Purpose and objectives


The purpose of this project is to develop a methodology for retrieving, processing and tailoring intense precipitation forecasts, primarily aimed at hydrologic and hydraulic assessment of e.g. flood risk. Requirements of the forecasts are a high resolution in both time (1 hour) and space (potentially below 10 km) and flexibility to enable usage in a variety of applications that need high-resolution forecasts.

Ultimately, the developed methodology is envisioned to form the basis of a warning system which:

• provides hourly probability forecasts for intense precipitation at a high spatial resolution;

• delivers warnings, information and data through a warning service web portal, including specialized web services related to specific end users’ needs and subscriptions (e.g. RSS feeds, risk maps etc.);

• is linked to hydrological forecasting models, with specific focus on e.g.
     1. high water levels in flood risk zones,
     2. flash floods in areas with thin soils and/or quick hydrologic response,
     3. urban flooding and inundation.


The main objective of the project is to assess the applicability of high-resolution precipitation data for improved knowledge of, preparedness for and cost-analysis of intense precipitation and mainly its hydrological consequences in terms of e.g. floods and debris flow.

Questions to be answered include:

• Knowledge: How can different types of sensors (e.g. radar and automatic gauges) be combined for optimal description of intense rainfall fields and their variability in time and space? What are the typical characteristics of intense rainfall fields in Sweden with respect to extension, duration and development over time? Is there any geographical dependence of the most intense rainfall events?

• Preparedness: With which accuracy can today’s forecasting system predict intense rainfall events? How can different observational data sources and forecast products be combined for different hydrological applications? Which is the potential added value of higher-resolution flood forecasts as compared with today’s 24 hr-based system?

• Cost analysis: Can costs for damages related to intense precipitation be meaningfully described in terms of damage functions with selected explanatory factors? Can rainfall and flood forecasts be used in a more efficient way to increase the benefit of warnings in specific applications and situations? Can planning and decision support at e.g. municipal level be improved by tools for risk assessment coupled to intense rainfall?

As a basis for the analyses, a national database will be compiled comprising historical high-reolution observations from different types of sensors as well as high-resolution precipitation forecasts from different models and institutes. The observations will be analysed and coupled to cases of natural disasters and rainfall-induced damages. The forecasts will be evaluated against observations and methods will be developed to combine different types of forecasts into optimal products for subsequent use in hydrological forecasting.

The project will be organised in four work packages (WPs), where WP1 will focus on the different precipitation products, their evaluation and integration into primary forecasts. In WP2, tools for post-processing primary forecasts into e.g. probability forecasts and distributing them will be developed. In WP3, forecasts will be tested and evaluated in a number of relevant case studies. WP4 will assess economical consequences in terms of damage functions and develop an economical tool.

Forecast tools

SMHI is currently not issuing warnings for local flooding where there are no observations or models available. Warnings of intense precipitation are issued in terms of Class 1 (Heavy rain) and Class 2 (Very large rainfall amounts).

These warnings cover large areas and do not provide information with enough temporal and spatial resolution to be of direct use for hydrological risk assessment. The ambition of SMHI is, however, to provide warnings of intense rainfall at higher resolution in time and space.

New methods of short-term forecasts are constantly being developed; methods that can be utilized for this purpose. SMHI has recently launched a system for short-term precipitation forecasts (KNEP; Ridal et al., 2011), a ‘now-casting’ approach using weather radar in combination with weather projections from the HIRLAM atmospheric model. KNEP delivers hourly precipitation forecast for the coming 36 hours with 11×11 km resolution nationwide.

Weather radars provide rainfall estimates with an excellent resolution in time (15 min) and space (2×2 km), but are associated with large errors. They need to be adjusted by using ground observations or a mesoscale analysis of the observational field, for example the Swedish system MESAN or an interpolated gridded precipitation field like the PTHBV. Both the PTHBV and the MESAN are part of SMHIs operational system.

Another new forecast tool is GLAMEPS (Grand Limited Area Model Ensemble Prediction System; Iversen et al., 2011), which is a multi-model forecast ensemble with high resolution (13 km). GLAMEPS uses different versions of the meteorological forecast models HIRLAM and ALADIN to create an ensemble of 52 forecasts and will become operational in 2011.

On a lower resolution (25 km) there are also global ensemble forecasts from the European Centre (ECMWF). See Table 1 for an overview of existing products in Sweden. Similar systems are operationally tested at different locations in Europe, for example the Extreme Rainfall Alert (ERA; Hurford et al., 2011) and the European Precipitation Index based on simulated Climatology (EPIC; Alfieri et al., 2011). Furthermore, the ECMWF Extreme Forecast Index (EFI; Lalaurette, 2003) can be used as an indicator of extreme rainfall.