Satellite inspired hydrology in an uncertain future: a H SAF and HEPEX workshop

Europe/London
ECMWF

ECMWF

Reading
Description

ECMWF hosted the H SAF and HEPEX joint workshop on “Satellite inspired hydrology in an uncertain future”, with the following objectives:

  • To review the status of the hydrological variables retrieval algorithms and to prepare future satellites (e.g. METEOSAT Third Generation, METOP‐SG, SWOT)
  • To characterize the hydrological variable accuracy and to discuss requirements and validation metrics 

  • To review the status of modelling and data assimilation for hydrological and NWP applications and to share new ideas
  • To strengthen links between the H SAF and the Risk management, Hydrological and NWP communities to promote a consistent Earth system approach. 

Registration
Satellite inspired hydrology in an uncertain future: a H-SAF and HEPEX workshop
Events team
    • 13:00 14:30
      Registration and welcome
      Convener: Patricia de Rosnay (ECMWF)
      • 13:00
        Registration and coffee 40m Weather Room

        Weather Room

      • 13:40
        Workshop information 10m
        Speaker: Patricia de Rosnay (ECMWF)
      • 13:50
        ECMWF Welcome 10m
        Speaker: Florence Rabier (ECMWF)
      • 14:00
        EUMETSAT Welcome 10m
        Speaker: Lothar Schueller (EUMETSAT)
      • 14:10
        H SAF Welcome 10m
        Speaker: Francesco Zauli (ITAF COMET)
      • 14:20
        HEPEX Welcome 10m
        Speaker: Fredrik Wetterhall (ECMWF)
    • 14:30 17:00
      H SAF products and quality assessment
      Conveners: David Fairbairn (ECMWF), Silvia Puca (Italian Civil Protection)
      • 14:30
        Soil moisture products 20m

        Soil moisture is a key parameter for environmental (e.g. flood, drought,fire) and weather prediction systems. The EUMETSAT H SAF products give information on both surface and root zone soil moisture. The ASCAT Surface Soil Moisture (SSM) products are produced globally in near real-time and represent the water content in the upper soil layer (< 2cm). The SSM products are distributed in two spatial resolutions (25/50km) for each Metop satellite. Beginning of every year a new Metop ASCATSurface Soil Moisture (SSM) Climate Data Record (CDR) is generated by reprocessing all historic Metop ASCAT data in a consistent manner. During the year an offline extension of the latest ASCAT SSM CDR is also made available for users.

        The ASCAT Root-zone Soil Moisture (RZSM) near real-time product is expressed as a liquid water index and is based on the assimilation of the ASCAT SSM products within the ECMWF/H SAF land data assimilation system. It is a global product available daily at 25km resolution for four layers of soil. Reprocessed versions of the RZSM NRT product are the multi-year time series RZSM CDR products covering the ERS/ASCAT scatterometer data record for the time period 1992-2018.

        Speakers: Apostolos Giannakos (ZAMG), Sebastian Hahn (TU Wien), David Fairbairn (ECMWF)
      • 14:50
        Soil moisture products: Quality assessment and hydrovalidation 20m

        H SAF provides satellite-derived products of precipitation, soil moisture and snow. These products are continuously validated to guarantee optimal performance. Various soil moisture (SM) products are distributed in near real-time, namely, the surface SM product derived from the radar backscattering coefficients measured by the Advanced Scatterometer (ASCAT) on-board the series of Metop satellites (H101 – H16 – H102 – H103), the disaggregated Surface SM product (H08) derived from the global ASCAT NRT product (H16), and, the root zone SM index product (H14), based on H102/H103 data assimilation in the ECMWF Land Data Assimilation System.
        In the last 5 years, the validation of SM has included both product validation and hydrological validation. The product validation has considered both the classical validation against ground-based observations and Triple Collocation, while the hydrological validation has been used to assess the product ability to improve the prediction of hydrological extremes in different European basins and, with the main purpose of using the products in an operational context.
        We present the summary of the validation of H SAF soil moisture products trying to highlight products benefits and limitations, and identify the open issues that should be addressed in the next phases of H SAF project.

        Speaker: Dr Christian Massari (CNR-IRPI)
      • 15:10
        Coffee break 30m Weather Room; Lobby

        Weather Room; Lobby

      • 15:40
        HSAF Precipitation products 20m

        Precipitation is the most important variable in Earth hydrological budget being the major component of water cycle. For this reason, the better under¬standing of the spatial and temporal distribution of precipitation, and its quantification, is fundamental for any hydrological and climatological application. Although surface precipitation gauges are considered the standard devices for precipitation measurements, in many areas they are sparse or not available. This, combined with the temporal and spatial variability of precipitation occurrence, phase, and intensity, makes monitoring and estimation of global precipitation very challenging.
        Meteorological satellites provide a unique opportunity for global precipita¬tion monitoring.
        Within H SAF precipitation products are mainly based on the exploitation of current (AMSU- MHS, SSMIS, ATMS, AMSR-2, GMI) and future, (e.g., MWI, MWS) microwave radiometers on board a constellation of Low Earth Orbit (LEO) satellites. The goal is to achieve the best temporal and spatial coverage by converting microwave radiation measurements, directly related to emission/scattering properties of liquid/solid hydrometeors, into surface precipitation rates. For near-real time monitoring and applications, products based on a combination of geostationary infrared observations with passive microwave (PMW) are also delivered.

