Operational Daily Streamflow Forecasts by coupling Variational Ensemble Forecasting and Machine Learning (VEF-ML) approaches

<p>The operational implementation of a Hydrologic Forecasting System (HFS) is limited in many catchments of the world by the lack of historical in-situ hydrologic data, i.e., long temporal records of rainfall or streamflow. By combining high-resolution Satellite Precipitation Products (SPPs), or Regional Climatological Models (RCMs), with Hydrologic Models, baselines can be established for the quantification and reduction of total hydrologic uncertainty in ungauged basins. We have studied how Variational Ensemble Forecasting (VEF) can be combined with Machine Learning (ML) techniques to improve a hydrologic system representation – i.e., raw data processing, model training, model evaluation, model selection, forecasts post-processing, etc. The VEF-ML method is applied and assessed with three general Hydrologic Processing Hypotheses (HPH): (1) Hydrologic Pre-processing (HPR), (2) Hydrologic Processing (HP), and (3) Hydrologic Post-processing (HPP). The operational implementation of VEF-ML was evaluated in the Upper Zambezi River Basin (UZRB) and its sub-basins, by using multiple precipitation products, multiple hydrologic models, and multiple optimal parameter sets. This extended VEF configuration and its coupling with ML techniques (VEF-ML) allows increasing the number of hydrologic ensembles available for the generation of operational streamflow forecasts products. The performance of VEF-ML is evaluated by comparing two hydrologic learning strategies (HLS) i.e. inference- and pattern-based approaches, which are used to improve hydrologic post-processing hypotheses (i.e. reduce total hydrologic uncertainty) in the poorly gauged UZRB.</p>

A crash-testing framework for predictive uncertainty assessment when forecasting high flows in an extrapolation context

An increasing number of flood forecasting services assess and communicate the uncertainty associated with their forecasts. While obtaining reliable forecasts is a key issue, it is a challenging task, especially when forecasting high flows in an extrapolation context, i.e. when the event magnitude is larger than what was observed before. In this study, we present a crash-testing framework that evaluates the quality of hydrological forecasts in an extrapolation context. The experiment set-up is based on (i) a large set of catchments in France, (ii) the GRP rainfall–runoff model designed for flood forecasting and used by the French operational services and (iii) an empirical hydrologic uncertainty processor designed to estimate conditional predictive uncertainty from the hydrological model residuals. The variants of the uncertainty processor used in this study differ in the data transformation they use (log, Box–Cox and log–sinh) to account for heteroscedasticity and the evolution of the other properties of the predictive distribution with the discharge magnitude. Different data subsets were selected based on a preliminary event selection. Various aspects of the probabilistic performance of the variants of the hydrologic uncertainty processor, reliability, sharpness and overall quality were evaluated. Overall, the results highlight the challenge of uncertainty quantification when forecasting high flows. They show a significant drop in reliability when forecasting high flows in an extrapolation context and considerable variability among catchments and across lead times. The increase in statistical treatment complexity did not result in significant improvement, which suggests that a parsimonious and easily understandable data transformation such as the log transformation or the Box–Cox transformation can be a reasonable choice for flood forecasting.

Is hydraulic modelling parametrization the major source of variability in flood hazard assessment? Insight into hydrologic uncertainty and the role of design rainfall in probabilistic flood maps

<p>In this work, a methodology for quantifying the relative impact of hydrological and hydraulic modelling parameterizations on uncertainty of inundation maps has been developed and applied in the Marta river basin, central Italy. A lumped rainfall-runoff forced by a synthetic hyetograph derived from regionalized IDF curves and a Quasi-2D hydraulic model were adopted to delineate the flood hazard maps related to different return periods. The uncertainty related to the design rainfall estimation method, given by the limited length of the time series from which the IDF curves fitted, was considered adopting a Monte Carlo approach.  On the other hand, the uncertainty related to floodplain roughness was considered adopting literature values. The above mentioned methodologies for representing both uncertainties were applied simultaneously and separately. Results, expressed in terms of variability of simulated flood extents and flow depths, suggest a significant predominance of the uncertainty related to hydrological modelling as respect to the hydraulic modelling.</p>

