Probabilistic Seasonal Forecasts Research Articles

Hypothesis testing for performance evaluation of probabilistic seasonal rainfall forecasts

Hypothesis testing for performance evaluation of probabilistic seasonal rainfall forecasts

Probabilistic seasonal precipitation forecasts using quantiles of ensemble forecasts

Seasonal precipitation forecasting is vital for weather-sensitive sectors. Global Circulation Models (GCM) routinely produce ensemble Seasonal Climate Forecasts (SCFs) but suffer from issues like low forecast resolution and skills. To address these issues in this study, we introduce a post-processing method, Quantile Ensemble Bayesian Model Averaging (QEBMA). It utilises quantiles from a GCM ensemble forecast to create a pseudo-ensemble forecast. Through their reasonable linear relationships with observations, each pseudo-member connects a hurdle distribution with a point mass at zero for dry months and a gamma distribution for wet months. These distributions are mixed to construct a forecast probability distribution with their weights, proportional to the quantiles’ historical forecast performance. QEBMA is applied to three GCMs, including GloSea5 from the United Kingdom, ECMWF from Europe and ACCESS-S1 from Australia, for monthly precipitation forecasts in 32 locations across four climate zones in Australia. Leave-one-month-out cross-validation results illustrate that QEBMA enhances forecast skills compared to raw GCMs and other post-processing techniques, including quantile mapping and Extended Copula Post-Processing (ECPP), for forecast lead time of 0 to 2 months, based on five metrics. The skill improvements achieved by QEBMA are often statistically significant, particularly when compared to raw GCM forecasts across the 32 study locations. Among these post-processing models, only QEBMA consistently outperforms the SCF benchmark climatology, offering a promising alternative for improving seasonal precipitation forecasts.

Probabilistic seasonal forecasts of North Atlantic atmospheric circulation using complex systems modelling and comparison with dynamical models

AbstractDynamical seasonal forecast models are improving with time but tend to underestimate the amplitude of atmospheric circulation variability and to have lower skill in predicting summer variability than in winter. Here, we construct Nonlinear AutoRegressive Moving Average models with eXogenous inputs (NARMAX) to develop the analysis of drivers of North Atlantic atmospheric circulation and jet‐stream variability, focusing on the East Atlantic (EA) and Scandinavian (SCA) patterns as well as the North Atlantic Oscillation (NAO) index. New time series of these indices are developed from empirical orthogonal function (EOF) analysis. Geopotential height data from the ERA5 reanalysis are used to generate the EOFs. Sets of predictors with known associations with these drivers are developed and used to formulate a sliding‐window NARMAX model. This model demonstrates a high degree of predictive accuracy, as indicated by its average correlation coefficients over the testing period (2006–2021): 0.78 for NAO, 0.83 for EA and 0.68 for SCA. In comparison, the SEAS5 and GloSea5 dynamical forecast models exhibit lower correlations with observed circulation changes: for NAO, the correlation coefficients are 0.51 for SEAS5 and 0.34 for GloSea5, for EA they are 0.15 and 0.09, respectively, and for SCA, they are 0.28 and 0.24, respectively. Comparison of NARMAX predictions with forecasts and hindcasts from the SEAS5 and GloSea5 models highlights areas where NARMAX can be used to help improve seasonal forecast skill and inform the development of dynamical models, especially in the case of summer.

Pairing monitoring datasets with probabilistic forecasts to provide early warning of drought in Australia