        Speaker: Dr Davide Melfi (ITAF COMET)
      • 16:00
        H SAF precipitation products: Quality assessment and hydro validation 20m

        The H SAF concept was born on the requirements of the hydrologic community and is focused on
        the development and operational delivery of satellite products that can be used in work pursued by
        the National Hydrological Services and water management institutions. In H SAF program are
        available two types of validation on precipitation products: quality assessment and hydrological
        validation. For the first one, the H SAF Precipitation Product (PP) Validation Group (VG) analyses
        annually all the PP released by the consortium in order to evaluate their performance over time. On
        the European area validation is carried out by comparison with ground radar and raingauge data
        available within the consortium. For products with hemispherical/global coverage the comparison
        is performed with precipitation estimates by the DPR (Dual-frequency Precipitation Radar) product
        onboard of GPM satellite that combines both Ku and Ka frequencies. In the second one, the H SAF
        Hydrological Program is focused on an assessment of PP from a hydrological point of view: it
        includes products interfacing with hydrologic models. The methodology developed, models and
        basins used, and instruments applied will be briefly described in order to show results achieved by
        the H SAF PP.

        Speaker: Marco Petracca (Civil Protection Department)
      • 16:20
        HSAF snow cover products: From developing to operation stage 20m

        Reliable snow cover extent is of vital importance in order to have a comprehensive understanding for present and future climate, hydrological, and ecological dynamics. Development of methodologies to obtain reliable snow cover information by means of optical and microwave remote sensing (RS) has long been one of the most active research topics of the RS community.
        Operational snow products namely H10 (Snow detection (snow mask) by VIS/IR radiometry), H11 (dry/wet by MW radiometry), H12 (Effective snow cover by VIS/IR radiometry AVHRR), H13 (Snow Water Equivalent by MW radiometry), H31 (Snow detection by VIS/IR radiometry), H32 (Effective snow cover by VIS/IR radiometry AVHRR) have been developed since 2008 within HSAF. The development of new snow products are in progress. Considering different characteristics of snow for mountainous and flat areas, various algorithms are used in producing the snow products for flat and mountainous areas, and then the products are merged to have a single snow product.

        Speakers: Zuhal Akyurek (METU), Ali Nadir Arslan (Finnish Meteorological Institute)
      • 16:40
        Operational validation of H SAF snow products 20m

        H SAF provides satellite-derived products of precipitation, soil moisture and snow. These products are yearly validated to guarantee optimal performance. Under the project, different snow products are distributed in near real-time, namely, SE-E-SEVIRI (Snow detection by VIS/IR radiometry); WS-E (Snow status - dry/wet - by MW radiometry); FSC-E (Effective snow cover by VIS/IR radiometry) and SWE-E (Snow water equivalent by MW radiometry). The snow products validation includes both product validation and hydrological validation.
        In product validation each product has a dedicated common validation procedure: SE-E-SEVIRI, WS-E and SWE-E are validated using ground data from stations over the European area while FSC-E is validated using high resolution satellite data from COPERNICUS Sentinel 2 satellites. In this study, we present the adopted methodologies for the validation and a summary of the results trying to highlight products benefits and limitations. The hydrological validation is used to assess the product ability to improve the prediction of hydrological modelling, it will be shown that the assimilation of snow products (SWE and SCA) in hydrological models, with different approaches, can improves the model states and runoff prediction performance.

        Speaker: Simone Gabellani (CIMA Research Foundation)
    • 17:00 18:30
      Ice breaker and poster session Weather Room; Lobby

      Weather Room; Lobby

    • 09:00 14:20
      Session 1: Remote sensing, hydrological modelling and data assimilation
      Conveners: Luca Ciabatta (CNR-IRPI), Patricia de Rosnay (ECMWF)
      • 09:00
        Challenges for the HEPEX community in the coming years 20m

        Since 2004, HEPEX (Hydrologic Ensemble Prediction Experiment) has been fostering a community of researchers and practitioners around the world. The mission is still to establish a more integrated view of hydrological forecasting, where data assimilation, hydro-meteorological modelling chains, pre- and post-processing techniques, expert knowledge, participatory co-evolution of knowledge and user needs, communication and visualisation tools, training material and games, and decision support systems are connected to enhance operational systems and water
        management applications. Massive progress has been made over the years in terms of using ensemble hydrometeorological forecasting, but there are still institutional, scientific and operational challenges that the community faces.
        Here we present a brief introduction to how data assimilation have been used within the HEPEX community with focus on uses in operational forecasts. We also outline the main challenges that still exist with regards to using data assimilation in operational hydrometeorological framework.

        Speaker: Dr Fredrik Wetterhall (ECMWF)
      • 09:20
        Data assimilation for continuous global assessment of severe conditions over terrestrial surfaces 30m

        This study investigates the capability of LDAS-Monde global offline LDAS to monitor and forecast the impact of extreme events on Land Surface Variables (LSVs). LDAS-Monde is driven by ERA-5 atmospheric forcing from ECMWF and is able to ingest information from satellite-derived surface soil moisture (SSM) and leaf area index (LAI) observations to constrain the ISBA land surface model.
        A global 2010-2018, 0.25°x0.25º spatial resolution, reanalysis of the LSVs is first evaluated thanks to global estimates of SSM, LAI, evapotranspiration , Gross Primary Production, Sun Induced Fluorescence and several in situ measurements of soil moisture, river discharge, and flux measurements. This 9-yr reanalysis is used to provide a climatology of the LSVs. Significant anomalies are used to decide on where to focus for a more detailed monitoring and forecasting activity. 19 regions across the globe were investigated for 2018. Two of them, presenting large negative anomalies of SSM and LAI were further analysed: Western-Europe and the Murray-Darling river basin in southeastern Australia. LDAS-Monde was operated forced by ECMWF IFS high-resolution atmospheric analysis leading to a 0.1°x0.1° reanalysis. It complements the coarse-resolution LDAS-Monde operated using ERA5. The IFS forecast capacity initialised by LDAS-Monde analysis is also presented.

