for the Shipshaw River in Quebec, Canada. Given that no natural streamflow records are available for the Shipshaw River, the HEC-HMS hydrological model is calibrated using a regionalization method based on physical similarity. The calibrated hydrological model is fed by ensemble meteorological forecasts that include 20 members, with a 10-day horizon and a 6-hour time step. The proposed decision-support system can help avoid small flooding events while potentially improving energy production by 2 to 60% for this case study. The proposed forecasting system also allows water-resource managers to anticipate events with a greater lead time.

Statistical Postprocessing of Water Level Forecasts Using Bayesian Model Averaging With Doubly Truncated Normal Components

AbstractAccurate and reliable probabilistic forecasts of hydrological quantities like runoff or water level are beneficial to various areas of society. Probabilistic state‐of‐the‐art hydrological ensemble prediction models are usually driven with meteorological ensemble forecasts. Hence, biases and dispersion errors of the meteorological forecasts cascade down to the hydrological predictions and add to the errors of the hydrological models. The systematic parts of these errors can be reduced by applying statistical postprocessing. For a sound estimation of predictive uncertainty and an optimal correction of systematic errors, statistical postprocessing methods should be tailored to the particular forecast variable at hand. Former studies have shown that it can make sense to treat hydrological quantities as bounded variables. In this paper, a doubly truncated Bayesian model averaging (BMA) method, which allows for flexible postprocessing of possibly multimodel ensemble forecasts of water level, is introduced. A case study based on water levels for a gauge of river Rhine reveals a good predictive skill of doubly truncated BMA compared both to the raw ensemble and the reference ensemble model output statistics approach. Using rolling training periods, BMA considerably outerperforms ensemble model output statistics. However, this gap shrinks drastically when using analog‐based training periods.

Characterizing the Atmospheric Conditions Leading to Large Error Growth in Volcanic Ash Cloud Forecasts

ABSTRACTVolcanic ash poses an ongoing risk to safety in the airspace worldwide. The accuracy with which volcanic ash dispersion can be forecast depends on the conditions of the atmosphere into which it is emitted. In this study, meteorological ensemble forecasts are used to drive a volcanic ash transport and dispersion model for the 2010 Eyjafjallajökull eruption in Iceland. From analysis of these simulations, the authors determine why the skill of deterministic-meteorological forecasts decreases with increasing ash residence time and identify the atmospheric conditions in which this drop in skill occurs most rapidly. Large forecast errors are more likely when ash particles encounter regions of large horizontal flow separation in the atmosphere. Nearby ash particle trajectories can rapidly diverge, leading to a reduction in the forecast accuracy of deterministic forecasts that do not represent variability in wind fields at the synoptic scale. The flow‐separation diagnostic identifies where and why large ensemble spread may occur. This diagnostic can be used to alert forecasters to situations in which the ensemble mean is not representative of the individual ensemble‐member volcanic ash distributions. Knowledge of potential ensemble outliers can be used to assess confidence in the forecast and to avoid potentially dangerous situations in which forecasts fail to predict harmful levels of volcanic ash.

Verification of ECMWF System 4 for seasonal hydrological forecasting in a northern climate

