Abstract. Sub-daily ensemble rainfall forecasts that are bias free and reliably quantify forecast uncertainty are critical for flood and short-term ensemble streamflow forecasting. Post processing of rainfall predictions from numerical weather prediction models is typically required to provide rainfall forecasts with these properties. In this paper, a new approach to generate ensemble rainfall forecasts by post processing raw NWP rainfall predictions is introduced. The approach uses a simplified version of the Bayesian joint probability modelling approach to produce forecast probability distributions for individual locations and forecast periods. Ensemble forecasts with appropriate spatial and temporal correlations are then generated by linking samples from the forecast probability distributions using the Schaake shuffle. The new approach is evaluated by applying it to post process predictions from the ACCESS-R numerical weather prediction model at rain gauge locations in the Ovens catchment in southern Australia. The joint distribution of NWP predicted and observed rainfall is shown to be well described by the assumed log-sinh transformed multivariate normal distribution. Ensemble forecasts produced using the approach are shown to be more skilful than the raw NWP predictions both for individual forecast periods and for cumulative totals throughout the forecast periods. Skill increases result from the correction of not only the mean bias, but also biases conditional on the magnitude of the NWP rainfall prediction. The post processed forecast ensembles are demonstrated to successfully discriminate between events and non-events for both small and large rainfall occurrences, and reliably quantify the forecast uncertainty. Future work will assess the efficacy of the post processing method for a wider range of climatic conditions and also investigate the benefits of using post processed rainfall forecast for flood and short term streamflow forecasting. HESS - Peer review - Post-processing rainfall forecasts from numerical weather prediction models for short-term streamflow forecasting Articles | Volume 17, issue 9 Article Peer review Metrics Related articles Articles | Volume 17, issue 9 https://doi.org/10.5194/hess-17-3587-2013 © Author(s) 2013. This work is distributed under the Creative Commons Attribution 3.0 License. https://doi.org/10.5194/hess-17-3587-2013 © Author(s) 2013. This work is distributed under the Creative Commons Attribution 3.0 License. Articles | Volume 17, issue 9 Article Peer review Metrics Related articles Research article | 27 Sep 2013 Research article | | 27 Sep 2013 Post-processing rainfall forecasts from numerical weather prediction models for short-term streamflow forecasting D. E. Robertson , D. L. Shrestha , and Q. J. Wang D. E. Robertson https://orcid.org/0000-0003-4230-8006 × CSIRO Land and Water, P.O. Box 56, Highett, 3190 Victoria, Australia D. L. Shrestha × CSIRO Land and Water, P.O. Box 56, Highett, 3190 Victoria, Australia Q. J. Wang https://orcid.org/0000-0002-8787-2738 × CSIRO Land and Water, P.O.

The accurate prediction of rainfall, and in particular rainfall extremes, remains challenging for numerical weather prediction models. This can be attributed to subgrid-scale parameterizations of processes that play a crucial role in the multi-scale dynamics, as well as the strongly intermittent nature and the highly skewed, non-Gaussian distribution of rainfall. Here we show that a specific type of deep neural networks can learn rainfall extremes from a numerical weather prediction ensemble. A frequency-based weighting of the loss function is proposed to enable the learning of extreme values in the distributions' tails. We apply our framework in a post-processing step to correct for errors in the model-predicted rainfall. Our method yields a much more accurate representation of relative rainfall frequencies and improves the forecast skill of extremes by factors ranging from two to above six, depending on the event magnitude.