        Speaker: Dr Clement Albergel (CNRM - Université de Toulouse, Météo-France, CNRS)
      • 09:50
        Joint assimilation of soil moisture and flood extent maps retrieved from satellite earth observation into a conceptual hydrological model for improving flood prediction: a proof of concept study. 20m

        The main objective of this study is to investigate how innovative satellite earth observation techniques that allow for the estimation of soil moisture and the mapping of flood extents can help in reducing errors and uncertainties in conceptual hydro-meteorological modelling especially in ungauged areas where potentially no or limited runoff records are available. A spatially distributed conceptual hydrological model is first developed allowing for the simulation of soil moisture and flood extent. Using as forcing of this model rainfall and air temperature time series provided in the globally and freely available ERA5 database it is then possible to carry out long-term simulations of soil moisture, discharge and flood extent. Next, time series of soil moisture and flood extent observations derived from freely available satellite image databases are jointly assimilated into the hydrological model in order to retrieve optimal parameter sets. For this assimilation experiment, we take benefit of recently introduced Particle Filters with tempering that circumvent some of the usual particle filter limitations such as degeneracy and sample impoverishment. As a proof of concept, we set up an identical twin experiment based on synthetically generated observations and we evaluate the performance of the calibrated model.

        Speaker: Dr Renaud Hostache (Luxembourg Institute of Science and Technology, Environmental Research and Innovation Department)
      • 10:10
        An automatic system for flood mapping based on Sentinel-1 data 20m

        Floods are the most frequent and costliest natural disasters worldwide. The potentiality of the images provided by synthetic aperture radar (SAR) systems for near real-time flood mapping was demonstrated by several past studies. Nowadays, scientific methods for daily automatically detection and identification of flood area or, more generally, areas where standing water is present from SAR data are mature and ready for an operational implementation which can complement the Copernicus Emergency Service (CEMS) for mapping and monitoring floods.
        A fully automated Sentinel-1 based flood service was designed to work at national scale on the framework of the convention between the Italian Department of Civil protection (DPC) and CIMA Research Foundation. The processing chain implementing the service includes the automatic and daily procurement of Sentinel-1 data and the run of a fully automatic flood mapping algorithm. The flood mapping algorithm firstly performs the geocoding and the calibration of the GRD data and then applies a combination of an Isodata clustering, an automatic thresholding and a Region Growing technique. Flood maps generated by the processing chain will be shown and discussed at the conference.

        Speaker: Dr Elisabetta Fiori (CIMA Research Foundation)
      • 10:30
        Coffee break 40m Weather Room; Lobby

        Weather Room; Lobby

      • 11:10
        Precipitation measurements for hydrological applications 20m

        The measurement of precipitation across the Earth’s surface is complex due to the vagaries in the occurrence and intensity of precipitation, together with the variations in the temporal and spatial distribution of precipitation across the global. The physical nature of precipitation necessitates that the observation, sampling and reporting of precipitation is adequate for the resulting end user and their applications, which are in turn constrained by engineering and technology.

        This paper provides and overview of precipitation measurement techniques, with particular reference to their use in hydrological applications. Gauge measurements provide the most direct measure of precipitation, although the temporal and spatial sampling vary greatly. Weather radars (where available) provide spatial maps of precipitation with frequent/regular sampling, although ultimately are calibrated against gauge data. Gauges and radars are essentially confined to global land surfaces: satellite observations are capable of providing precipitation measurements on a global basis. Techniques have been devised to extract precipitation from these observations and presently provide precipitation measurements of up to 10km every 30m.

        The relative importance and accuracy of the different sources and products at different spatial and temporal resolutions will be assessed. The paper will conclude by providing an insight into future measurements of precipitation.

        Speaker: Dr Christopher Kidd (UMD/ESSIC & NASA/GSFC)
      • 11:30
        Projected Advances in the Remote Sensing of Precipitation 30m

        The talk will review the current status of satellite and surface networks capable of providing precipitation with high space/time resolution with a view towards the future. While ground based radar networks continue to see systematic although measured progress, there will be a rapid increase in satellite capabilities both in terms of microwave sensors made possible by new Cube- and MicroSat technology, as well increased sampling frequency from the next generation of geostationary VIS/IR sensors. This will stress our ability to create general products that satisfy broad user categories with varying requirements for resolution, quality and stability. Instead it will likely require the production of more focused products such as hydrology. A second emerging trend is for satellite algorithms, through maturation and the recent increased use of machine learning, to fully exploit the available information content. This will likely lead to greater emphasis being placed on error estimation, while existing challenges, posed by such conditions as drizzle and orographic precipitation is requiring greater model input leading to more synergistic frameworks.

        Speaker: Christian Kummerow (Colorado State University)
      • 12:00
        Snow depth observations from Sentinel-1 over the Northern Hemisphere mountain ranges 30m

        The snow depth in the world’s mountains ranks among the most uncertain variables in hydrology. Estimates from the interpolation of local measurements are unrealistic where they are sparse, atmospheric models poorly estimate snowfall, and current snow remote sensing observations have inherent limitations. Yet, accurate snow depth estimates are critically needed to provide information on the associated water resources. More than a billion people rely on water from snow, most of which originates in the Northern Hemisphere mountain ranges. Here, we demonstrate the unprecedented ability of the Sentinel-1 mission to map ~weekly snow depth in the Northern Hemisphere mountains at 1-km² resolution. An evaluation with measurements from ~4,000 sites and reanalysis data demonstrates that the Sentinel-1 observations capture the spatial variability between and within mountain ranges, as well as their inter-annual differences. The latter is showcased with the contrasting snow depths between 2017 and 2018 in the US Sierra Nevada and European Alps. The Sentinel-1 observations offer new opportunities for data assimilation, to improve the initialization of numerical weather, seasonal and climate predictions. The long-term continuity of the ESA and Copernicus Sentinel-1 constellation is a strong asset, offering the frequent and continual observations that are required for monitoring the cryosphere.

