Floods and droughts develop over different timeframes, from fast developing events to slowly evolving processes, but all require timely decisions for effective preparedness. SEED-FD is validating new forecasting products to support better-informed decisions and strengthen preparedness worldwide. The prototypes, developed by scientists from Politecnico di Milano and ECMWF, include forecasts for flash flood risk, flash drought conditions, as well as seasonal hydrological and agricultural drought.
Flash Flood Risk Forecasts: Identifying Rainfall-Driven Flood Risks
Flash floods can develop within hours when intense rainfall hits a vulnerable area. They often strike communities unexpectedly, making flash floods a dangerous weather-related hazard for people and their livelihoods. Many areas do not have early warning systems for flash flooding and currently CEMS-Floods GloFAS does not include flash flooding in its global forecast product offering.
To address this challenge, SEED-FD has developed a prototype global flash flood risk forecast product. It combines a precipitation hazard forecast with dynamic flash flood susceptibility in an impact matrix. In simple terms, the product not only considers where heavy rainfall may occur, but also where this rainfall could lead to a risk of severe impacts.
Flash flood susceptibility is assessed through four key factors: how quickly rivers respond to rainfall, soil and hydrological conditions before the event, population exposure, and each country’s capacity to cope with natural hazards. By bringing these elements together, the new forecasting product can provide valuable information for preparedness and risk reduction.
Flash Flood Case Studies: Testing Global Applicability
During the SEED-FD validation phase, the prototype global flash flood risk product has been tested in several case studies. These case studies examined the product’s ability to predict past flash flood events and its applicability across different regions of the world.
The results showed a good agreement between reported events and the prototype product. For example, on 23 April 2026 authorities in Bangladesh issued flash flood warnings for the north east of the country, however no signal of any flooding was forecasted in this area by CEMS-Floods GloFAS. The SEED-FD prototype flash flood risk product highlights the potential for severe flash flood risk in this area up to 5-days ahead. For another flash flood near Nairobi on 6 March 2026, the prototype product highlighted the risk in this area two days before the event.
Figure 1: Case study of the prototype SEED-FD flash flood risk forecast for the Bangladesh and India flood event in April 2026, showing forecast signals for possible flash flooding up to five days in advance across different lead times.
Flash Drought Forecasts: Anticipating Rapidly Intensifying Droughts
Some drought events intensify rapidly over short periods; these are known as flash droughts. Their sudden onset makes them difficult to manage in operational drought monitoring and early warning systems, impacting especially in agricultural production.
SEED-FD has developed a prototype forecast product specifically for flash droughts, aiming to anticipate their onset. A key element of this work is the IDR3 index. It represents how quickly droughts intensify during their first 30 days. Based on extensive soil moisture data analysis, the index reliably distinguishes between slow-evolving and fast-intensifying droughts, with flash events characterized by higher values of the index. This allows flash droughts to be identified globally at an early stage, before they reach full intensity.
Flash Drought Case Studies: Identifying Global Hotspots
The prototype SEED-FD flash drought forecasting product has been tested during validation through an implementation of the IDR3 index into GloFAS, the Global Flood Awareness System of the Copernicus Emergency Management Service. The analysis produced spatial patterns showing how frequently flash drought events occur in different parts of the world.
The results highlighted areas with a high frequency of flash droughts, including Brazil, central India, and Mexico. These patterns are consistent with findings described in the available scientific literature. This agreement supports the relevance of the prototype product for global drought monitoring and future early warning applications.
Figure 2: Case studies of the prototype flash drought forecast product, showing global patterns in the frequency of flash and slow drought events based on the IDR₃ index between 1991 and 2023.
Seasonal Hydrological Drought Forecasts: River Water Availability
Drought is often associated with a lack of rainfall, but the most damaging impacts usually appear later. They occur when water stored in soil, rivers, and groundwater becomes scarce. These hydrological droughts are the stage at which water restrictions, agricultural losses, energy supply issues, and ecosystem impacts can become unavoidable.
SEED-FD has developed a prototype seasonal hydrological drought forecast product that can identify such events up to six months in advance. It is based on the Standardized Streamflow Index (SSI), which measures how strongly river discharge differs from long-term normal conditions. The product also enables an assessment of forecast skill using the signal-to-noise ratio (SNR).
Seasonal Hydrological Drought Case Studies: Capturing Past Events
The prototype seasonal hydrological drought forecast product has also been tested during SEED-FD’s validation phase. Global simulations show good agreement between the forecast and reanalysis data, which serves as a reference for actual conditions.
Figure 3 shows, for example, a case study covering the period from April to September 2010. During this period, a mega-heatwave in Russia led to severe drought conditions, widespread wildfires, and significant impacts on food security. The prototype forecast product successfully identified this event up to six months in advance, demonstrating its potential to support earlier preparedness and more informed decision-making in hydrological drought risk management.
Figure 3: Case study of the prototype seasonal hydrological drought forecast for the 2010 Russia heatwave, showing good agreement between forecast and reanalysis data for drought conditions.
Agricultural Drought Forecasts Based on Tracking
For agriculture, soil moisture is crucial. It influences crop growth, irrigation needs, and ultimately yield. Effective decision-making requires information that supports both short-term action and planning over several months.
SEED-FD has developed a prototype agricultural drought forecast product with a seasonal forecast timeframe. It builds on an advanced three-dimensional drought tracking algorithm, which identifies soil moisture droughts by clustering information across latitude, longitude, and time.
By tracking past and forecast soil moisture conditions, the product can provide indications of possible agricultural drought development up to six months into the future. This supports more informed water and crop management in the agricultural sector.
Agricultural Drought Case Studies: Identifying Past Events
The prototype agricultural drought forecast product has been tested in case studies covering selected periods with well-documented major drought events and recorded impacts. Global simulations show a good match between the forecast and the reanalysis data used as a reference for actual drought conditions.
Figure 4 shows examples of well-known droughts that were successfully identified by the developed forecast. These include the European drought in 2003, the central US drought in 2012, the drought in China in 2011, and the Amazon basin drought in 1997.
Figure 4: Case studies of the prototype SEED-FD agricultural drought forecast product, showing major historical drought events successfully identified.
Better Preparedness for Floods and Droughts
SEED-FD’s new forecasting products have been developed to support decision-makers involved in water management, agriculture, civil protection, and natural hazard risk management. By providing forecasts for events that are often not yet sufficiently covered by existing systems, the products can contribute to better preparedness. They help identify where risks may emerge, how severe impacts may become, and how reliable the forecast information is. This can support more informed decisions on resource allocation, emergency planning, and water and agricultural management strategies.
Outlook: Validation, Fine-Tuning, and Next Steps
As part of SEED-FD’s final project year, extensive validation and fine-tuning of the new forecasting products will continue in order to further improve their performance and prepare them for future operational use.
The prototypes are being developed with potential integration into the Copernicus Emergency Management Service in mind. At the same time, they can be adopted independently in other forecasting and early warning systems, depending on user needs and operational contexts.
Further information will be shared in the coming months, so stay tuned. For the latest updates on SEED-FD’s results and case studies, follow SEED-FD on LinkedIn, X, and Facebook.