Droughts are a cyclical feature of Australia's climate that have compounding impacts on agricultural productivity and the wellbeing of drought-affected communities. Understanding anticipated drought conditions in context of antecedent observations is critical to providing early warning of drought. In this study we paired probabilistic seasonal rainfall forecasts with precipitation, soil moisture and evapotranspiration data that were objectively combined using Principal Component Analysis (PCA). The final Drought Early Warning System (DEWS) maps overlay forecasting information with the multi-variate PCA-weighted maps at 1-, 3- and 6-month timescales over the common period of overlap between all datasets (1982–2018). In this study period, the 1982–1983 Ash Wednesday tinder drought and the 1997–2001 Millennium drought were investigated. We validated PCA-weighted maps with satellite vegetation data and found performance was strongest over the Murray Darling Basin region (R = 0.63, p = 0.009) and poorest over Central interior Australia (insignificant correlations). We also validated PCA-weighted maps using agricultural commodity data from ABARES. Significant negative correlations at 95 %, 99 % and 99.9 % confidence intervals were found between %-Area in drought category and crop cultivation area; export volume/value; crop yield; and rural debt. Our findings indicate that early warning of drought can be categorised by concern – wherein dry antecedent conditions and dry forecasted conditions are of the highest concern. The developed proof-of-concept DEWS contributes to the growing body of proactive drought management research. In a drought vulnerable future, operationalising and communicating drought early warnings will be critical to reducing the harmful impacts of drought on communities, economies, and environments.

Seasonal seed scenario planning: co-design of a generic framework for matching seed supply and demand using seasonal climate forecasts

To cope with interannual climate variability, many farmers in tropical and sub-tropical regions choose crop varieties that fit seasonal climate conditions. Therefore, seed demand for different varieties, such as early- or late-maturing cultivars, varies between years. Resulting mismatches between relatively constant supply and variable demand create losses for both seed suppliers and farmers. Because demand for seed of different varieties is influenced by seasonal climate, however, probabilistic seasonal rainfall forecasts could help seed suppliers better anticipate upcoming seed demand. To explore this idea, we engaged decision-makers from seed supply organizations in Zimbabwe and Ethiopia. Through a participatory design process, we identified opportunities and challenges for using seasonal rainfall forecasts to inform seed supply decisions. In a case study of maize seed sales in Zimbabwe, we tested our assumptions and iteratively devised a systematic procedure for forecast-based planning in seed supply, relying on free online data sources and expert deliberations. We found that currently accessible rainfall forecasts could indeed be useful for prioritizing likely high-demand varieties during the stages of seed treatment, packaging, and logistics. In practice, though, more flexible and adaptive management of seed supply pipelines might be required to make use of seed demand forecasts. In the future, targeting farmers with climate forecasts along with recommended variety portfolios may strengthen the association between seasonal climate and farmers’ variety demand, increasing the accuracy of demand anticipation. This study highlights opportunities for increased case-specific collaboration between climate scientists and the seed sector to make seasonal forecast information operational.

A Machine Learning Approach for Probabilistic Multi-Model Ensemble Predictions of Indian Summer Monsoon Rainfall

Due to the uncertainty associated with Indian summer monsoon rainfall (ISMR), probabilistic seasonal forecasts which can convey the inherent uncertainty of ISMR are more useful to the user community than a single deterministic forecast. While such probabilistic seasonal forecasts can be produced from general circulation model (GCM) output, one single model generally does not represent all sources of error. The probabilistic multi model ensemble (PMME) is a well-accepted way to improve on the skill of probabilistic forecasts by individual GCMs. PMME can be constructed with one of two approaches: non-parametric, or parametric with respect to the occurrence of three categories of seasonal total rainfall—below, near, and above normal as defined by the climatological base period. Both the methods have their limitations. Non-parametric PMME use a smaller ensemble size which results in overconfident forecasts, and parametric PMME make the inaccurate assumption that total rainfall follows a Gaussian distribution. To avoid these problems, we propose the use of the Extreme Learning Machine (ELM), a novel machine learning (ML) approach, to construct PMME for ISMR forecasting. ELM is a state-of-the-art generalized form of single-hidden-layer feed-forward neural network. However, since the traditional ELM network only produces a deterministic outcome, we use a modified version of ELM called Probabilistic Output Extreme Learning Machine (PO-ELM). PO-ELM uses sigmoid additive neurons and slightly different linear programming to make probabilistic predictions. The performance of such PO-ELM based PMME is assessed rigorously in terms of Generalized Receiver Operating Characteristic scores and reliability diagrams over a 37 years period spanning from 1982 to 2018 following a leave-three-year-out cross-validation scheme. It is demonstrated that our new strategy for PMME based on ML is capable of producing skillful MME forecasts over large regions of India.