        Speaker: Dr Hans Lievens (Department of Earth and Environmental Sciences, KU Leuven)
      • 12:30
        The impact of satellite data assimilation on hydrologic model perfor- mance 30m

        The very scales at which data may be required and the accessibility of sampling points quickly renders certain direct measurements impractical e.g., changes in land cover, sea surface temperature, snow in mountains, etc. Satellite-derived products are a valuable source of such data and overcomes the large spatial extent limitations. The data is sometimes at a reasonable temporal resolution as well, which addresses the sampling frequency challenges. The satellite data, at the large spatial scales can provide boundary conditions for constraining numerical models but can also be assimilated to improve their predictions. In this study we show the effect of assimilating the actual and potential evapotranspiration, the snow water equivalent and the fractional snow cover; and the different combinations of the four products into the hydrologic model HYPE. We assessed the dependency of the hydrologic model performance by assimilating different products on: i) whether the model is calibrated to the data to be assimilated or not, ii) observation error model assumptions, iii) gaussification of non-Gaussian model and observation fields iv) ensemble size, v) assumptions regarding error model for perturbing the model forcing data (ensemble generation).

        Speaker: Jude Musuuza (SMHI)
      • 13:00
        Lunch break 1h ECMWF

        ECMWF

        Reading
      • 14:00
        Characterization and monitoring of heavy precipitation events in the Mediterranean area using the H-SAF precipitation products 20m

        The Mediterranean region is often impacted by heavy precipitation events, responsible for damages and sometimes casualties. Because of its peculiar geographical characteristics, ground-based observations (e.g. raingauges and radars) cannot be fully exploited for the monitoring, characterization and forecasting of severe events, in particular during their offshore development. Therefore in the last decades satellite observations, both over geostationary and LEO platforms, have been recognized as a powerful tool to overcome the limitations of ground-based observations.
        In this study we analyze severe events that have recently affected the Mediterranean area, by exploiting the Global Precipitation Measurement (GPM) mission active and passive MW measurements. We also analyze MW and MW/IR H SAF precipitation product capabilities to estimate and monitor the precipitation throughout the evolution of these events. We highlight how the H SAF products, specifically developed for the European area, and thus tailored for the Mediterranean region, show better performances than algorithms designed for global application. We also show how the use of a MW radiometer constellation, in conjunction with other satellite and ground-based observations, allows for the characterization of precipitation structure and microphysics of severe systems in the Mediterranean area, in support of operational forecasting activities in a climate change perspective.

        Speaker: Dr Giulia Panegrossi (Institute of Atmospheric Sciences and Climate, National Research Council (ISAC/CNR))
    • 14:20 16:30
      Session 2: Hydrological validation and benchmarking
      Conveners: Fredrik Wetterhall (ECMWF), Zuhal Akyurek (METU)
      • 14:20
        EO-based retrieval of snow cover, overview of selected snow products and their quality assessment 30m

        Monitoring terrestrial snow cover on continental and global scales is complicated by large differences on regional and local conditions as well as large gaps in surface observing networks. Therefore satellite-based observations provide the best spatially and temporally extensive means for snow cover monitoring on hemispherical scale.

        The objective of the ESA Snow CCI is to generate homogenized long-time series of daily global snow extent (SE) maps from optical and daily global snow water equivalent (SWE) products from passive microwave satellite data.

        The goal of the EUMETSAT H SAF “snow cluster” and the Copernicus Global Land Service “cryosphere theme”, is production of satellite based near real time maps of various snow cover parameters.

        The uncertainty associated with current hemispherical datasets on both SE and SWE are significant. There are significant differences even between the well-established snow products which has been made obvious in the ESA SnowPEx project that has investigated and inter-compared the available SE and SWE products.

        A brief overview of these ESA, EUMETSAT and Copernicus frameworks is presented. Overview will consider snow cover extent and snow water equivalent retrieval approaches and products, both for operational near-real time purposes and historical climate data records of snow.

        Speaker: Dr Kari Luojus (Finnish Meteorological Institute (FMI) Arctic Space Centre)
      • 14:50
        Coffee break 1h Weather Room; Lobby

        Weather Room; Lobby

      • 15:50
        Improving the snowmelt modelling in mesoscale Hydrological Model (mHM) with satellite based dynamically calculated degree-day factor 20m

        Modelling snow events stayed as challenge as it depends on multiple parameters. For the sake of performance and fewer requirements, temperature-index models employed. These models generally calibrate a basin-wide threshold temperature, a degree-day factor (DDF) which accepted as constant. Some hydrologic models such as Mesoscale Hydrologic Model (mHM), that employs multiscale parameter regionalization approach (MPR) to model sub-grid variability, improves DDF calibration spatially, based on land cover, yet temporal changes remained ignored. In this study, daily DDFs calculated using 500 m2 resolution MODIS snow-covered area product (MOD10A1) and snow depth measurements. Snow density calculated in days of accumulation and using this value DDFs are calculated in days of ablation. mHM is fed with daily DDF values and simulations done on two mountain basins in Turkey: small-scale Karanlikdere basin (area: ~29.1 km2 and elevation range: 1295 - 2332m) and mesoscale Bahcelik basin (area: ~2378.6 km2 and elevation range: 1497 - 2705m). Preliminary results indicate that remotely-sensed data improves model performance: discharge estimations at gauging points are slightly improved while snow-water-equivalent calculations became more realistic. For Karanlikdere Basin, mHM-calibrated DDF (2.395) converges to mean daily dynamic DDF (2.431) and NSE for the discharge simulation improved from 0.45078 to 0,45785.

