WMO Regional Association VI
Working Group on Climate and Hydrology (WG -CH)
Task Team on Flood Forecasting and Warning (TT /FFW)
Progress report nr 2 (October 2011)
1
General issues

Documentation from TT FFW is now available on the RAVI web site.
http://www.wmo.int/pages/prog/dra/eur/RA6_WG_CH_WrkProg.php

The WMO Congress XVI was held in Geneva from 16 May to 3 June 2011.
Hydrology was considered under many different agenda items (e.g. GFCS – the
Global Framework for Climate Services). The Hydrology and Water Resources
Programme (HWRP) was agenda item 3.4 and the approved document can be
found on ftp://ftp.wmo.int/Documents/SESSIONS/CgXVI/English/Approved%26Corrected/
The document 3.4 contains four resolutions, as follows:


Adoption of the Hydrology and Water Resources Programme (2012-2015)
as described in the document.

Establishing a Quality management framework – hydrology (QMF-H),
within the overall WMO QMF, covering data observations as well as
hydrological products and services in accordance with the recommendation
of CHy.

Urged all members to actively support the implementation of World
Hydrological Cycle Observing System (WHYCOS), both with manual
resources as well as financial.

To define the scope of the WMO Flood Forecasting Initiative (FFI) to
include all the hydrological forecasting activities such as those related to
flash floods, riverine flood including seasonal forecasts and coastal flooding
due to storm surges and to establish an Advisory Group (FFI-AG)
A Workshop on Intercomparison of Flood Forecasting Models - Decision making for
their Selection and Application was held from 14 to 16 September 2011 in Koblenz
Germany. The workshop was organized by the International Commission for the
Hydrology in the Rhine basin (CHR) in cooperation with the Federal Institute of
Hydrology in Germany. The workshop was in response to a recommendation made
during a workshop held in Geneva in December 2009 on the Strategy and Action
Plan of the WMO Flood Forecasting Initiative. Based on the exchange of
experiences with different flood forecasting models under vastly differing
environmental and institutional conditions, the primary output from the workshop
should be a decision-aiding matrix to help with the selection of appropriate flood
1
forecasting models under differing conditions and objectives. The report will
probably soon be found on the CHR web site. http://www.chr-khr.org/