Seasonal fire danger forecasts for supporting fire prevention management in an eastern Mediterranean environment: the case of Attica, Greece

Abstract. Forest fires constitute a major environmental and socioeconomic hazard in the Mediterranean. Weather and climate are among the main factors influencing forest fire potential. As fire danger is expected to increase under changing climate, seasonal forecasting of meteorological conditions conductive to fires is of paramount importance for implementing effective fire prevention policies. The aim of the current study is to provide high-resolution (∼9 km) probabilistic seasonal fire danger forecasts, utilizing the Canadian Forest Fire Weather Index (FWI) for the Attica region, one of the most fire-prone regions in Greece and the Mediterranean, employing the fifth-generation ECMWF seasonal forecasting system (SEAS5). Results indicate that, depending on the lead time of the forecast, both the FWI and ISI (Initial Spread Index) present statistically significant high discrimination scores and can be considered reliable in predicting above-normal fire danger conditions. When comparing the year-by-year fire danger predictions with the historical fire occurrence recorded by the Hellenic Fire Service database, both seasonal FWI and ISI forecasts are skilful in identifying years with a high number of fire occurrences. Overall, fire danger and its subcomponents can potentially be exploited by regional authorities in fire prevention management regarding preparedness and resources allocation.

Flexible forecast presentation overcomes longstanding obstacles to using probabilistic seasonal forecasts

We describe an innovative forecast presentation that aims to overcome obstacles to using seasonal climate forecasts for decision making, trace factors that influenced how seasonal forecast conventions have evolved, and describe a workshop process for training and supporting farmers in sub-Saharan Africa to use probabilistic seasonal forecasts. Mainstreaming seasonal climate forecasts through Regional Climate Outlook Forums (RCOFs) was an important milestone in the development of climate services. Most RCOFs and National Meteorological Services (NMS) adopted a subjective process to arrive at a consensus among different sources of prediction, and express the forecast as probabilities that rainfall in the upcoming season will fall in “below-normal,” “normal” or “above-normal” historical tercile categories. The Flexible Forecast is an online presentation that rectifies the main criticisms of the tercile convention by presenting downscaled forecasts as full probability distributions in probability-of-exceedance format along with the historical climate distribution. A map view provides seasonal forecast quantities, anomalies, or probabilities of experiencing above or below a user-selected threshold in amount or percentile, at the spatial resolution of the underlying gridded data (typically 4 to 5 km). We discuss factors that contributed to the persistence of the tercile convention, and milestones that paved the way to adopting seasonal forecast methods and formats that better align with user needs. The experience of adopting the new flexible forecast presentation regionally and at a national level in Eastern Africa illustrates the challenges and how they can be overcome. We also describe a seasonal forecast training and planning workshop process that has been piloted with smallholder farmers in several African countries. Beginning with participants' collective memory of past seasonal climate variations, the process leads them incrementally to understand the forecast presented in probability-of-exceedance format, and apply it to their seasonal planning decisions.

Probability Games, Workshops, and the Social World of International Science Communication

Abstract Probabilistic thinking underpins a wide range of scientific claims, but effectively communicating probabilistic information across audiences is challenging. In this article, we present a political–institutional approach to science that harnesses the social relationships between the people working as scientists and the public using scientific innovations. First, we show how we learned to use games and local analogies to effectively communicate probabilistic seasonal forecasts of weather and crop yields with farmers, extension workers, and water managers in Ethiopia. Second, we show how workshops—the unglamorous institutional workhorse of international development and scientific enterprises—became warmhearted events when organized around the fundamental fact of social connections between researchers and the community members and between the community members themselves. Scientists in an international scientific collaboration may not be able to become longstanding members of every community, but our approach to workshopping—and to research networks—allowed us to be engidoch (in English, guests), to tap into rich social ties to harness the humor, goodwill, and commitment that is hard to muster when scientists engage with community members as unconnected, nameless “workshop participants.”