Abstract. Hydropower production requires optimal dam and reservoir management to prevent flooding damage and avoid operation losses. In a northern climate, where spring freshet constitutes the main inflow volume, seasonal forecasts can help to establish a yearly strategy. Long-term hydrological forecasts often rely on past observations of streamflow or meteorological data. Another alternative is to use ensemble meteorological forecasts produced by climate models. In this paper, those produced by the ECMWF (European Centre for Medium-Range Forecast) System 4 are examined and bias is characterized. Bias correction, through the linear scaling method, improves the performance of the raw ensemble meteorological forecasts in terms of continuous ranked probability score (CRPS). Then, three seasonal ensemble hydrological forecasting systems are compared: (1) the climatology of simulated streamflow, (2) the ensemble hydrological forecasts based on climatology (ESP) and (3) the hydrological forecasts based on bias-corrected ensemble meteorological forecasts from System 4 (corr-DSP). Simulated streamflow computed using observed meteorological data is used as benchmark. Accounting for initial conditions is valuable even for long-term forecasts. ESP and corr-DSP both outperform the climatology of simulated streamflow for lead times from 1 to 5 months depending on the season and watershed. Integrating information about future meteorological conditions also improves monthly volume forecasts. For the 1-month lead time, a gain exists for almost all watersheds during winter, summer and fall. However, volume forecasts performance for spring varies from one watershed to another. For most of them, the performance is close to the performance of ESP. For longer lead times, the CRPS skill score is mostly in favour of ESP, even if for many watersheds, ESP and corr-DSP have comparable skill. Corr-DSP appears quite reliable but, in some cases, under-dispersion or bias is observed. A more complex bias-correction method should be further investigated to remedy this weakness and take more advantage of the ensemble forecasts produced by the climate model. Overall, in this study, bias-corrected ensemble meteorological forecasts appear to be an interesting source of information for hydrological forecasting for lead times up to 1 month. They could also complement ESP for longer lead times.

Assimilation of water temperature and discharge data for ensemble water temperature forecasting

Recent work demonstrated the value of water temperature forecasts to improve water resources allocation and highlighted the importance of quantifying their uncertainty adequately. In this study, we perform a multisite cascading ensemble assimilation of discharge and water temperature on the Nechako River (Canada) using particle filters. Hydrological and thermal initial conditions were provided to a rainfall-runoff model, coupled to a thermal module, using ensemble meteorological forecasts as inputs to produce 5day ensemble thermal forecasts. Results show good performances of the particle filters with improvements of the accuracy of initial conditions by more than 65% compared to simulations without data assimilation for both the hydrological and the thermal component. All thermal forecasts returned continuous ranked probability scores under 0.8°C when using a set of 40 initial conditions and meteorological forecasts comprising 20 members. A greater contribution of the initial conditions to the total uncertainty of the system for 1-dayforecasts is observed (mean ensemble spread=1.1°C) compared to meteorological forcings (mean ensemble spread=0.6°C). The inclusion of meteorological uncertainty is critical to maintain reliable forecasts and proper ensemble spread for lead times of 2days and more. This work demonstrates the ability of the particle filters to properly update the initial conditions of a coupled hydrological and thermal model and offers insights regarding the contribution of two major sources of uncertainty to the overall uncertainty in thermal forecasts.

Moving beyond the cost–loss ratio: economic assessment of streamflow forecasts for a risk-averse decision maker

Abstract. A large effort has been made over the past 10 years to promote the operational use of probabilistic or ensemble streamflow forecasts. Numerous studies have shown that ensemble forecasts are of higher quality than deterministic ones. Many studies also conclude that decisions based on ensemble rather than deterministic forecasts lead to better decisions in the context of flood mitigation. Hence, it is believed that ensemble forecasts possess a greater economic and social value for both decision makers and the general population. However, the vast majority of, if not all, existing hydro-economic studies rely on a cost–loss ratio framework that assumes a risk-neutral decision maker. To overcome this important flaw, this study borrows from economics and evaluates the economic value of early warning flood systems using the well-known Constant Absolute Risk Aversion (CARA) utility function, which explicitly accounts for the level of risk aversion of the decision maker. This new framework allows for the full exploitation of the information related to a forecasts' uncertainty, making it especially suited for the economic assessment of ensemble or probabilistic forecasts. Rather than comparing deterministic and ensemble forecasts, this study focuses on comparing different types of ensemble forecasts. There are multiple ways of assessing and representing forecast uncertainty. Consequently, there exist many different means of building an ensemble forecasting system for future streamflow. One such possibility is to dress deterministic forecasts using the statistics of past error forecasts. Such dressing methods are popular among operational agencies because of their simplicity and intuitiveness. Another approach is the use of ensemble meteorological forecasts for precipitation and temperature, which are then provided as inputs to one or many hydrological model(s). In this study, three concurrent ensemble streamflow forecasting systems are compared: simple statistically dressed deterministic forecasts, forecasts based on meteorological ensembles, and a variant of the latter that also includes an estimation of state variable uncertainty. This comparison takes place for the Montmorency River, a small flood-prone watershed in southern central Quebec, Canada. The assessment of forecasts is performed for lead times of 1 to 5 days, both in terms of forecasts' quality (relative to the corresponding record of observations) and in terms of economic value, using the new proposed framework based on the CARA utility function. It is found that the economic value of a forecast for a risk-averse decision maker is closely linked to the forecast reliability in predicting the upper tail of the streamflow distribution. Hence, post-processing forecasts to avoid over-forecasting could help improve both the quality and the value of forecasts.

