Hydrological forecasts strongly rely on predictions of precipitation amounts and temperature as meteorological forcings for hydrological models. Ensemble weather predictions provide a number of different scenarios that reflect the uncertainty about these meteorological inputs, but these are often biased and under‐dispersive, and therefore require statistical postprocessing. In addition to correcting the marginal distributions of the two weather variables, postprocessing methods must reconstruct their spatial, temporal, and intervariable dependence in order to generate physically realistic forecast trajectories that can be used as forcings of hydrological streamflow forecast models. For many years, a sample reordering method referred to as “Schaake shuffle” has been used successfully to address this multivariate aspect of forecast distributions by using historical observation trajectories as multivariate “dependence templates.” This paper proposes a variant of the Schaake shuffle, in which the historical dates are selected such that the marginal distributions of the corresponding observation trajectories are similar to the forecast marginal distributions, thus making it more likely that spatial and temporal gradients are preserved during the reordering procedure. This new approach is demonstrated with temperature and precipitation forecasts over four river basins in California, and it is shown to improve upon the standard Schaake shuffle both with respect to verification metrics applied to the forcings, and verification metrics applied to the resulting streamflow predictions.

This study verifies the skill and reliability of ensemble water supply forecasts issued by an innovative operational Hydrologic Ensemble Forecast Service (HEFS) of the U.S. National Weather Service (NWS) at eight Sierra Nevada watersheds in the State of California. The factors potentially influencing the forecast skill and reliability are also explored. Retrospective ensemble forecasts of April-July runoff with 60 traces for these watersheds from 1985 to 2010 are generated with the HEFS driven by raw precipitation and temperature reforecasts from operational Global Ensemble Forecast System (GEFS) for the first 15 days and climatology from day 16 up to day 365. Results indicate that the forecast skill is limited when the lead time is long (over three months or before January) but increases through the forecast period. There is generally a negative bias in the most probable forecast (median forecast) for most study watersheds. When the mean forecast is investigated instead, the bias becomes mostly positive and generally smaller in magnitude. The forecasts, particularly the wet forecasts (with less than 10% exceedance probability) are reliable on the average. The low April-July flows (with higher than 90% exceedance probability) are forecast more frequently than their actual occurrence frequency, while the medium April-July flows (90% to 10% exceedance) are forecast to occur less frequently. The forecast skill and reliability tend to be sensitive to extreme conditions. Particularly, the wet extremes show more significant impact than the dry extremes. Using different forcing data, including pure climatology and Climate Forecast System version 2 (CFSv2) shows no consistent improvement in the forecast skill and reliability, neither does using a longer (than the study period 1985-2010) period of record. Overall, this study is meaningful in the context of (1) establishing a benchmark for future enhancements (i.e., newer version of HEFS, GEFS and CFSv2) to ensemble water supply forecasting systems and (2) providing critical information (on what skill and reliability to expect at a given lead time, water year type and location) to water resources managers in making uncertainty-informed decisions in maximizing the reliability of the water supply.

This study presents a comprehensive trend analysis of precipitation, temperature, and runoff extremes in the Central Valley of California from an operational perspective. California is prone to those extremes of which any changes could have long-lasting adverse impacts on the society, economy, and environment of the State. Available long-term operational datasets of 176 forecasting basins in six forecasting groups and inflow to 12 major water supply reservoirs are employed. A suite of nine precipitation indices and nine temperature indices derived from historical (water year 1949-2010) six-hourly precipitation and temperature data for these basins are investigated, along with nine indices based on daily unimpaired inflow to those 12 reservoirs in a slightly shorter period. Those indices include daily maximum precipitation, temperature, runoff, snowmelt, and others that are critical in informing decision making in water resources management. The non-parametric Mann-Kendall trend test is applied with a trend-free pre-whitening procedure in identifying trends in these indices. Changes in empirical probability distributions of individual study indices in two equal sub-periods are also investigated. The results show decreasing number of cold nights, increasing number of warm nights, increasing maximum temperature, and increasing annual mean minimum temperature at about 60% of the study area. Changes in cold extremes are generally more pronounced than their counterparts in warm extremes, contributing to decreasing diurnal temperature ranges. In general, the driest and coldest Tulare forecasting group observes the most consistent changes among all six groups. Analysis of probability distributions of temperature indices in two sub-periods yields similar results. In contrast, changes in precipitation extremes are less consistent spatially and less significant in terms of change rate. Only four indices exhibit statistically significant changes in less than 10% of the study area. On the regional scale, only the American forecasting group shows significant decreasing trends in two indices including maximum six-hourly precipitation and simple daily intensity index. On the other hand, runoff exhibits strong resilience to the changes noticed in temperature and precipitation extremes. Only the most southern reservoir (Lake Isabella) shows significant earlier peak timing of snowmelt. Additional analysis on runoff indices using different trend analysis methods and different analysis periods also indicates limited changes in these runoff indices. Overall, these findings are meaningful in guiding reservoir operations and water resources planning and management practices. IntroductionClimatic and weather-induced hazards including excessive heat, flooding, and drought are often economically, environmentally, and societally disruptive [1][2][3]. Previous studies have suggested that such hazards are typically caused by changes in the frequency and intensity rather than the mean of hydro-climatic variables including precipita...

In the post-processing 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 covariabilty in space, time, and across variables, via so-called ”reordering” methods. Within this framework though, post-processors 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 (NWS)’s Hydrologic Ensemble Forecast Service (HEFS), although users were left with material in the grey 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 multi-day 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.