A simplified seasonal forecasting strategy, applied to wind and solar power in Europe

We demonstrate levels of skill for forecasts of seasonal-mean wind speed and solar irradiance in Europe, using seasonal forecast systems available from the Copernicus Climate Change Service (C3S). While skill is patchy, there is potential for the development of climate services for the energy sector. Following previous studies, we show that, where there is skill, a simple linear regression-based method using the hindcast and forecast ensemble means provides a straightforward approach for producing calibrated probabilistic seasonal forecasts. This method extends naturally to using a larger-scale feature of the climate, such as the North Atlantic Oscillation, as the climate model predictor, and we show that this provides opportunities to improve the skill in some cases.We further demonstrate that, on seasonal-average and regional (e.g. national) average scales, wind and solar power generation are highly correlated with single climate variables (wind speed and irradiance). The detailed non-linear transformations from meteorological quantities to energy quantities, which are essential for detailed simulation of power system operations, are usually not necessary when forecasting gross wind or solar generation potential at seasonal-mean regional-mean scales.Together, our results demonstrate that where there is skill in seasonal forecasts of wind speed and irradiance, or a correlated larger-scale climate predictor, skilful forecasts of seasonal mean wind and solar power generation can be made based on the climate variable alone, without requiring complex transformations. This greatly simplifies the process of developing a useful seasonal climate service.

Improving early-season wheat yield forecasts driven by probabilistic seasonal climate forecasts

Seasonal climate forecasts (SCF) are evolving rapidly alongside improvements in climate modelling and downscaling research, and have great potential for weather-sensitive sectors, especially agriculture, by reducing weather-related risks and increasing productivity. Skilful yield forecasts at the beginning of, or before, a cropping season can provide farmers and other stakeholders in agribusiness with the necessary information for early planning and actions. Only a few yield forecast studies have a forecast lead time of four months or longer due to the problem complexity. To enable SCFs from Global Climate Models (GCMs) to be used for early-season yield forecasts, this paper uses a statistical downscaling technique, Extended Copula Post-Processing (ECPP) and the Schaake shuffle, to downscale four climate variables to generate weather-like daily data that are suitable for agricultural applications. Climate forecasts drive a process-based crop model APSIM (Agricultural Production Systems sIMulator) to simulate crop forecasts on 50 stations, well-distributed across the Australian grain zone. To focus on yield forecast skills attributable to SCF, we propose best practice management rules to predict water-limited winter wheat yield. Yield forecasts from ECPP have a significant improvement over quantile mapping downscaling and raw SCF from the Australian recent seasonal forecast model ACCESS-S1 in terms of bias, accuracy, reliability, and overall forecast skill. In addition, even at the beginning of a cropping season with a forecast lead time of four or more months, yield forecasts driven by ECPP illustrate higher skill than climatology, a benchmark for yield forecast. Early-season yield forecasts driven by SCFs provides a promising alternative to regression/machine-learning-based forecasts. Performance sensitivity and issues, and gaps on using skilful SCFs to help growers with their farming decision-making are discussed.

The iCOLT climate service: Seasonal predictions of irrigation for Emilia‐Romagna, Italy

AbstractEmilia‐Romagna is a region of Northern Italy whose land use is mainly devoted to agriculture, especially over the plain area. Since the region is characterized by summer water shortages and recurrent drought conditions, a summer irrigation water demand forecast system has been developed, operating both at regional and Land Reclamation and Irrigation Board levels. Remote sensing crop detection is used in combination with a soil water balance model as a monitoring tool in support to water management and to optimize the use of available water resources. These instruments are also integrated with operational calibrated probabilistic seasonal forecast products currently based on Copernicus Climate Services, in order to evaluate summer irrigation water needs in advance in support to decision‐making by water managers. Results obtained using this prediction system are characterized by a significant forecast skill. The sources of the predictability of this system are specifically evaluated, showing that the correct representation of crop geographical distribution is crucial for the ability of the system to capture spatial variability in irrigation demand. Furthermore, the system forecast skill is mostly due to the use of an empirical model to assess the hypodermic water table depth initialized by means of the observed meteorological data of the months before the beginning of the summer season. Finally, calibrated probabilistic seasonal predictions also contribute to a minor extent to the good performance of the system as a whole.