The safety of earth-rockfill dams during construction is critically challenged by transient hydrological loads and evolving boundary conditions. While numerical and data-driven models exist, their standalone application is limited; the former is computationally prohibitive for real-time forecasting, and the latter often lacks physical interpretability. To bridge this gap, we introduce a novel hybrid framework that tightly couples Finite Element analysis (FEM) with deep learning (ANN-LSTM) for the physics-constrained forecasting of rainfall-induced instability. The model leverages FEM to simulate the hydro-mechanical response, the LSTM to capture temporal patterns in monitoring data, and an ANN to map strength degradation, with an attention mechanism identifying critical antecedent failure sequences. Validated on a two-year monitoring dataset from the Megech Dam, which experienced documented instabilities, our framework significantly outperformed established baselines. It achieved a superior MAE of 0.027 (vs. 0.081 for SVM, 0.067 for Random Forest, and 0.052 for standalone LSTM, p < 0.05) and provided an early-warning lead time of up to 3.5 weeks by identifying the critical lag between rainfall peaks and pore-pressure buildup. The integrated attention mechanism autonomously highlighted weeks 25–30 and 75–80 as high-risk periods, aligning with field observations. This work demonstrates that a physics-informed hybrid approach offers a more reliable and interpretable tool for early-warning systems than purely data-driven methods. The proposed framework is adaptable to other earth-rockfill dams, providing a pathway from reactive monitoring to proactive risk management during critical construction phases.

Hydrometeorological Prediction Center The Hydrometeorological Prediction Center (a.k.a. HPC) is one of nine Service Centers that are part of the National Centers for Environmental Prediction (NCEP), which is part of the National Weather Service, which in turn is part of the National Oceanic and Atmospheric Administration (NOAA) of the U.S. government. The HPC serves as a center of excellence in Quantitative Precipitation Forecasting, Medium Range Forecasting (three to seven days) and the interpretation of numerical weather prediction models. The HPC issues storm summaries on storm systems bringing significant rainfall and snowfall to portions of the United States. Advisories are also issued for tropical cyclones which have moved inland and are no longer the responsibility of the National Hurricane Center. The HPC also acts as the backup office to the National Hurricane Center in the event of a complete communications failure. The HPC was created in October 1995, when the former National Meteorological Center was reorganised into the NCEP. The Weather Prediction Center (WPC), located in College Park, Maryland, is one of nine service centers under the umbrella of the National Centers for Environmental Prediction (NCEP), a part of the National Weather Service (NWS), which in turn is part of the National Oceanic and Atmospheric Administration (NOAA) of the U.S. government. Until March 5, 2013, the Weather Prediction Center was known as the Hydrometeorological Prediction Center (HPC). The Weather Prediction Center serves as a center for quantitative precipitation forecasting, medium range forecasting (three to eight days), and the interpretation of numerical weather prediction computer models. The Weather Prediction Center issues storm summaries on storm systems bringing significant rainfall and snowfall to portions of the United States. They also forecast precipitation amounts for the lower 48 United States for systems expected to impact the country over the next seven days. Advisories are also issued for tropical cyclones which have moved inland, weakened to tropical depression strength, and are no longer the responsibility of the National Hurricane Center. The Weather Prediction Center also acts as the backup office to the National Hurricane Center in the event of a complete communications failure. Long range climatological forecasts are produced by the Climate Prediction Center (CPC), a branch of the National Weather Service. These include 8–14 day outlooks, monthly outlooks, and seasonal outlooks. The QPF desks prepare and issue forecasts of accumulating (quantitative) precipitation, heavy rain, heavy snow, and highlights areas with the potential for flash flooding, with forecasts valid over the following five days. These products are sent to the National Weather Service forecast offices and are available on the Internet for public use. Heavy snow forecast products, in association with the short-range public forecast products (described below), serve as a coordinating mechanism for the national winter storm watch and warning program. One desk of the National Environmental Satellite Data and Information Service (NESDIS) is co-located with the WPC QPF desks, which together form the National Precipitation Prediction Unit (NPPU). NESDIS meteorologists prepare estimates of rainfall and current trends based on satellite data, and this information is used by the Day 1 QPF forecaster to help create individual 6-hourly forecasts that cover the next 12 hours. With access to WSR-88D/Doppler weather radar data, satellite estimates, and NCEP model forecast data as well as current weather observations and WPC analyses, the forecaster has the latest data for use in preparation of short-range precipitation forecasts. Meteorological reasoning discussions are regularly written and issued with the forecast packages to explain and support the forecast. Medium range forecasters are responsible for preparing forecasts for three to seven days into the future. Surface pressure forecasts are issued three times per day, with temperature and probability of precipitation products issued twice per day, using guidance from the NWS medium range forecast model (GFS) as well as models from the European Centre for Medium Range Weather Forecasting (ECMWF), the United Kingdom's Meteorology Office (UKMET), Canadian model, the Navy NOGAPS model, and ensemble guidance from the GFS, ECMWF, Canadian, and North American Ensemble Forecast System (NAEFS). The medium range forecast products include surface pressure patterns, circulation centers and fronts, daily maximum and minimum temperatures and anomalies, probability of precipitation in 12-hour increments, total 5-day precipitation accumulation for the next five days, and 500 hPa (mb) height forecasts for days 3–7. In addition, a narrative is issued for each set of forecasts highlighting forecast reasoning and significant weather over the Continental United States. Separate forecasts, similar to the 5-day mean products, are prepared for Hawaii.

