ITSC 15th, 8-12 October 2006, Maratea, ItalyThe EUMETSAT Satellite Application Facility on support to

Operational Hydrology and Water Management (H-SAF)Luigi De Leonibus (HSAF Science Manager, USAM, Italy)

ABSTRACT

In recent years, the interest of the hydrological community for using satellite data has rapidly increased. This is a consequence of improved satellite data quality, and improved performance of hydrological models including their capability to assimilate observational data. As a consequence, scientific demonstration works and operationally-oriented initiatives are proliferating. One of these is the EUMETSAT Satellite Application Facility on support to operational hydrology and water management (H-SAF). It was established by the EUMETSAT Council in July 2005 and the Development phase started on 1st September 2005.

H-SAF membership includes 11 EUMETSAT member or cooperating States (Austria, Belgium, Finland, France, Germany, Hungary, Italy, Poland, Romania, Slovakia and Turkey) and ECMWF. The Italian Meteorological Service serves as Host Institute. In several Countries the activity is run by more institutes, of either operational or scientific nature. Partners of the Italian Meteorological Service are: the Department of Civil Protection (DPC) and the CNR Institute of Atmosphere and Climate Sciences (ISAC). Since the earliest discussions that finally led to the establishment of H-SAF, the interest for new satellite products focused on: • precipitation rate and cumulate precipitation, including liquid/solid discrimination; • soil moisture in the surface layer and possibly in the roots region; • snow parameters such as effective cover, wet/dry discrimination, water equivalent. The key element for the feasibility of generating these products with the required quality is the current or expected availability of highly-performing satellite instruments, such as: SEVIRI on Meteosat; AVHRR/3, AMSU-A and AMSU-B/MHS on Metop and NOAA; SSM/I and SSMIS on DMSP; ASCAT on Metop; and others, including some embarked on satellites of R&D nature such as AMSR-E and MODIS on EOS-Aqua, and TMI, PR and LIS on TRMM. Instruments will be used, that will have a long-term operational future either in their current configuration or as evolutions, e.g. VIIRS, CMIS and ATMS on NPOESS to replace AVHRR/3, MODIS, SSM/I, SSMIS and AMSU. The suite of currently available instruments and processing methods will allow early start of pre-operational product generation. Progressive availability of improved instruments and processing methods will enable progressive improvement of products quality in the course of the 5-year Development phase. While progressing with the products generation activity, a hydrological validation programme will have to demonstrate the cost effectiveness of the novel data so as to support the case for a follow-on Operational phase. The ingredients of the hydrological validation programme will be: • development of techniques to up/downscale the information for use at basin level; • merging satellite and conventional data, and assimilation in hydrological models; • assessment of the benefit of the new satellite data on the performance of operational hydrological models on the

basis of actual experimentation on selected test sites. Whereas the product generation activity will be performed by meteorological services supported by scientific institutes specialised in remote sensing, the hydrological validation programme will be performed by hydrometeorological services, hydrological scientific institutes and operational units of Civil Defence. Cooperation with other SAF’s is envisaged and although time de-phasing between H-SAF and others SAFs, efforts to tune some activities with SAF operations starting in 2007 will be pursued. HSAF has also strong interest in the activities of other international groups like as ITSC, which carries on studies on applications of atmospheric sounding and related issues, focusing mainly on generation of geophysical parameters with emphasis on the potential of MW water vapour absorbing channel in precipitation retrieval.

SATELLITE APPLICATION FACILITIES (SAF’s)

Eumetsat is the European agency for the management of operational meteorological satellites, it organises its application ground segments of all its satellites into two parts a centralized one in Darmstadt (Germany) at Eumetsat HQs premises and a decentralized one distributed among consortia held by its member states. These consortia are called Satellite Application Facilities (SAFs) which start with a development Phase , lasting 5 years, and which continue with the Operational Phase. Seven SAFs are already initiating their operations, the official start of their operations will take place for all them on march 2007. Since the 2005 a new SAF on support to Hydrology (HSAF) joined the former bunch of SAFs.