Assessment and weighting of meteorological ensemble forecast members based on supervised machine learning with application to runoff simulations and flood warning

Numerical weather forecasts, such as meteorological forecasts of precipitation, are inherently uncertain. These uncertainties depend on model physics as well as initial and boundary conditions. Since precipitation forecasts form the input into hydrological models, the uncertainties of the precipitation forecasts result in uncertainties of flood forecasts. In order to consider these uncertainties, ensemble prediction systems are applied. These systems consist of several members simulated by different models or using a single model under varying initial and boundary conditions. However, a too wide uncertainty range obtained as a result of taking into account members with poor prediction skills may lead to underestimation or exaggeration of the risk of hazardous events. Therefore, the uncertainty range of model-based flood forecasts derived from the meteorological ensembles has to be restricted.In this paper, a methodology towards improving flood forecasts by weighting ensemble members according to their skills is presented. The skill of each ensemble member is evaluated by comparing the results of forecasts corresponding to this member with observed values in the past. Since numerous forecasts are required in order to reliably evaluate the skill, the evaluation procedure is time-consuming and tedious. Moreover, the evaluation is highly subjective, because an expert who performs it makes his decision based on his implicit knowledge.Therefore, approaches for the automated evaluation of such forecasts are required. Here, we present a semi-automated approach for the assessment of precipitation forecast ensemble members. The approach is based on supervised machine learning and was tested on ensemble precipitation forecasts for the area of the Mulde river basin in Germany. Based on the evaluation results of the specific ensemble members, weights corresponding to their forecast skill were calculated. These weights were then successfully used to reduce the uncertainties within rainfall-runoff simulations and flood risk predictions.

A Stratified Sampling Approach for Improved Sampling from a Calibrated Ensemble Forecast Distribution

Abstract Before using the Schaake shuffle or empirical copula coupling (ECC) to reconstruct the dependence structure for postprocessed ensemble meteorological forecasts, a necessary step is to sample discrete samples from each postprocessed continuous probability density function (pdf), which is the focus of this paper. In addition to the equidistance quantiles (EQ) and independent random (IR) sampling methods commonly used at present, the stratified sampling (SS) method is proposed. The performance of the three sampling methods is compared using calibrated GFS ensemble precipitation reforecasts over the Xixian basin in China. The ensemble reforecasts are first calibrated using heteroscedastic extended logistic regression (HELR), and then the three sampling methods are used to sample calibrated pdfs with a varying number of discrete samples. Finally, the effect of the sampling method on the reconstruction of ensemble members with preserved space dependence structure is analyzed by using EQ, IR, and SS in ECC for reconstructing postprocessed ensemble members for four stations in the Xixian basin. There are three main results. 1) The HELR model has a significant improvement over the raw ensemble forecast. It clearly improves the mean and dispersion of the predictive distribution. 2) Compared to EQ and IR, SS can better cover the tails of the calibrated pdfs and a better dispersion of calibrated ensemble forecasts is obtained. In terms of probabilistic verification metrics like the ranked probability skill score (RPSS), SS is slightly better than EQ and clearly better than IR, while in terms of the deterministic verification metric, root-mean-square error, EQ is slightly better than SS. 3) ECC-SS, ECC-EQ, and ECC-IR all calibrate the raw ensemble forecast, but ECC-SS shows a better dispersion than ECC-EQ and ECC-IR in this study.