        Speaker: Mr Birkan Erguc (Middle East Technical University)
      • 16:10
        Combining Passive and Active Microwave Remote Sensing Data to Assess the Impact of Forest Fires on the Hydrology of Boreal Forests 20m

        The observed warming of temperatures in boreal regions in the past decades has led to an increase in the frequency, intensity and size of forest fires with numerous consequences on ecosystems and water budget. Hydrological models can be used to assess the impact of these fires on the hydrological regime, but they necessitate sufficient input data to produce reliable results and ground measurements can be difficult to obtain in such remote regions. The objective of this project is to use a combination of passive and active microwave remote sensing data to estimate soil moisture in boreal forests and to see if the impact of forest fires can be modeled using these inputs. The study area is situated in Northern Quebec within the Rupert and Manicouagan river catchments, which were affected by large forest fires during the summer of 2013. The first step consists in downscaling data from passive microwave sensors (SMOS and SMAP) at the watershed scale. This is accomplished by using a relationship between Synthetic Aperture Radar (SAR) backscattering from Sentinel-1 data and optical thickness measured with SMOS. The produced soil moisture map will then be introduced into a semi-distributed hydrological model to assess the impact of forest fires.

        Speaker: Dr Yannick Duguay (Université de Sherbrooke)
    • 16:30 18:00
      Poster session
    • 09:00 09:50
      Session 2: Hydrological validation and benchmarking
      Conveners: Fredrik Wetterhall (ECMWF), Zuhal Akyurek (METU)
      • 09:00
        Optimization of the satellite datasets to study the water cycle at the global scale 30m

        Over the last decades, satellite observations have increasingly been used to study the global terrestrial Water Cycle (WC). We will present here applications to retrieve for instance soil moisture, surface waters or underground water storage. However, the use of satellite products is still limited by their uncertainties and inconsistencies like their inability to close the WC budget. We show that it is possible to optimise these datasets by integrating them based on the closure of the water budget at the basin scale. This is performed using observations only and no model assimilation is required. It is possible to close simultaneously the terrestrial, oceanic and atmospheric WC budgets. The obtained optimized hydrological coherent dataset can be used to design an independent and simple calibration of each satellite dataset to reduce the overall WC budget residuals. This calibration procedure is extrapolated from several basins around the world to the global scale. The reference dataset can also be used to estimate the uncertainties of each data source. This opens new perspectives to generate long-term global datasets based purely on all the available satellites observations. Several applications will be presented over South Asian basins, over the Mediterranean region, and over the Amazon.

        Speaker: Filipe Aires (LERMA / Observatoire de Paris / CNRS)
      • 09:30
        SMOS Soil Moisture and its potential within the Copernicus Emergency Management Service for Flood Forecasting at ECMWF 20m

        The flood forecasting component of the Copernicus Emergency Management Service Early Warning System (CEMS-EWS Floods), run at the European Centre for Medium-range Weather Forecasts (ECMWF), provides medium range flood forecasts at both the European and global scales. Forecast accuracy correlates strongly with accurately representing the initial status of hydrological variables including soil moisture. Initial soil moisture status within CEMS-EWS Floods is currently obtained by a hydrological analysis of point scale temperature and precipitation observations. Remotely sensed soil moisture from SMOS may provide a better representation of the initial soil moisture status than the current methodology.

        This study evaluates the potential use of SMOS soil moisture data within CEMS-EWS Floods, using the level 1 Near Real Time (NRT) SMOS soil moisture Neural Network product trained on the ECMWF land surface scheme. First a long-term comparison against existing re-analysis datasets was made, including the CEMS-EWS Floods analysis in Europe and ERA5 globally, to ascertain spatial and temporal correlations. Data assimilation was then conducted within the European and global components of CEMS-EWS Floods. The resulting changes in streamflow accuracy provide a good test of benefits of assimilating SMOS soil moisture data.

        Speaker: Dr Calum Baugh (ECMWF)
    • 09:50 12:30
      Session 3: Hydrological data assimilation for NWP
      Conveners: David Fairbairn (ECMWF), Samantha Pullen (Met Office)
      • 09:50
        Precipitation data assimilation at ECMWF 20m

        The number of precipitation-related observations that are used in ECMWF’s operational 4D-Var data assimilation system has been steadily increasing over the past two decades, bringing a significant improvement in the quality of atmospheric analyses and subsequent forecasts. This presentation will review the status of ECMWF’s operational assimilation of all-sky satellite microwave brightness temperatures, ground-based radar composites (NEXRAD over the United States) and summarize recent developments toward the assimilation of even more observations from OPERA precipitation composites over Europe, from space-borne radar and lidar as well as from lightning observing systems such as GOES GLM.

        Speaker: Philippe Lopez (ECMWF)
      • 10:10
        Impact of UKV soil moisture data assimilation on potential operational forecast of river flows 20m

        The UKV is the Met Office operational modelling system for the UK area. An hourly cycling 4DVAR algorithm provides atmospheric initial conditions; meanwhile the initial land state is provided by interpolating a global soil moisture analysis at a daily frequency. The forecast step uses the Unified Model for the atmosphere coupled to JULES for the land state at 1.5km.

        The Met Office has recently developed a Soil Moisture analysis for the UKV where MetOp A&B ASCAT soil wetness, and screen observations of temperature and relative humidity, are assimilated in a Simplified Extended Kalman Filter. Although verification against screen observations shows an overall neutral impact, the hydrology estimation is significantly improved.