2
2.1
The Strategic and Action Plan (SAP) of the WMO Flood Forecasting Initiative (FFI)
states that flood forecasting services in many parts of the world need to improve
their services. This situation requires the development of a framework for the
assessment of service delivery capabilities of hydrological services with regard to
flood forecasting. In particular, the methodology to be developed aims to provide
tools and methods on how a flood forecasting service could be assessed and
opportunities identified for its improvement. An expert meeting was held in Geneva
from 12 to 14 October 2011. The scope of the meeting was “Improving the
efficiency of flood forecasting services – Development of a framework for the
assessment of service delivery capabilities of hydrological services. The minutes
from the meeting will hopefully be available on the web soon. This initiative is also
important for RAVI, giving the opportunity to be assessed and given input for
improvement. This activity does not contribute directly to our FFW activities but may
give valuable input for improvements
5th International conference on flood management was held in Tokyo in September
2011. Declaration arisen from the conference is available on the web page:
(http://www.ifi-home.info/icfm-icharm/icfm5.html).
Participants agreed that:
There is an increased emphasis in modern flood management thinking on nonstructural methods, such as early warning as in many nations, many people have
no choice but to live in flood prone areas. Continued economic development,
therefore, requires a rational balance of structural and non-structural infrastructure.
Scientific Advancement of Prediction. Floods are largely predictable; disasters of
hydro-meteorological origin can be far better managed through the applied use of
science and technology. Research and knowledge-based decision making should
be greatly promoted.
A PhD report on a decision tool based on flood forecast for optimizing reservoir
operations in a view to reduce flood peaks has been published.
http://library.epfl.ch/theses/?au=&ti=&sc=&yr=&nr=5093&adv=&access=&send.x=0
&send.y=0
Report of the Flood forecasting project, concerning the use of satellite areal snow
water equivalents in operational forecasting models. The report can be found on:
http://floodfore.fmi.fi/public_docs/FloodFore_report_final.pdf
A short description of the Hydrological Forecasting System (WSFS) in Finland can
be found in Annex 1.
Tasks/Actions in the Work Plan
Task: To look at the use of probabilities in both meteorological and
hydrological forecasts in order to improve the hydrological
forecast/warning, for all kinds of floods.
Ensemble predictions are implemented for global and regional NWP. For
example, Austrian meteorological service operates ALADIN-LAEF (16
members ensemble). Outputs are statistical post processed and then
input for testing in flood forecasting systems in the Czech
Hydrometeorological Institute (CHMI).
However meteorological probabilities in operation still consider only the
initial conditions as a source of uncertainty to produce ensemble
forecasts. Correspondingly hydrological forecast mostly consider only
meteorological inputs variants, not the other sources of uncertainty.
2.2
Task: To look at methods for estimating the uncertainties propagating
within the forecasting process from observation to hydrological
forecast/warning.
CHMI has performed evaluation of hydrological model forecasts for 20022010. It proved the dominance of QPF impact on forecast performance
(http://hydro.chmi.cz/hpps/index.php)
2.3
Task: To look at methods for verification.
Methods for evaluation of probabilistic forecast are well known (Brier
scores etc.). However there is not a single measure available for
deterministic forecasts. The text below is from a paper from Mr Jan
Danhelka from CHMI.
Nash-Sutcliffe criterion (eq. 1) is generally applied as a basic measure of model
performance in cases of long term simulation. But its application on single forecasted
hydrograph is not proper because of small dataset (in our case 48 values of hourly
forecasts) and change of leading processes and factors that affect the forecast error
with the lead time. It, as well, can not be used to evaluate more forecasts as they
usually overlap and/or made discontinuous time series. As a result there is no single
objective criteria generally accepted as a standard measure of real-time forecast
performance and eye control remains the most important evaluation technique.
However, as discussed during IAHS Workshop “Expert judgment vs. statistical
goodness-of-fit for hydrological model evaluation: Results of experiment“ in the frame
of XXV IUGG 2011 General Assembly, some pattern matching criteria provide
reasonable performance in evaluating high flow (see the report at http://www.stahy.org
). We especially point the attention to the procedure “Series Distance” proposed by
Ehret and Zehe (2011).
More
on
web
page
of
(http://hydro.chmi.cz/hpps/index.php ).
2.4
CHMI
evaluation
of
forecasts
Task: To look at methods for forecasting flash floods, including both
meteorological and hydrological input.
Several services operate system for evaluation of soil saturation which could be used
to estimate potential dangerous rainfall amount to cause significant runoff (e.g. CHMI
operates FFG-CZ and publish it on the web page: http://hydro.chmi.cz/hpps/index.php,
during the summer season). Such systems known as Flash flood guidance are basic
tool for real-time evaluation of flash flood risk. It could be also combined with simple
hydrological models to estimate run off, however it looks reasonable to transfer the
output to categorical explanation of risk to make it understandable for users.
2.5
Task: To look at methods for low flood forecasting
2.6
Task: To monitor the process of making EFAS operational
According to the GMES Regulation Work Programme 2011, EFAS is now a part of the
GMES Emergency Management Services (Early warning Systems).
An invitation to tender (ITT) was issued in June 2011 and last date for submitting a bid
was October 3.
The future EFAS will build upon four operational centres.
 Meteorological data collection
 Hydrological data collection
 The computational running of EFAS
 Dissemination of information to end users.
The above mentioned ITT covers the hydrological data collection, the computational
part and the dissemination.
The bids are evaluated by the Institute for Environment and Sustainability of the Joint
Research Centre (JRC) of the European Commission (EC), denominated IES. The
duration of the contract is one year after the date of signature and may be renewed
three times for an additional period of one year up to a total of four years maximum.
3
Outlook