Understanding the hydrology of runoff source areas is crucial for predicting floods and evaluating chemical transport. Numerically modeling water fluxes in the source areas is quite complex. However, the self-organization of complex hydrological systems makes it possible to simplify watershed models by considering the landscape functions. The limitation is that input values are not known a priori. This study seeks to find the soil physical parameters governing the hydrology of runoff source areas in humid climates and use them in a surrogate simulation model to predict the runoff and the perched water table height. The site chosen was a 5.4-ha, periodically saturated runoff source area with a shallow perched water table and a humid temperate climate. The only inflow was from precipitation. Perched water table depths at five locations and the outflow were measured continuously. Measurements showed that the outflow was negligible 24 h after a rain event. It indicated that a quasi-static equilibrium had been established, in which the capillary pressure decreased linearly with depth to zero at the shallow groundwater. The soil–water retention function determined the soil water distribution and the drainable porosity. Runoff was generated by saturation excess and equaled the rainfall minus the empty pore volume. Based on these observations, a simple spreadsheet-based surrogate model was developed to calculate the air-filled pore volume by accounting for daily precipitation and eva

Monitoring of fecal indicator bacteria at recreational waters is an important public health measure to minimize water-borne disease, however traditional culture methods for quantifying bacteria can take 18-24 hours to obtain a result. To support real-time notifications of water quality, models using environmental variables have been created to predict indicator bacteria levels on the day of sampling. We conducted a systematic review of predictive models of fecal indicator bacteria at freshwater recreational sites in temperate climates to identify and describe the existing approaches, trends, and their performance to inform beach water management policies. We conducted a comprehensive search strategy, including five databases and grey literature, screened abstracts for relevance, and extracted data using structured forms. Data were descriptively summarized. A total of 53 relevant studies were identified. Most studies (n = 44, 83%) were conducted in the United States and evaluated water quality using E. coli as fecal indicator bacteria (n = 46, 87%). Studies were primarily conducted in lakes (n = 40, 75%) compared to rivers (n = 13, 25%). The most commonly reported predictive model-building method was multiple linear regression (n = 37, 70%). Frequently used predictors in best-fitting models included rainfall (n = 39, 74%), turbidity (n = 31, 58%), wave height (n = 24, 45%), and wind speed and direction (n = 25, 47%, and n = 23, 43%, respectively). Of the 19 (36%) studies that measured accuracy, predictive models averaged an 81.0% accuracy, and all but one were more accurate than traditional methods. Limitations identifed by risk-of-bias assessment included not validating models (n = 21, 40%), limited reporting of whether modelling assumptions were met (n = 40, 75%), and lack of reporting on handling of missing data (n = 37, 70%). Additional research is warranted on the utility and accuracy of more advanced predictive modelling methods, such as Bayesian networks and artificial neural networks, which were investigated in comparatively fewer studies and creating risk of bias tools for non-medical predictive modelling.