HSAF BACKGROUND IInn rreecceenntt yyeeaarrss,, tthhee iinntteerreesstt ooff tthhee hhyyddrroollooggiiccaall ccoommmmuunniittyy ffoorr uussiinngg ssaatteelllliittee ddaattaa hhaass rraappiiddllyyiinnccrreeaasseedd mmaaiinnllyy bbeeccaauussee ooff improved satellite data quality and improved performance ofhydrological models including their capability to assimilate observational data. Taking in consideration these elements, Eumetsat started some steps towards the establishment of a SAF in support to Hydrology.A specific “Working Group on a Potential SAF” was set and after two years (22000011--22000022)) iittccoonnssoolliiddaatteedd HHyyddrroollooggiiccaall eenndd uusseerr rreeqquuiirreemmeennttss aanndd iitt established ssoommee achievable objectives. During the 2003-2004 a “SAF Hydrology Framework Working Group” defined a scientific strategy and a long term vision proper for a SAF in support to Hydrology . In 2005 the Italian Air Force Meteorological Service set up a Consortium for the HSAF gathering ten national institutes and one international agency, the European Centre for Medium rangeWeather Forecast, the full list of the participants with their main roles is reported in fig.1. TheConsortium presented to Eumetsat a Proposal for the HSAF Development Phase (2005-2010) which was approved by Eumetsat Council on 3 July 2005. TThhee HHSSAAFF KKiicckk--OOffff mmeeeettiinngg wwaass hheelldd oonn 1155 SSeepptteemmbbeerr 22000055 iinn RRoommee aanndd tthhee aaccttiivviittiieess ssttaarrtteedd..TThhee ffiirrsstt rreevviieeww ttooookk ppllaaccee on 26-27 April 2006 on the requirements the next review on the preliminary design is going to take place on 11-12 December 2006. The first Workshop of HSAF is being prepared to take place in Rome on October 2007. The HSAF will join other SAFs in operations after 2010.

OBJECTIVES OF H-SAF

Starting form the HSAF theme which is to exploit meteorological satellites observations to support Hydrology, the objectives of H-SAF have been established taking in consideration Hydrologicalrequirement, with main emphasis for those ones coming from Numerical Hydrological modelling.So that main objectives are to provides products and services. As regards products, the HSAF will provide the following satellite-derived products: - precipitation (liquid, solid, rate, cumulate), see fig.2, - soil moisture (at surface, in the roots region), see fig.3, - snow parameters (cover, melting conditions, water equivalent), see fig.4As regard service the HSAF will perform independent validation with Hydrological models of the usefulness of the new products for hydrological applications. The programme will also carrying on of studies on new algorithms operational implementation.COMPOSITION OF THE H-SAF CONSORTIUM

The consortium is leaded by an host Country and it is broke down into four cluster according to the products and service. The Italian Air Force Meteorological Service, with The Italian Civil Protection Agency and the National Research Council lead the consortium, details regarding other participants and their roles are given in fig.1.

No. Country Main Unit in the Country Role 01 Austria Zentral Anstalt für Meteorologie und Geodynamik Leader for soil moisture 02 Belgium Royal Meteorological Institute of Belgium 03 ECMWF N/A Contributor for “core” soil moisture 04 Finland Ilmatieteen Laitos Leader for snow parameters 05 France Météo-France 06 Germany Bundesanstalt für Gewässerkunde 07 Hungary Hungarian Meteorological Service 08 Italy Servizio Meteorologico dell’Aeronautica Host + Leader for precipitation 09 Poland Institute of Meteorology and Water Management Leader for Hydrology 10 Romania National Institute for Meteorology and Hydrology 11 Slovakia Slovakia Hydro-Meteorological Institute 12 Turkey Turkish State Meteorological Service Contributor for “core” snow parameters

Fig.1.

