This paper presents a new operational forecasting system (driven by atmospheric forecasts), predicting river flow in European rivers for the next 7 months. For the first month only, these river flow forecasts are, on average, better than predictions that do not make use of atmospheric forecasts. Overall, this forecasting system can predict whether abnormally high or low river flows will occur in the next 7 months in many parts of Europe, and could be valuable for various applications.

Hydropower makes up nearly half of Sweden's electrical energy production. Careful reservoir management is required for optimal production throughout the year and accurate seasonal forecasts are essential for this. In this work we develop a seasonal forecast prototype and evaluate its ability to predict spring flood volumes, a critical variable, in northern Sweden. We show that the prototype is better than the operational system on average 65 % of the time and reduces the volume error by ~ 6 %.

This study aims to bridge the gap between water resources forecasts from the short-range (days) up to the long-range (seasonal) timescale. Applications of such a system are typically the waterpower industry, river and lake transports, and water resources management. The study uses a new meteorological forecast product combined with a hydrological model to predict river discharge over the major European river basins. The results show an improvement in comparison with the current system.

We benchmarked when and where ensemble streamflow prediction (ESP) is skilful in the UK across a diverse set of 314 catchments. We found ESP was skilful in the majority of catchments across all lead times up to a year ahead, but the degree of skill was strongly conditional on lead time, forecast initialization month, and individual catchment location and storage properties. Results have practical implications for current operational use of the ESP method in the UK.

Predictions of irrigation-season water availability are important for water-limited Pakistan. We assess a Bayesian joint probability approach, using flow and climate indices as predictors, to produce streamflow forecasts for inflow to Pakistan's two largest dams. The approach produces skilful and reliable forecasts. As it is simple and quick to apply, it can be used to provide probabilistic seasonal streamflow forecasts that can inform Pakistan's water management.

We describe a forecasting procedure that is based on a semi-distributed hydrological model using two types of weather ensembles for the lead time period: observed weather data, constructed on the basis of the ESP methodology, and synthetic weather data, simulated by a weather generator. We compare the described methodology with the regression-based operational forecasts that are currently in practice and show the increased informational content of the ensemble-based forecasts.

Meteorological forecasting agencies constantly work on pushing the limit of predictability farther in time. However, some end users need proof that climate model outputs are ready to be implemented operationally. We show that bias correction is crucial for the use of ECMWF System4 forecasts for the studied area and there is a potential for the use of 1-month-ahead forecasts. Beyond this, forecast performance is equivalent to using past climatology series as inputs to the hydrological model.

This work developed models to predict how much water will be available in the next month to maximise environmental and social outcomes in southern Australia. Initialising the models with observed streamflow data, instead of warmed up by rainfall data, improved the results, even at a monthly lead time, making sure only data representative of the forecast period to develop the models were also important. If this step was ignored, and instead all data were used, poor predictions could be produced.

The skill of an experimental streamflow forecast system in the Ahlergaarde catchment, Denmark, is analyzed. Inputs to generate the forecasts are taken from the ECMWF System 4 seasonal forecasting system and an ensemble of observations (ESP). Reduction of biases is achieved by processing the meteorological and/or streamflow forecasts. In general, this is not sufficient to ensure a higher level of accuracy than the ESP, indicating a modest added value of a seasonal meteorological system.

The present study evaluates the skill of a seasonal forecasting system for hydrological relevant variables in Denmark. Linear scaling and quantile mapping were used to correct the forecasts. Uncorrected forecasts tend to be more skillful than climatology, in general, for the first month lead time only. Corrected forecasts show a reduced bias in the mean; are more consistent; and show a level of accuracy that is closer to, although no higher than, that of ensemble climatology, in general.

We assess a new streamflow forecasting system in Australia. The system is designed to meet the need of water agencies for 12-month forecasts. The forecasts perform well in a wide range of rivers. Forecasts for shorter periods (up to 6 months) are generally informative. Forecasts sometimes did not perform well in a few very dry rivers. We test several techniques for improving streamflow forecasts in drylands, with mixed success.

Forecasting at the seasonal timescale aims to answer questions such as the following: how much water do we have next summer? Is next winter going to be extremely cold? Constrained by computer power, earth system models (ESMs) do not resolve all environmental variables of interest. Our study tests a method to refine the output of such an ESM for streamflow forecasting in the Rhine basin. The results show that the method is able to translate skill at different spatial scales.