Benchmarking an operational hydrological model for providing seasonal forecasts in Sweden

Abstract. Probabilistic seasonal forecasts are important for many water-intensive activities requiring long-term planning. Among the different techniques used for seasonal forecasting, the ensemble streamflow prediction (ESP) approach has long been employed due to the singular dependence on past meteorological records. The Swedish Meteorological and Hydrological Institute is currently extending the use of long-range forecasts within its operational warning service, which requires a thorough analysis of the suitability and applicability of different methods with the national S-HYPE hydrological model. To this end, we aim to evaluate the skill of ESP forecasts over 39 493 catchments in Sweden, understand their spatio-temporal patterns, and explore the main hydrological processes driving forecast skill. We found that ESP forecasts are generally skilful for most of the country up to 3 months into the future but that large spatio-temporal variations exist. Forecasts are most skilful during the winter months in northern Sweden, except for the highly regulated hydropower-producing rivers. The relationships between forecast skill and 15 different hydrological signatures show that forecasts are most skilful for slow-reacting, baseflow-dominated catchments and least skilful for flashy catchments. Finally, we show that forecast skill patterns can be spatially clustered in seven unique regions with similar hydrological behaviour. Overall, these results contribute to identifying in which areas and seasons and how long into the future ESP hydrological forecasts provide an added value, not only for the national forecasting and warning service, but also, most importantly, for guiding decision-making in critical services such as hydropower management and risk reduction.

A theoretical relationship between probabilistic relative operating characteristic skill and deterministic correlation skill in dynamical seasonal climate prediction

Understanding the relationship between probabilistic and deterministic prediction skills is of important significance for the study of seasonal forecasting and verification. Based on the Brier skill score methodology, we have previously found a theoretical relationship between the probabilistic resolution skill and the deterministic correlation (i.e., anomaly correlation; AC) skill and a lack of necessary or consistent relationship between the probabilistic reliability skill and the deterministic skill in dynamical seasonal prediction. Here, we further theoretically investigate the relationship between the probabilistic relative operating characteristic (ROC) skill and the deterministic skill. The ROC measures the discrimination attribute of probabilistic forecast quality, another important attribute besides the resolution and reliability. With some simplified assumptions, we first derive theoretical expressions for the hit and false-alarm rates that are basic ingredients for the ROC curve, then demonstrate a sole dependence of the ROC curve on the AC, and finally analytically derive a relationship between the related ROC score and the AC. Such a theoretically derived ROC-AC relationship is further examined using dynamical models’ ensemble seasonal hindcasts, which is well verified. The finding here along with our previous findings implies that the discrimination and resolution attributes of probabilistic seasonal forecast skill are intrinsically equivalent to the corresponding deterministic skill, while the reliability appears to be the fundamental attribute of the probabilistic skill that differs from the deterministic skill, which constitutes an understanding of the fundamental similarities and difference between the two types of seasonal forecasting skills and predictability and can offer important implications for the study of seasonal forecasting and verification.

Machine learning-based integration of large-scale climate drivers can improve the forecast of seasonal rainfall probability in Australia