SATELLITE PRECIPITATION PRODUCTS AND SATELLITE DATA SOURCES

The precipitation products are retrieved from satellite data according to information content of radiative channels: - precipitation rate by MW from conical scanning SSM/I and SSMIS; - precipitation rate by MW from cross-nadir scanning AMSU and MHS; - precipitation rate by blending MW from LEO and IR from GEO; - accumulated precipitation from LEO/MW or LEO/MW+GEO/IR. Products characteristics are reported in fig. 2. As regard calibration and validation it was considered that satellite retrieved products are instantaneous rate and their integration required for achieving any comparison with soil observation has inner error structure. The accumulated precipitation products are provided to face these problems.

Fig.2.

Products Cycle ranges from 6 hours for the polar orbiting satellite data up to 15 minutes for Geostationary satellite data. The 3 hours cycle for product derive from polar orbiting satellites has been set considering the avilabilty of NPOESS data by the end of the development phase (2010) Most challenging goal is to achieve fine spatial resolution up to 10 km, this is a very sensible requirement for hydrology because their numerical model run over very fine grid mesh, even less than 1 km. The prototype precipitation products generation chain is based on the direct reception of geostationary data (SEVIRI) and the real time redissemination of TOVS data performed by the Eumetsat broadcast (EUMETcast) which allow a full coverage of the HSAF area. For the DMSP data (SSMI/S) it has been set up a procedure for getting data from UK Meteorological Office with a delay of about 1 hours and half. The general scheme is reported in Fig.3.

Fig.3.

The prototype generation chain is divided into three chains for the instantaneous rain rate products, these chains converge to the chain of the cumulated precipitation product generation. The reference chain is that for the rain rate production from SSMI/S instrument MW data, because these channels (mainly the window ones) are contaminated only by the scattering produced by the hydrometeors. The retrieval is based on the collection of a cloud radiation data base which allow the retrieval (Bayesian) with the observed radiances. The scheme is given in fig.4. The WV absorption channels from METOP and NOAA satellites (AMSU and MHS instruments) with the highest resolution of AMSU channel give rise to the second MW product generation chain. Here the basic algorithm is statistic (neural network) and the main task is to perform the best calibration over the SAF area, see the scheme in fig.5. MW derived rain rats are the input for the high resolution precipitation generation chain which blends these products with the SEVIRI data to obtain rain rate every 10km and 15 minutes. The basic algorithm relays on space time regression which generate lookup tables for the rapid update. A technique which make use of dynamic information (morphing) is also under implementation. See fig.6.

Fig.4

The use the instantaneous products to compute the cumulated precipitation, require the assessment of their error structure which is performed using fine mesh ground network available within the HSAF area. An additional product offered out of the budget by HSAF host institute (Italy) is the Quantitative Precipitation Field (QFP) derived from the NWP model running operatively at the Host Institute itself. A study is undergoing to correct the QFP with the fine details smoothed by the NWP model assimilation and present in the satellite observations. The study will make use of AMSU data and IR and it is focused only on convective precipitation with an object approach (see ITSC15th lecture n. 11.7). For the generation of this product it will make use of the EuroLM (CNMCA operationalset-up of the non-hydrostatic regional model named Lokal Modell (LM) developed by theConsortium for Small-Scale Modelling (COSMO). A specific operational chain will produce QFP and derived fields: •accumulated precipitation •large-scale soil moisture (experimental)•snow water equivalent (experimental The objective is the improvement of analysed and forecast precipitation fields through the use ofmicrowave satellite observations in spectral ranges with proven sensitivity to precipitation andatmospheric humidity AMSUB, MHS and SSMI/S. Main activities are:•setting up of the RTTOV radiative transfer model and its interface with the EuroLM very highresolution NWP model; •treatment of microwave window channels’ radiances from available sensors, quality control and bias correction; •setting up of a variational (1D-VAR) retrieval algorithm in order to optimally combine theEuroLM first-guess hydrometeors’ profiles with the observed radiance information; •ingestion of retrieved profiles in EuroLM model (see ITSC15th lecture 7.5 by Vocino et al.).