Rainfall forecasts from dynamical global climate models (GCMs) require post-processing before use in hydrological models. Existing methods generally lack the sophistication to achieve calibrated forecasts of both daily amounts and seasonal accumulated totals. We develop a new statistical method to post-process Australian GCM rainfall forecasts for 12 perennial and ephemeral catchments. Our method produces reliable forecasts and outperforms the most commonly used statistical method.

This article presents predictability analyses of snow accumulation for the upcoming winter season. The results achieved using two coupled atmosphere–ocean general circulation models and a water balance model show that the tendency of snow water equivalent anomalies (i.e. the sign of anomalies) is correctly predicted in up to 11 of 13 years. The results suggest that some seasonal predictions may be capable of predicting tendencies of hydrological model storages in parts of Europe.

Central Asia crucially depends on water resources supplied by snow melt in the mountains during summer. To support water resources management we propose a generic tool for statistical forecasts of seasonal discharge based on multiple linear regressions. The predictors are observed precipitation and temperature, snow coverage, and discharge. The automatically derived models for 13 different catchments provided very skilful forecasts in April, and acceptable forecasts in January.

The paper contributes a comparative analysis of the operational value of different forecast products for the agricultural sector. The underlying idea is that forecast accuracy per se might not be the most effective indicator of the goodness of a forecast but the behaviour of the farmers using the forecast should somehow also be taken into consideration and, correspondingly, the benefit generated by the forecast. This paper explores and validates this idea with a model-based analysis.

Inland waterway transport is a commercial sector relying on hydrological forecasts on different timescales. This paper describes the set-up of a monthly to seasonal forecasting system for the German waterways. Multiple approaches are tested, compared and combined. Despite the predictive limitations on longer lead times, this study reveals the existence of a valuable predictability on monthly up to seasonal timescales. Forecast quality depends on forecast location, lead time and season.

The theme of the paper is exploration of the potential for seasonal flow forecasting for large lakes using a range of stochastic transfer function techniques with additional insights gained from simple analytical approximations. The methods were evaluated using records for two of the largest lakes in the world. The paper concludes with a discussion of the relevance of the results to operational flow forecasting systems for other large lakes.

Some do not realize the intrinsic importance of water until there is no water left. This is the reality faced by people in southern Peru, a dry area of the world where several economic activities and cities vie for scarce water resources. With the developed season-ahead precipitation prediction model, stakeholders and decision makers in this region will have another tool in their belt to respond to and plan for the negative impacts brought on by drought.

Process prediction of seasonal drought is the goal of our study. We developed a drought prediction model based on atmospheric–oceanic anomalies. It is essentially the synchronous statistical relationship between atmospheric–oceanic anomalies and precipitation anomalies, forced by seasonal climate forecast models. It can predict seasonal drought development very well, despite its weakness in drought severity.

The research presented here provides the first evaluation of the skill of a seasonal hydrological forecast for the UK. The forecast scheme combines rainfall forecasts from the Met Office GloSea5 forecast system with a national-scale hydrological model to provide estimates of river flows 1 to 3 months ahead. The skill in the combined model is assessed for different seasons and regions of Britain, and the analysis indicates that Autumn/Winter flows can be forecast with reasonable confidence.

The study proposes advancing local-level seasonal rainfall predictions by first conditioning on regional-level predictions, as defined through cluster analysis. This statistical approach is applied to western Ethiopia, where lives and livelihoods are vulnerable to its high spatial–temporal rainfall variability, particularly given the high reliance on rain-fed agriculture. The statistical model improves in skills versus the non-clustered case or dynamical models for some critical regions.

This work provides forecasts of water supply for the semi-arid Elqui River Valley, Chile, at periods prior to the October–January growing season. Forecasts are constructed provide water rights holders, whose allocations are subject to annual change, with an advanced indication of expected allocations. Forecasts, based on global and local indicators, are best suited to provide an initial indication of allocation category (above, near, or below normal) in May and are quantified in September.

Disaster managers would like to use seasonal forecasts to anticipate flooding months in advance. However, current seasonal forecasts give information on rainfall instead of flooding. Here, we find that the number of extreme events, rather than total rainfall, is most related to flooding in different regions of Africa. We recommend several forecast adjustments and research opportunities that would improve flood information at the seasonal timescale in different regions.

Water supply forecasts are critical to support water resources operations and planning. The skill of such forecasts depends on our knowledge of (i) future meteorological conditions and (ii) the amount of water stored in a basin. We address this problem by testing several approaches that make use of these sources of predictability, either separately or in a combined fashion. The main goal is to understand the marginal benefits of both information and methodological complexity in forecast skill.