Probabilistic seasonal rainfall forecasting is of great importance for stakeholders such as farmers and policymakers to assist in developing risk management strategies and to inform decisions. In practice, there are two kinds of commonly used tools, dynamical models and statistical models, to provide probabilistic seasonal rainfall forecasts. Dynamical models are based on physical processes but are usually expensive to operate and implement, and rely overly on initial conditions. Statistical models are easy to implement but are usually based on simple or linear relationships between observed variables. Recently, machine learning techniques have been widely used in climate projection and perform well in reproducing historical climate. For these reasons, we conducted a case study in Australia by developing a machine learning-based probabilistic seasonal rainfall forecasting model using multiple large-scale climate indices from the Pacific, Indian and Southern Oceans. Rainfall probabilities of exceeding the climatological median for upcoming seasons from 2011 to 2018 were successively forecasted using multiple climate indices of precedent six months. The performance of the model was evaluated by comparing it with an officially used forecasting model, the SOI (Southern Oscillation Index) phase model (SP) operated by Queensland government in Australia. Results indicated that the random forest (RF) model outperformed the SP model in terms of both distinct forecasts and forecasting accuracy. The RF model increased the percentages of distinct forecasts to 64.9% for spring, to 71.5% for summer, to 65.8% for autumn, and to 63.9% for winter, 1.4 ∼ 3.2 times of the values from the SP model. Forecasting accuracy was also greatly increased by 28%, 167%, 219%, and 76% for four seasons respectively, compared to the SP model. The proposed rainfall forecasting model is based on readily available data, and we believe it can be easily extended to other regions to provide seasonal rainfall outlooks.

Evaluation of a New Precipitation-Based Index for Global Seasonal Forecasting of Unusually Wet and Dry Periods

AbstractThis paper describes the assessment of the performance of a method for providing early warnings of unusually wet and dry precipitation conditions globally. The indicator that is used for forecasting these conditions is computed from forecasted standardized precipitation index (SPI) values for accumulation periods of 1, 3, and 6 months. The SPI forecasts are derived from forecasted precipitation produced by the latest probabilistic seasonal forecast of ECMWF. Early warnings of unusual precipitation periods are shown only when and where the forecast is considered robust (i.e., with at least 40% of ensemble members associated with intense forecasts), and corresponding with significant SPI values (i.e., below −1 for dry, or above +1 for wet conditions). The intensity of the forecasted events is derived based on the extreme forecast index and associated shift of tails products developed by ECMWF. Different warning levels are then assessed, depending on the return period of the forecast intensity, and the coherence of the ensemble forecast members. The assessment of the indicators performance is based on the 25-member ensemble forecast system that is carried out every month during the 36 years of the hindcast period (1981–2016). The results show that significant information is provided even for the longest lead time, albeit with a large variability across the globe with the highest scores over central Russia, Southeast Asia, and the northern part of South America or Australia. Because of the loss of predictability, each SPI is based on the first lead time. A sensitivity test highlights the influence on the robustness of the forecasts of the warning levels used, as well as the effects of prior conditions and of seasonality.

Improving Seasonal Precipitation Forecasts for Agriculture in the Orinoquía Region of Colombia

Abstract Canonical correlation analysis (CCA) is used to improve the skill of seasonal forecasts in the Orinoquía region, where over 40% of Colombian rice is produced. Seasonal precipitation and frequency of wet days are predicted, as rice yields simulated by a calibrated crop model are better correlated with wet-day frequency than with precipitation amounts in June–August (JJA). Prediction of the frequency of wet days, using as predictors variables from the NCEP Climate Forecast System, version 2 (CFSv2), results in a forecast with higher skill than models predicting seasonal precipitation amounts. Using wet-day frequency as an alternative climate variable reveals that the distribution of daily rainfall is both more relevant for rice yield variability and more skillfully predicted than seasonal precipitation amounts. Forecast skill can also be improved by using the Climate Hazards Infrared Precipitation with Stations (CHIRPS) merged satellite–station JJA precipitation as the predictand in a CCA model, especially if the predictor is CFSv2 vertically integrated meridional moisture flux (VQ). The probabilistic hindcast derived from the CCA model using CHIRPS as the predictand can successfully discriminate above-normal, normal, and below-normal terciles of over 80% of the stations in the region. This is particularly relevant for stations that, due to discontinuity in their time series, are not included in station-only CCA models but are still in need of probabilistic seasonal forecasts. Finally, CFSv2 VQ performs better than precipitation as the predictor in CCA, which we attribute to CFSv2 being more internally consistent in regards to sea surface temperature (SST)-forced VQ variability than to SST-forced precipitation variability in the Orinoquía region.

