Hybrid hydroclimatic forecasting systems employ data-driven (statistical or machine learning) methods to harness and integrate a broad variety of predictions from dynamical, physics-based models – such as numerical weather prediction, climate, land, hydrology, and Earth system models – into a final prediction product. They are recognized as a promising way of enhancing the prediction skill of meteorological and hydroclimatic variables and events, including rainfall, temperature, streamflow, floods, droughts, tropical cyclones, or atmospheric rivers. Hybrid forecasting methods are now receiving growing attention due to advances in weather and climate prediction systems at subseasonal to decadal scales, a better appreciation of the strengths of AI, and expanding access to computational resources and methods. Such systems are attractive because they may avoid the need to run a computationally expensive offline land model, can minimize the effect of biases that exist within dynamical outputs, benefit from the strengths of machine learning, and can learn from large datasets, while combining different sources of predictability with varying time horizons. Here we review recent developments in hybrid hydroclimatic forecasting and outline key challenges and opportunities for further research. These include obtaining physically explainable results, assimilating human influences from novel data sources, integrating new ensemble techniques to improve predictive skill, creating seamless prediction schemes that merge short to long lead times, incorporating initial land surface and ocean/ice conditions, acknowledging spatial variability in landscape and atmospheric forcing, and increasing the operational uptake of hybrid prediction schemes.

Flooding in urban areas represents a particular challenge to modellers and flood risk managers because of the complex interactions of surface and sewer flows. Quantified flood risk estimates provide a common metric that can be used to compare risks from different sources. In situations where there are several organisations responsible for flood risk management we wish to be able to disaggregate the total risk and attribute it to different components in the system and/or agents with responsibility for risk reduction in order to target management actions. Two approaches to risk attribution are discussed: Standards-based attribution, which is a deterministic approach, based upon the performance of different engineering components in the system at their “design standard”. Sensitivity-based attribution, which apportions risk between the variables that influence the total flood risk. Whilst both these approaches are feasible for the small system considered here, in practice urban flooding systems involve tens of thousands of variables. The only feasible approach to tackling this problem for large urban systems is therefore by hierarchical simplification of the system, with the attribution analysis being applied in several tiers of detail. In this paper, the applicability of a hierarchical approach is demonstrated in the context of sewer pipe blockages. The results demonstrate the potential of attribution methods to support the development of integrated urban flood risk management strategies, as they can identify the forcing variables and infrastructure components that have the most influence upon flood risk.

Surface water (or pluvial) flooding is caused by intense rainfall before it enters rivers or drainage systems. As the climate changes and urban populations grow, the number of people around the world at risk of surface water flooding is increasing. Although it may not be possible to prevent such flooding, reliable and timely flood forecasts can help improve preparedness and recovery. Unlike riverine and coastal flooding where forecasting methods are well established, surface water flood forecasting presents a unique challenge due to the high uncertainties around predicting the location, timing, and impact of what are typically localized events. Over the past 5 years, there has been rapid development of convection‐permitting numerical weather prediction models, ensemble forecasting, and computational ability. It is now theoretically feasible to develop operational surface water forecasting systems. This paper identifies three approaches to surface water forecasting utilizing state‐of‐the‐art meteorological forecasts: empirical‐based scenarios, hydrological forecasts linked to presimulated impact scenarios, and real‐time hydrodynamic simulation. Reviewing operational examples of each approach provides an opportunity to learn from international best practice to develop targeted, impact‐based, surface water forecasts to support informed decision‐making. Although the emergence of new meteorological and hydrological forecasting capabilities is promising, there remains a scientific limit to the predictability of convective rainfall. To overcome this challenge, we suggest that a rethink of the established role of flood forecasting is needed, alongside the development of interdisciplinary solutions for communicating uncertainty and making the best use of all available data to increase preparedness.This article is categorized under: Engineering Water > Engineering Water

Recent surface-water and flash floods have caused millions of pounds worth of damage in the UK. These events form rapidly and are difficult to predict due to their short-lived and localised nature. The interdisciplinary Flooding From Intense Rainfall (FFIR) programme investigated the feasibility of enhancing the integration of an end-to-end forecasting system for flash and surface-water floods to help increase the lead time for warnings for these events. Here we propose developments to the integration of an operational end-to-end forecasting system based on the findings of the FFIR programme. The suggested developments include methods to improve radar-derived rainfall rates and understanding of the uncertainty in the position of intense rainfall in weather forecasts; the addition of hydraulic modelling components; and novel education techniques to help lead to effective dissemination of flood warnings. We make recommendations for future advances such as research into the propagation of uncertainty throughout the forecast chain. We further propose the creation of closer bonds to the end users to allow for an improved, integrated, end-to-end forecasting system that is easily accessible for users and end users alike, and will ultimately help mitigate the impacts of flooding from intense rainfall by informed and timely action.

Scientific and technical changes to flood forecasting models are implemented to improve forecasts. However, responses to such changes are complex, particularly in global models, and evaluation of improvements remains focussed on generalised skill assessments and not on the most relevant outcomes for those taking decisions. Recently, the Global Flood Awareness System (GloFAS) flood forecasting model has been upgraded from version 2.1 to 3.1 with a significant change to its hydrological model structure. In the updated version 3.1, a single fully configured hydrological model (LISFLOOD) has been adopted, including ground water and river routing processes, instead of two coupled models, a land surface and a simplified hydrological model, of the previous version 2.1. This study aims to evaluate changes in the simulated behaviour of floods and the forecast skill of the two GloFAS versions based on different decision criteria for early action. We evaluate GloFAS reforecasts for the Brahmaputra and the Ganges Rivers in Bangladesh for the period 1999–2018. For the Brahmaputra River, the old GloFAS 2.1 version performs better than the 3.1 version, especially in predicting low‐ (90th percentile) and medium‐level (95th percentile) floods. For the Ganges, GloFAS 3.1 shows improved probability of detection of low‐ to medium‐level floods compared to version 2.1, especially for lead times longer than 10 days. Both versions show limited skill for more extreme floods (99th percentile) but results are less robust for these less frequent floods given the lower number of events. Using lead‐time dependent thresholds improves the false alarm ratio while reducing the probability of detection. The changes in model structures influence the model performance in a complex and varied way and forecast skill needs further investigation across regions and decision‐making criteria. Understanding the skill changes between different model versions is important for decision‐makers; however, focused case studies such as this should also be used by model developers to guide future changes to the system to ensure that they lead to improvements in decision‐making ability.