Technical Note: Initial assessment of a multi-method approach to spring-flood forecasting in Sweden

Abstract. Hydropower is a major energy source in Sweden, and proper reservoir management prior to the spring-flood onset is crucial for optimal production. This requires accurate forecasts of the accumulated discharge in the spring-flood period (i.e. the spring-flood volume, SFV). Today's SFV forecasts are generated using a model-based climatological ensemble approach, where time series of precipitation and temperature from historical years are used to force a calibrated and initialized set-up of the HBV model. In this study, a number of new approaches to spring-flood forecasting that reflect the latest developments with respect to analysis and modelling on seasonal timescales are presented and evaluated. Three main approaches, represented by specific methods, are evaluated in SFV hindcasts for the Swedish river Vindelälven over a 10-year period with lead times between 0 and 4 months. In the first approach, historically analogue years with respect to the climate in the period preceding the spring flood are identified and used to compose a reduced ensemble. In the second, seasonal meteorological ensemble forecasts are used to drive the HBV model over the spring-flood period. In the third approach, statistical relationships between SFV and the large-sale atmospheric circulation are used to build forecast models. None of the new approaches consistently outperform the climatological ensemble approach, but for early forecasts improvements of up to 25 % are found. This potential is reasonably well realized in a multi-method system, which over all forecast dates reduced the error in SFV by ∼ 4 %. This improvement is limited but potentially significant for e.g. energy trading.

Probabilistic flood forecasting on the Rhone River: evaluation with ensemble and analogue-based precipitation forecasts

Hydrological ensemble forecasting performances are analysed over 5 basins up to 2000 km2 in the French Upper Rhone region. Streamflow forecasts are issued at an hourly time step from lumped ARX rainfall-runoff models forced by different precipitation forecasts. Ensemble meteorological forecasts from ECMWF and NCEP are considered, as well as analogue-based forecasts fed by their corresponding control forecast. Analogue forecasts are rearranged using an adaptation of the Schaake-Shuffle method in order to ensure the temporal coherence. A new evaluation approach is proposed, separating forecasting performances on peak amplitudes and peak timings for high flow events. Evaluation is conducted against both simulated and observed streamflow (so that relative meteorological and hydrological uncertainties can be assessed), by means of CRPS and rank histograms, over the 2007-2014 period. Results show a general agreement of the forecasting performances when averaged over the 5 basins. However, ensemble-based and analogue-based streamflow forecasts produce a different signature on peak events in terms of bias, spread and reliability. Strengths and weaknesses of both approaches are discussed as well as potential improvements, notably towards their merging.

Real-time drought forecasting system for irrigation management

Abstract. In recent years frequent periods of water scarcity have enhanced the need to use water more carefully, even in European areas which traditionally have an abundant supply of water, such as the Po Valley in northern Italy. In dry periods, water shortage problems can be enhanced by conflicting uses of water, such as irrigation, industry and power production (hydroelectric and thermoelectric). Furthermore, in the last decade the social perspective in relation to this issue has been increasing due to the possible impact of climate change and global warming scenarios which emerge from the IPCC Fifth Assessment Report (IPCC, 2013). Hence, the increased frequency of drought periods has stimulated the improvement of irrigation and water management. In this study we show the development and implementation of the PREGI real-time drought forecasting system; PREGI is an Italian acronym that means "hydro-meteorological forecast for irrigation management". The system, planned as a tool for irrigation optimization, is based on meteorological ensemble forecasts (20 members) at medium range (30 days) coupled with hydrological simulations of water balance to forecast the soil water content on a maize field in the Muzza Bassa Lodigiana (MBL) consortium in northern Italy. The hydrological model was validated against measurements of latent heat flux acquired by an eddy-covariance station, and soil moisture measured by TDR (time domain reflectivity) probes; the reliability of this forecasting system and its benefits were assessed in the 2012 growing season. The results obtained show how the proposed drought forecasting system is able to have a high reliability of forecast at least for 7–10 days ahead of time.