A Grand Prediction: Communicating and Evaluating 2018 Summertime Upper Blue Nile Rainfall and Streamflow Forecasts in Preparation for Ethiopia's New Dam

When complete, the Grand Ethiopian Renaissance Dam (GERD) will be the largest hydropower dam in Africa. The GERD has become a focal point of geopolitical tensions because it will allow Ethiopia greater control over the Blue Nile River, Egypt’s main source of freshwater. To inform discussions of filling plans and responses, we created a probabilistic seasonal forecast for Upper Blue Nile rainfall and streamflow in the GERD basin. Eight statistical models and eight dynamical models were used to forecast the rainy season (June-September), which were then converted into river flow for June-December 2018. Both statistical and dynamical models predicted a high probability of average to above average rainfall as well as Upper Blue Nile flow in the GERD basin. Actual summer precipitation in 2018 was slightly below the long-term mean but well within the range considered to be “near normal.” Leveraging the increasingly online media landscape for science communication, we made the forecast publicly available through a blog and shared with regional decision-makers in advance of the 2018 rainy season. The blog attracted news coverage in the region focusing primarily on the relatively low likelihood of below-average Nile flow across the forecast ensemble. When asked for feedback on the blog, Ethiopian decision-makers and forecasters reported that flow predictions included in our blog were useful and not part of existing products. Access and comprehension were noted barriers to the use of these types of forecasts, consistent with prior research in forecast communication and dissemination. Forecasts available on such blogs can inform a shared understanding among decision-makers in the management of transboundary waters, yet effective communication and dissemination remain a challenge.

Application of a Hybrid Statistical–Dynamical System to Seasonal Prediction of North American Temperature and Precipitation

Abstract Recent research demonstrates that dynamical models sometimes fail to represent observed teleconnection patterns associated with predictable modes of climate variability. As a result, model forecast skill may be reduced. We address this gap in skill through the application of a Bayesian postprocessing technique—the calibration, bridging, and merging (CBaM) method—which previously has been shown to improve probabilistic seasonal forecast skill over Australia. Calibration models developed from dynamical model reforecasts and observations are employed to statistically correct dynamical model forecasts. Bridging models use dynamical model forecasts of relevant climate modes (e.g., ENSO) as predictors of remote temperature and precipitation. Bridging and calibration models are first developed separately using Bayesian joint probability modeling and then merged using Bayesian model averaging to yield an optimal forecast. We apply CBaM to seasonal forecasts of North American 2-m temperature and precipitation from the North American Multimodel Ensemble (NMME) hindcast. Bridging is done using the model-predicted Niño-3.4 index. Overall, the fully merged CBaM forecasts achieve higher Brier skill scores and better reliability compared to raw NMME forecasts. Bridging enhances forecast skill for individual NMME member model forecasts of temperature, but does not result in significant improvements in precipitation forecast skill, possibly because the models of the NMME better represent the ENSO–precipitation teleconnection pattern compared to the ENSO–temperature pattern. These results demonstrate the potential utility of the CBaM method to improve seasonal forecast skill over North America.

From Probabilistic Seasonal Streamflow Forecasts to Optimal Reservoir Operations: A Stochastic Programming Approach

Abstract We investigate the potential use of seasonal streamflow forecasts for the realtime operation of Angat reservoir (Philippines). The system is characterized by a strong intra- and inter-annual variability in the inflow process, which is further amplified by the El Nino Southern Oscillation (ENSO). We bank on the relationship between ENSO indices and local hydro-climatological processes to issue probabilistic streamflow forecasts (with a 3-month forecast horizon) and then integrate them within a Multistage Stochastic Programming (MSP) approach. The rolling-horizon, forecast-informed scheme is adopted for the period 1968–2014 and benchmarked against deterministic optimization solutions with perfect forecasts, climatology, and mean forecasts. We also compare its performance with the current operating rules, and the operating rules obtained by solving a Stochastic Dynamic Programming problem. Results show that the MSP approach can help reduce the severity of failures during prolonged droughts caused by ENSO.