The second meeting of the WG CH will be held in Yerevan, Armenia 14-16 of
November 2011.
Conference on warnings, forecasts and lake regulations January 20 2012 (
http://www.swv.ch/Weiterbildung/Veranstaltungen-SWV/KOHS-TagungenHochwasserschutz/KOHS-Tagung-2012 )
The Hydrology Forum is planned for 8-10 May 2012 in Koblenz.
The next Commission for Hydrology meeting is planned for November 2012.
Flood risk conference is planned to be held in Rotterdam in November 2012.
Annex 1
HYDROLOGICAL FORECASTING SYSTEM (WSFS) IN FINLAND
Bertel Vehviläinen
Finnish Environment Institute SYKE, PO Box 140, 00251 Helsinki, Finland
bertel.vehvilainen@ymparisto.fi
Abstract: WSFS (Watershed Simulation and Forecasting System) is widely used in
Finland for real time hydrological simulation and forecasting. WSFS development has
started from a rainfall-runoff model with the same basic structure as the HBV-model
widely used in Scandinavia. WSFS covers the land area of Finland including crossboundary watersheds, 390 000 km2. The distribution of the model is based on the third
level watershed division with 60 - 100 km2 sub-basins. In research version 1x1 km2 grid
is used. Main meteorological inputs are precipitation, temperature and cloudiness.
Simulated variables are areal precipitation, evapotranspiration, lake evaporation, snow,
soil moisture, surface, sub-surface and ground water flow and storage, runoff,
discharges and water levels of rivers and lakes. An elevation model is included to
simulate effects of elevation and slope gradient on areal precipitation, temperature,
snow accumulation and melt. Land-use, vegetation and soil type data are used to
simulate snow and to develop simulation of soil moisture, soil frost and
evapotranspiration. A nutrient (phosphorus, nitrogen and solid sediments) simulation
model covering also whole Finland is developed and operational based on hydrological
simulation model
Remote sensing data used in the system are satellite data of snow cover extent
and water equivalent and precipitation from weather radars. Assimilation methods for
the use of satellite snow data in WSFS have been developed for operational use.
Assimilation methods of flood area and soil moisture (SMOS) data from satellites are
under development. Weather radar has been in operational use since 1998 and is still
under development to increase the accuracy of weather radar areal precipitation
estimates. Weather radar gives more realistic rainfall distribution estimates than gauge
network. In summer, gauge measurements can be replaced by radar precipitation. In
winter, snow accumulation simulation using only radar data is not possible.
Automatic model updating system developed in SYKE is an important part of
the WSFS. Model state updating is done against water level, discharge, snow line and
satellite snow observations. The updating procedure corrects the model simulation by
changing the areal values of temperature, precipitation and potential evaporation (i.e.
finally the corresponding storage) so that the observed and simulated discharges,
water levels and snow values are as equal as possible. The WSFS obtains daily
meteorological data from 200 precipitation and 40 temperature stations, weather radar
precipitation over 300 000 km2. Updating of WSFS is done against 160 water level
stations, 130 discharge stations, snow covered area and water equivalent over Finland
from satellites and against 178 snow lines obtained once or twice per month.
Main operating part of WSFS is the distributed hydrological catchment model
over Finland, model updating software, data collection and forecast transfer software,
www-user interface for lake regulation control and data storage and www based
forecast distribution. WSFS is connected to hydrological registers and real-time
observation network, to synoptic weather station and weather radar network of the
Finnish Meteorological Institute (FMI). WSFS obtain weather forecasts (VAREPS,
monthly and seasonal EPS) from European Centre of Medium-Range Weather
Forecasts via FMI. In operation WSFS automatically collects meteorological and
hydrological data from the registers, runs hydrological forecasts and distributes
forecasts into the Internet www.environment.fi/waterforecast and registers for different
users.