This study investigates the relationship between skill and value of ensemble seasonal streamflow forecasts. Using data from a modern forecasting system, we show that skilled forecasts are more likely to provide benefits for reservoirs operated to maintain a target water level rather than reservoirs operated to satisfy a target demand. We identify the primary causes for this behaviour and provide specific recommendations for assessing the value of forecasts for reservoirs with supply objectives.

This paper presents the development and skill analysis of WUSHP, a system that makes hydrological forecasts for time horizons up to 7 months. Hot spots of significant skill in river discharge were identified in Fennoscandia (from January to October), the southern part of the Mediterranean (from June to August), Poland, north Germany, Romania and Bulgaria (mainly from November to January), and west France (from December to May). Some skill is left at the end of the forecasts.

This paper explains why forecasts of river flow in Europe for a time between 1 and 7 months have skill. The forecasts were produced with a water model. The model reacts to forecasts of weather variables like precipitation, which tend to have little skill and hence hardly contribute to the skill in the forecasts of river flow. The paper shows when and where these forecasts have skill; this is mostly due to knowledge of the amount of water in the soil at the time the forecasts are made.

This paper presents a skill assessment of the global seasonal streamflow forecasting system FEWS-World. For 20 large basins of the world, forecasts using the ESP procedure are compared to forecasts using actual S3 seasonal meteorological forecast ensembles by ECMWF. The results are discussed in the context of prevailing hydroclimatic conditions per basin. The study concludes that in general, the skill of ECMWF S3 forecasts is close to that of the ESP forecasts.

This study evaluated dynamic climate model sub-seasonal forecasts for important precipitation and temperature indices over the contiguous United States. The presence of active Madden-Julian Oscillation (MJO) events improved weekly mean precipitation forecast skill over most regions. Sub-seasonal forecast indices calculated from the daily forecast showed higher skill than temporally downscaled forecasts, suggesting the usefulness of the daily forecast for sub-seasonal hydrological forecasting.

The use of general circulation model outputs for streamflow forecasting has developed in the last decade. In parallel, traditional streamflow forecasting is commonly based on historical data. This study investigates the impact of conditioning historical data based on circulation model precipitation forecasts on seasonal streamflow forecast quality. Results highlighted a trade-off between the sharpness and reliability of forecasts.

Australian seasonal streamflow forecasts are issued by the Bureau of Meteorology with up to two weeks' delay. Timelier forecast release will enhance forecast value and enable sub-seasonal forecasting. The bureau's forecasting approach is modified to allow timelier forecast release, and changes in reliability and skill are quantified. The results are combined with insights into the forecast production process to recommend a more flexible forecasting system to better meet the needs of users.

This study examined the potential of snow water equivalent data assimilation to improve seasonal streamflow predictions. We examined aspects of the data assimilation system over basins with varying climates across the western US. We found that varying how the data assimilation system is implemented impacts forecast performance, and basins with good initial calibrations see less benefit. This implies that basin-specific configurations and benefits should be expected given this modeling system.

Most statistically based seasonal precipitation forecast models utilize a small set of well-known climate indices as potential predictor variables. However, for many target regions, these indices do not lead to sufficient results and customized predictors are required for an accurate prediction. This study presents a statistically based routine, which automatically identifies suitable predictors from globally gridded SST and climate variables by means of an extensive data mining procedure.

This study investigates the way bias correcting precipitation forecasts can improve the skill of streamflow forecasts at extended lead times. Eight variants of bias correction approaches based on the linear scaling and the distribution mapping methods are applied to the precipitation forecasts prior to generating the streamflow forecasts. One of the main results of the study is that distribution mapping of daily values is successful in improving forecast reliability.

Oceanic–atmospheric climate modes, such as El Niño–Southern Oscillation (ENSO), are known to affect the streamflow regime in many rivers around the world. A new method is presented for ENSO conditioning of the ensemble streamflow prediction (ESP) method, which is often used for seasonal streamflow forecasting. The method was tested on three tributaries of the Columbia River, OR. Results show an improvement in forecast skill compared to the standard ESP.

Seasonal early warning is vital for drought management in arid regions like the Limpopo Basin in southern Africa. This study shows that skilled seasonal forecasts can be achieved with statistical methods built upon driving factors for drought occurrence. These are the hydrological factors for current streamflow and meteorological drivers represented by anomalies in sea surface temperatures of the surrounding oceans, which combine to form unique combinations in the drought forecast models.