        In our work, surface and sub-surface runoffs from the UKV are routed through a standalone river scheme and compared against river flow observations. Results shows that the current operational system produces river flows that are several times larger than observations, while the new soil moisture analysis, planned for implementation in November 2019, is able to provide estimations which reproduce the magnitude and day-to-day variability reported by the measurements. This opens the possibility to develop an operational hydrology system with applications areas such as flood forecasting, coupled modelling or land verification.

        Speaker: Breo Gomez (Met Office)
      • 10:30
        Satellite data and operational hydrology - A WMO perspective 30m

        Many global initiatives launched in recent years by the international community, such as the SDG (especially SDG-6 water and sanitation) and WMO’s WHOS, HydroSOS and others have highlighted the growing need for reliable hydrological data. However the outcomes of recent survey carried out by WMO and WB indicate that particularly in low and middle-income countries, Hydrological Services are unable to respond to this growing demand for accessible, quality assured, and timely information on the availability, distribution and trends of water resources. Among the various causes obsolescent ground based hydrometeorological monitoring networks and their insufficient maintenance play a major role in restricting the availability of data. While the ground truth provided by these observations remains a central element of the hydrological monitoring, it is necessary to complement and expand the availability of data and information by leveraging new technologies and international data resources. However only a few NMHSs are effectively using the opportunities offered by satellite information and other new sources of information. Interactions between WMO hydrology and satellite community should be coordinated in a well-defined framework to provide user tailored information and services.

        Speaker: Tommaso Abrate (World Meteorological Organization)
      • 11:00
        Coffee break 50m Weather Room; Lobby

        Weather Room; Lobby

      • 11:50
        Assimilation of SCATSAR-SWI with SURFEX: Resolution studies over Austria 20m

        The performance of numerical weather prediction (NWP) can be improved significantly by assimilating remotely sensed soil moisture in the included land surface model. In our study, we investigate the improvement of the NWP in the Austrian domain by assimilating the multi-layer fused data product SCATSAR-SWI (Soil Water Index) with the surface model SURFEX. The SURFEX assimilation employs a multi-layer diffusion scheme approach and the simplified Extended Kalman Filter. The data has a grid sampling of 500 m, whereas the assimilation system is run with different grid resolutions. Our goal is to find the optimum relation between exploiting the high-resolution information and reasonable computational effort as well as to test the limits of the underlying physical parametrisations in the NWP model.

        The soil moisture analysis with 1.25 km resolution shows a positive impact on the NWP skill metrics compared to 2.5 km when using the same setup for the atmospheric forecast model. To probe the quality of the soil moisture analysis directly, we plan furthermore a comparison with in-situ measurements in Austria.

        Speaker: Dr Jasmin Vural (ZAMG Vienna, Austria)
      • 12:10
        Development of snow depth assimilation for the Met Office UK forecasting system 20m

        An accurate representation of snow extent and depth is of great importance in Numerical Weather Prediction (NWP) models for calculations of surface fluxes which provide the lower boundary conditions for the atmosphere. In the UK the model representation of snow is itself of great interest to forecasters, who must provide current information on and predictions of snow positioning, depths and evolution of lying snow.

        At the Met Office a new snow depth assimilation scheme has been developed for the high-resolution UK Numerical Weather Prediction system. The scheme is based on a 2D Optimal Interpolation method and makes use of observations of snow depth from the ground-based SYNOP network as well as observations of snow cover, derived from MSG SEVIRI and provided by the EUMETSAT H SAF. They provide improved coverage, for cloud-free areas, and valuable observations of snow-free ground, which enable better spatial analysis of the snow-affected surface.

        In this presentation we describe the assimilation method that has been developed and show the performance of the snow analysis for particular UK snow events, along with results from assimilation trials in the Met Office UK NWP system, in preparation for implementation in the operational forecasting system.

        Speaker: Samantha Pullen (Met Office)
    • 12:30 16:50
      Session 4: Impacts of hydrological uncertainty, hydrological forecasting and modelling
      Conveners: Marie-Amélie Boucher (Université de Sherbrooke), Simone Gabellani (CIMA Research Foundation)
      • 12:30
        Snow processes in bucket-type hydrological models – does increased realism lead to better simulations? 30m

        Bucket-type hydrological models such as the HBV model are widely popular, because of their relatively low data and computational demands. This is the case of the snow routine in the HBV model. While this approach usually results in good simulations, improvements are possible. We explored and tested different alternatives to the design of the snow routine of HBV-light such as considering a gradual transition between snowfall and rainfall, or implementing a seasonally variable degree-day factor. Furthermore, we evaluate the value of different data sources for model calibration and testing as well as the importance of different model formulation in the snow routine for simulations for changed climate conditions. We quantify the usefulness of the snow-routine modifications for simulations in alpine and other snow-covered areas. We also discuss the balance between increasing the realism and the preservation of the inherent simplicity of the HBV model. Results indicate that few model modifications result in clear model performance improvements whereas many modifications, despite seemingly increasing the model realism, did not lead to improved model performances. Furthermore, we demonstrate how different formulations of the snow routine might lead to different simulations for changed climate conditions and potentially can create artefacts, which are often overlooked.