Evaluating joint operation rules for connecting tunnels between two multipurpose dams

This study aims to provide a practically efficient approach for determining the most efficient joint operation rule for two reservoirs connected by a waterway tunnel. For this purpose, the connecting tunnel's effect was assessed and three heuristic joint operation rules accounting for the connecting tunnel were evaluated. A standard operation policy with the connecting tunnel led to positive effects on the water resource system of the target basin with regard to a reliable water supply. The connecting tunnel provides an additional water supply of 12.4 million m3/year to the basin, and the reliability of the two reservoirs increased. Among the three rules, the equivalent reservoir (ER) rule led to the most positive effect on water supply. We found that the ER rule could maximize the positive effects of the connecting tunnel by maintaining the effective water storage rates of the two reservoirs. Moreover, the effects of hydrologic uncertainty on the joint operation rules were discussed using the synthetically generated multiple streamflow traces.

Probabilistic Flood Forecasting Using Hydrologic Uncertainty Processor with Ensemble Weather Forecasts

Recent advances in the field of flood forecasting have shown increased interests in probabilistic forecasting as it provides not only the point forecast but also the assessment of associated uncertainty. Here, an investigation of a hydrologic uncertainty processor (HUP) as a postprocessor of ensemble forecasts to generate probabilistic flood forecasts is presented. The main purpose is to quantify dominant uncertainties and enhance flood forecast reliability. HUP is based on Bayes’s theorem and designed to capture hydrologic uncertainty. Ensemble forecasts are forced by ensemble weather forecasts from the Global Ensemble Prediction System (GEPS) that are inherently uncertain, and the input uncertainty propagates through the model chain and integrates with hydrologic uncertainty in HUP. The bias of GEPS was removed using multivariate bias correction, and several scenarios were developed by different combinations of GEPS with HUP. The performance of different forecast horizons for these scenarios was compared using multifaceted evaluation metrics. Results show that HUP is able to improve the performance for both short- and medium-range forecasts; the improvement is significant for short lead times and becomes less obvious with increasing lead time. Overall, the performances for short-range forecasts when using HUP are promising, and the most satisfactory result for the short range is obtained by applying bias correction to each ensemble member plus applying the HUP postprocessor.

A crash-testing framework for predictive uncertainty assessment when forecasting high flows in an extrapolation context

An increasing number of flood forecasting services assess and communicate the uncertainty associated with their forecasts. While obtaining reliable forecasts is a key issue, it is a challenging task, especially when forecasting high flows in an extrapolation context, i.e. when the event magnitude is larger than what was observed before. In this study, we present a crash-testing framework that evaluates the quality of hydrological forecasts in an extrapolation context. The experiment setup is based on (i) a large set of catchments in France, (ii) the GRP rainfall-runoff model designed for flood forecasting and used by the French operational services and (iii) an empirical hydrologic uncertainty processor designed to estimate conditional predictive uncertainty from the hydrological model residuals. The variants of the uncertainty processor used in this study differ in the data transformation they used (log, Box–Cox and log–sinh) to account for heteroscedasticity. Different data subsets were selected based on a preliminary event selection. Various aspects of the probabilistic performance of the variants of the hydrologic uncertainty processor, reliability, sharpness and overall quality, were evaluated. Overall, the results highlight the challenge of uncertainty quantification when forecasting high flows. They show a significant drop in reliability when forecasting high flows in an extrapolation context and considerable variability among catchments and across lead times. The increase in statistical treatment complexity did not result in significant improvement, which suggests that a parsimonious and easily understandable data transformation such as the log transformation or the Box–Cox transformation with a parameter between 0.1 and 0.3 can be a reasonable choice for flood forecasting.