Fig.5

Fig.6

SATELLITE SOIL MOISTURE PRODUCTS AND SATELLITE DATA SOURCES

The soil moisture products are retrieved from satellite data according to information content of active and passive MW radiometers embarked on some polar orbiting satellites as reported in fig.7:

Fig.7

Soil moisture is given into two products. The first product gives the surface soil moisture and it is retrieved directly from radar data (ASCAT). The second one is the soil moisture in the layer of ground from surface down to 1.69 meter and it is obtained from the 4Dvariational analysis of the ECMWF operational model, so that it depends strongly on the model parameterisation of ground. The time cycle of the product of 36 hours it was established in agreement to Hydrological requirements. The prototype precipitation products generation chain is based on the soil moisture surface global product, which will be operationally produced by the Central Facility of Eumetsat. This product will be the basic satellite observing input for both for the HSAF soil moisture processing chain at ZAMG to obtain the HSAF surface product (1km) and for the ECNWF model analysis to obtain the root product. The ZAMG shall generate the disaggregated surface product and it shall collect the root region product and it shall disseminate to the end users the soil moisture products. The Functional concept of the processing chain for soil moisture processing, at the end of the Development Phase, is reported in the scheme of fig.8.

Fig.8

The operational chain of the disaggregated product will benefit from the information of a scaling layer. This layer allows the interpretation of coarse resolution soil moisture information at 1 km resolution. The algorithm makes use of an additional data set, which for the prototype chain is mainly derived from the combination of the soil moisture trends observed in the coarse resolution for identified targets which have similar backscatter characteristics at both local (observed with the ENVISAT ASAR Global Mode sensor) and regional (observed with the MetOp ASCAT) scale. These additional data allow to compute soil moisture instantaneously at 1 km scale. The scheme of the chain is reported in fig.9. Several other data sources are under study to complete this auxiliary data set with additional quality information from various NRT datasets available for Europe like as snow (from H-SAF) and freeze/thaw (ECMWF, or also from H-SAF itself).

Fig.9

The prototype operational chain for the root zone soil moisture relays on the soil moisture climatology of the Integrated Forecast System of the ECMWF model derived from SCAT instruments and the real line ASCAT observations. Both are ingested by the Surface Data Assimilation System which will produce the soil moisture root zone product. The scheme of the generation chain is given at fig.10.

Fig.10

SATELLITE SNOW PRODUCTS AND SATELLITE DATA SOURCES The HSAF snow products are divided into two sets the flat terrain products and mountainous ones. Both the two sets relay on optical channels for the detection and they use different algorithm with optical and MW channels for the status and the water equivalence of the snow. The table of all products is reported in fig.11.

Fig.11

Snow products are based mainly on infrared and visible data for coverage and mainly on MW data for status and water equivalent. Furthermore operational product generation is strongly dependent on surface auxiliary data, this means that in the early prototype two separate products generation chain are kept, one for flat lands (fig. 12) and one for mountains areas (fig.13).

Fig. 12

Fig. 13

Along the programme this separation will be overcame and a unique product for both terrain types will be generated in accordance to one functional concept of the processing chain for soil moisture processing, the final scheme which the Consortium aims to achieve at the end of the Development Phase, is reported in fig.14.