        Speaker: Prof. Jan Seibert (Department of Geography, University of Zurich)
      • 13:00
        Lunch break 1h ECMWF

        ECMWF

        Reading
      • 14:00
        Improving hydrological prediction through data assimilation: results from the IMPREX and eWaterCycle II projects 30m

        Improving sharpness/reliability of hydrological forecasts is key to increase the value of a warning service. Hydrological data assimilation is one possible way to increase the accuracy (and possibly reliability). Various studies in assimilation of various ECVs (water level, discharge, soil moisture) show that accuracy can be indeed be improved. Here, we show results from the H2020 IMPREX project on assimilation of lake levels and discharge into a hydrological model of the Rhine using OpenDA (www.openda.org) and an open source hydrological modelling framework wflow (https://github.com/openstreams/wflow). This tool is also used to assimilate available discharge measurements in the W3RA model connected with a kinematic wave subsurface routing model in NRT in an operational global flow forecasting system (GLOFFIS).
        Within the ongoing eWatercylce II project in cooperation with the Dutch eScience Centre (https://www.esciencecenter.nl/project/ewatercycle-ii) we aim to enable joint assimilation of discharges and for instance soil moisture using local analysis, we hope to present first results from this work

        Speaker: Albrecht Weerts (Deltares)
      • 14:30
        The use of H-SAF soil moisture products for event-based hydrological modelling in Liguria (north of Italy) 20m

        Event-based hydrological modelling disregards some hydrological processes that during an event resulted neglectable. A reliable initial condition of model is fundamental for discharges prediction and, consequently, for flood risk mitigation. Satellite observations can be exploited to estimate soil moisture and these can be used to enhance the predictions of hydrological models.
        The aim of this work is to equalize the initial soil moisture of hydrological model through a statistical model. The current study investigates how these two operational soil moisture (SM) products provided by the H-SAF can improve the simulation during the flood event.
        The analysis is conducted for some events with different initial conditions and different modelling of soil moisture, evaluating where remote sensing provides better performance. The analysis of hydrographs through Nash–Sutcliffe model efficiency coefficient contains useful indication for flood forecasting system, particularly in this case where a continuous modelling approach is not implemented.

        Speakers: Dr Federica Martina (Agenzia Regionale per la Protezione dell’Ambiente Ligure (ARPAL)), Dr Martina Raffellini (International Centre on Environmental Monitoring (CIMA) Research Foundation)
      • 15:00
        Coffee break 50m Weather Room; Lobby

        Weather Room; Lobby

      • 15:50
        Assimilation of flood maps derived from satellite SAR data into a flood forecasting model 20m

        The objective of this study is to evaluate the potential of assimilating flood extent information derived from satellite Earth observation into a flood forecasting chain for reducing predictive uncertainty. A recent promising study introduces a DA framework based on probabilistic flood maps derived from SAR images considering uncertain rainfall. In our study, we carry out an identical twin experiment using synthetically generated observations to further evaluate and develop this DA framework. Synthetic probabilistic flood maps are assimilated, using a particle filter, into a flood forecasting chain composed of a rainfall-runoff and a shallow water model. The ‘truth’ is generated based on a forward run of the forecasting chain using ERA-interim data as meteorological forcings. Synthetic probabilistic flood maps are derived from the ‘truth’ flood extent maps via generating synthetic satellite images. For the DA experiment, an ensemble of particles is generated by perturbing the rainfall data and propagating these perturbations through the forecasting system. The DA allows updating the particle posterior distribution each time a synthetic satellite image is generated. Results show that discharge and water elevation predictions are significantly improved as a result of the assimilation.

        Speaker: Concetta Di Mauro (Luxembourg Institute of Science and Technology)
      • 16:10
        Understanding Water Availability Within Ugandan through the Drought and Flood Mitigation Service 20m

        A consortium led by the RHEA Group, working with the Ugandan Ministry of Water and Environment and local Non-Governmental Organizations (NGOs, AgriTechTalk Uganda and Mercy Corps) has developed a Drought and Flood Mitigation Service (DFMS), funded as part of the UK Space Agency’s International Partnership Programme.

        DFMS is using satellite Earth observation (EO) data alongside meteorological and hydrological modelling and ground-based data within an innovative cloud computing-based platform. EO products come from multiple missions and are the basis for the onward development of information services with the modelling activities allowing for future predictions to be made.

        For EO, related to water availability, this includes:
        • 1 km spatial resolution maps of evapotranspiration, Land Surface Temperature (LST) and soil moisture;
        • 10 m resolution vegetation indices and water extent maps;
        • Point-based water height data.

        The different products are being combined to provide insights into the soil / LST / vegetation ‘triangle’. The presentation will showcase the products and relationships between them, with the EO data being compared to ERA-Interim / ERA5 and hydrological model outputs.

        Speaker: Samantha Lavender (Pixalytics Ltd)
    • 19:00 21:00
      Workshop dinner at Côte Brasserie
    • 09:00 11:20
      Session 5: Novel hydrological data sources and assimilation techniques
      Conveners: Filipe Aires (LERMA / Observatoire de Paris / CNRS), Dr Renaud Hostache (Luxembourg Institute of Science and Technology)
      • 09:20
        Hydrological data assimilation using machine learning 20m

        Data assimilation (DA) allows for updating state variables in a model to represent reality more accurately than the initial (open loop) simulation. In hydrology, DA is often a pre-requisite for forecasting. Neural networks (NN) can learn almost any nonlinear relationship between inputs and outputs. Here, we hypothesize that NN could learn the relationship between the simulated streamflow (from a hydrological model) and the corresponding state variables. Once learned, this relationship can be used to obtain corrected state variables by applying it to observed rather than simulated streamflow. Based on this, we propose a novel, ensemble-based, DA approach. To verify the above mentioned hypothesis, we used an international testbed comprising of four contrasted watersheds. We applied the new DA method to the lumped hydrological model GR4J, which has two state variables. Within this framework, we compared two types of NN, namely Extreme Learning Machines (ELM) and Multilayer Perceptrons (MLP). Using well-known metrics such as the CRPS, we compared the assimilated streamflow series with the open loop streamflow series and with the observed streamflow. We show that NN are effective for DA, with a noticeable improvement over the open loop simulation for all watersheds.