Improved Spring Peak-Flow Forecasting Using Ensemble Meteorological Predictions

The potential of ensemble meteorological forecasts in improving ensemble spring peak flow prediction up to 14 days ahead is investigated for the Saguenay-Lac-Saint Jean watershed located in northeastern Canada. Large-scale ensemble meteorological forecasts (precipitation and temperature) generated by the National Center for Environmental Prediction’s (NCEP) Global Forecast System (GFS) are bias corrected for two meteorological stations in the watershed. The bias corrected NCEP ensemble meteorological forecast data are used as input in the hydrological model Hydrologiska Byråns Vattenbalan-avdelning (HBV) to simulate ensemble reservoir inflows and Serpent River flows up to 14 days ahead. The ensemble inflow and flow forecasts are compared with climatology as well as with the case in which only observed historical data are used for spring peak flow forecasting. The study results show that there is a significant improvement for the longer forecast range in the model forecast performance when bias-corrected NCEP forecast data are used. The improvement for forecasts for the spring season as well as for the entire year is revealed by Brier and rank probability skill score (BSS and RPSS, respectively) for reservoir inflow and Serpent River flow forecasts. Visual inspection of scatter plots between observed and simulated flows, and hydrographs of ensemble mean and ensemble members also reveal the potential of NCEP meteorological forecasts for improving spring flow forecasting.

Verification of temperature, precipitation, and streamflow forecasts from the NOAA/NWS Hydrologic Ensemble Forecast Service (HEFS): 2. Streamflow verification

Summary Retrospective forecasts of precipitation, temperature, and streamflow were generated with the Hydrologic Ensemble Forecast Service (HEFS) of the U.S. National Weather Service (NWS) for a 20-year period between 1979 and 1999. The hindcasts were produced for two basins in each of four River Forecast Centers (RFCs), namely the Arkansas–Red Basin RFC, the Colorado Basin RFC, the California–Nevada RFC and the Middle Atlantic RFC. In a companion paper, temperature and precipitation hindcasts were produced with the Meteorological Ensemble Forecast Processor (MEFP) and verified against observed temperature and precipitation, respectively. Inputs to the MEFP comprised raw precipitation and temperature forecasts from the frozen (circa 1997) version of the NWS Global Forecast System (MEFP-GFS) and a conditional or “resampled” climatology (MEFP-CLIM). For this paper, streamflow hindcasts were produced with the Community Hydrologic Prediction System and were bias-corrected with the Ensemble Post-processor (EnsPost). In order to separate the meteorological and hydrologic uncertainties, the raw streamflow forecasts were verified against simulated streamflows, as well as observed flows. Also, when verifying the bias-corrected streamflow forecasts, the total skill was decomposed into contributions from the MEFP-GFS and the EnsPost. In general, the streamflow forecasts are substantially more skillful when using the MEFP-GFS together with the EnsPost than using the MEFP with resampled climatology alone. However, both the raw and bias-corrected streamflow forecasts have lower biases, stronger correlations and are more skillful in CB- and CN-RFCs than AB- and MA-RFCs. In addition, there are strong variations in forecast quality with streamflow amount, forecast lead time, season and aggregation period. The relative importance of the meteorological and hydrologic uncertainties also varies between basins and is modulated by the same controls on forecast quality. For example, the MEFP-GFS accounts for the majority of skill in the CNRFC basins. This is associated with the greater predictability of large storms in the North Coast Ranges during the winter months. In CBRFC, much of the skill in the streamflow forecasts originates from the hydrologic modeling and the EnsPost, particularly during the snowmelt period. In AB- and MA-RFCs, the contributions from the MEFP and the EnsPost are more variable. This paper summarizes the verification results, describes the expected performance and limitations of the HEFS for short- to medium-range streamflow forecasting, and provides recommendations for future research.