Fig.14

The snow products generation chain for flat land make use of optical bands from SEVIRI and MODIS instruments for snow detection, it is a mask obtained with multispectral threshold parameter technique. The effective snow cover is computed from polar orbiting satellite data in optical band. The various area overages are assembled in a mosaic to obtain the full HSAF area coverage. During the effective snow cover processing, the snow status from MW data ( form radar instrument like as AMSR-E and ASCAT) is computed as auxiliary data and it is given like as a flag to the coverage product. Snow water equivalent product is obtained form the Tb brightness inversion of a snow pack emission model. The more sensitive MW channels (brightness temperature), which are affected by snow pack characteristics of snow water equivalent and grain, are at 19 and 37 GHz. The various steps of the algorithms extract the desired snow water information. The functional concept of the generation for flat and forest land of these products with their interconnections are reported in the scheme of fig. 15.

Fig. 15

For mountains area there is a dedicate generation chain which take in accounts terrain elevation and inclination with the their effect on reflectivity and shadows. The prototype algorithm relays on a basic technique for computing the Bi-directional Reflectance Distribution Function which makes use of a linear physically based model in which a mixed spectrum is modelled as a combination of pure spectra. The functional concept of the generation chain for mountain areas of the products with their interconnections are reported in the scheme of fig.16, main difference with the flat land processing chain is the absence of connection among MW and optical channel.

Fig.16

THE HYDROLOGICAL VALIDATION PROGRAMME The hydrological validation programme shall provide independent assessment of the impact of the new data on operational hydrology by means of Hydrological models. This cluster of the HSAF programme is very relevant of the SAF and it aims to set up and to maintain a bridge between meteorological satellite people and hydrology people which are very related but still use two specific languages. For the above reasons the basic task of the validation is the Hydrological requirements analysis which allow to establish accepted interfaces between HSAF products and Hydrological models. The result of that task are shown in fig.17, where HSAF products are enhanced (in orange, white and light blu), among other input data of Hydro model, with spatial and temporal resolution required.

Fig.17

Assimilation of satellite derived products by Hydrological model means to face the problem of developing methodologies algorithms for products in order to fit them to model spatial and time domain (catchment’s areas). These algorithms have to give solutions to up-scaling, down-scaling, averaging over catchments etc.. Other basic task is the integration of satellite products with in situ observations to improve the information on the specific input parameters. This methodology is called data fusion. The core of validation is the use of hydrological/hydrodynamic models for impact studies over sites where there is the availability of very valuable on ground observations. Various sites have been selected and the suite for the validation activity has been implemented. The sites are spread over the HSAF area to describe the specific regimes and scales which are present in the area itself; in the fig. 18.

Fig.18

The concept of validation activity is sketched in the fig.19. where the flow of HSAF products is described through the hydro models and the comparison of model outputs over test catchments, with the feed back to product improvement.

Fig.19

The validation activity will consolidate a structure for on-going H-SAF products validation and quality assessment which will be an operational service added to products delivering. The development structure arranged for the development phase is given in the breakdown scheme of fig.20, where an idea of its complexity is given.

Fig.20

CONCLUSION

H-SAF shall provide operationally retrieved parameters regarding water mass at ground related to meteorology like as precipitation, soil moisture and snow. These information can meet a wide variety of end users, but the HSAF focus mainly Hydrology, to this respect a specific operational service is considered the continuous validation with Hydro models of the HSAF products. This activity has to be will turn out to give updated quality/confidence information on the products. The basic goal of the HSAF is to expand the use of satellites in Hydrology. In the short term, the Development Phase (2005-2010) will: - make available new products (precipitation, soil moisture, snow parameters) in a pre-operational

way; - progressively improve products quality through a continuous development programme - parallel to the pre-operational activity; - independently assess the benefit of the new products through a hydrological validation

programme. Consequent to a• demonstrated feasibility and affordability of generating the new products and a demonstrated cost-effectiveness for application to Hydrology, Operational Phase (2010-2015) could follow. HSAF can be considered end user of retrieval products and it is strongly interested into establishing tight relation with the International ATOVS Group with the aim of establishing a lasting and fruitful cooperation as HSAF can be considered end user of retrieval products and it is strongly interested into establishing tight relation with the International ATOVS Group with the aim of establishing a lasting and fruitful cooperation.