        Speaker: Prof. Marie-Amélie Boucher (Université de Sherbrooke)
      • 09:40
        Retrieval of soil moisture using neural networks 20m

        A retrieval methodology based on Neural Networks was proposed (Aires et al. 2005) to retrieve and assimilate Soil Moisture (SM) from satellite observations. It is based on the fact that no radiative transfer model was satisfactory enough for a physically-based algorithm. An innovative aspect is to train the NN using modelled SMs. The resulting retrievals are not a reproduction of the model: the temporal and spatial variability is driven by the satellite observations and retrievals can actually correct erroneous models (Aires et al. 2005; Jimenez et al. 2013). The algorithm can work with passive or active microwave observations and the NN is able to combine them to exploit their synergy (Kolassa et al. 2013; 2016; 2017). This approach has been developed using recent instruments (e.g. SMOS, Rodrigues et al. 2015). The scheme uses implicitly a general CDF matching (at the global scale) that facilitates the assimilation of the retrieved SMs; avoiding traditional pixel-based CDF matching that modifies the satellite spatial patterns towards the model. Assimilation of the retrieved SM was recently tested at ECMWF using SMOS data (Rodriguez et al. 2018) and downscaling of coarse resolution SMs has been successfully performed (Alemohammad et al. 2018).

        Speaker: Filipe Aires (LERMA / Observatoire de Paris / CNRS)
      • 10:00
        On the impacts of location, timing, and frequency of inundation extent assimilation on flood forecast skill 20m

        Recent studies have demonstrated the potential of assimilating probabilistic inundation maps derived from Synthetic Aperture Radar (SAR) imagery for improved flood forecasts. However, high resolution SAR acquisition can only provide partial coverage of large catchments. Consequently, information on the impacts of location, timing, and frequency of inundation extent assimilation on flood forecast skill can provide guidance for acquisition planning. To investigate this issue, twin experiments were set up at 90m grid resolution using the two dimensional hydraulic model LISFLOOD-FP for the 2011 flood event in the Clarence Catchment, Australia. The truth run was setup using the observed inflow hydrograph, calibrated parameters, and LiDAR topography, while the open loop considered errors in all these datasets. Error characteristics of COSMO-SkyMed SAR images available for the event, were estimated and used to generate synthetic images, subsequently converted to probabilistic flood extents for assimilation. Synthetic observations were assimilated into three catchment sub-regions, delineated by flow distances based on reach flow behaviour, using a new Particle Filter-based algorithm. Preliminary results indicate that the forecast performance is highly sensitive to the spatiotemporal characteristics of the observation and that flood extent assimilation can even lead to forecast degradation.

        Speaker: Dr Renaud Hostache (Luxembourg Institute of Science and Technology)
      • 10:20
        Sequential and variational assimilation of satellite snow data through a conceptual hydrological model in a mountainous catchment 20m

        Analyzing and forecasting the variability of snow is essential for runoff prediction especially in mountainous regions where the optimal operation of reservoirs is important. Remote sensing information has been extensively developed over the last decay with enhanced snow data sets at high resolutions. The implementation of satellite facilities in operational runoff forecasting systems by means of data assimilation provides an improvement in the initial conditions of streamflow forecasts. The snowmelt is the main component of runoff in the Eastern Turkey despite the limited availability of ground snow observations. Therefore, incorporating different DA techniques together with observed discharge and satellite snow data is very crucial in the runoff predictions over the region. The work includes implementation of sequential and variational techniques through a conceptual hydrological model to assimilate H-SAF snow products on snow covered area (SCA) data during the period of 2008-1012. Sequential techniques have been commonly applied to hydrological processes however their application over snow dominated catchments is rather limited. On the other hand, variational techniques have been seldom used. Data assimilation results show a progress in the lead time performance of streamflow forecasts by using perfect forecast data beside an improvement in the forecast skill of modelled snow states.

        Speaker: Dr Gokcen Uysal (Eskisehir Technical University)
      • 10:40
        Implementation of a coupled land-atmosphere modeling system within a northwestern Mexican river basin 20m

        Fully coupled modelling of atmospheric and hydrological processes is of growing interest among the hydrometeorology community. Understanding multi-time scale processes within a land-atmosphere modelling system is of importance for improving forecast between medium-range and seasonal forecasts. Improving forecast capabilities between sub-seasonal to seasonal forecasting, can help shape the design of strategies and programs for water resource management, flood risk management, water supply and irrigation design. In this study, we conduct coupled land-atmosphere simulations within a large river basin in northwest México, for an extreme hidro-meteorological event that occurred in September 2018, simulating precipitation and runoff during the extreme event, through the use and implementation of the fully coupled WRF/WRF-Hydro modeling system. Firstly, the WRF-Hydro modelling system is tested as a stand-alone hydrological model for parameter calibration and validation using observed streamflow data, and purposely assessing model reliability for the study area. Secondly, fully coupled land-atmosphere simulations are conducted during a 1 year long simulation in the Sinaloa Hydrologic Region. Performance assesment is accomplished by comparing observed and simulated variables: precipitation, runoff and soil moisture. Lastly, for uncertainty assessment we perfome simulations after perturbing model input data and initial conditions, within an ensamble bayesian forecasting system framework.

        Speaker: Ms Jocelyn Betsabe Serrano Barragan (National Autonomous University of Mexico (UNAM))
      • 11:00
        Coffee break 20m Weather Room; Lobby

        Weather Room; Lobby

    • 11:20 12:20
      Sessions summary and discussions
      Convener: H SAF project manager and session chairs
    • 12:30 14:00
      Lunch break 1h 30m
    • 13:00 13:30
      Tour of supercomputer hall (optional)
    • 13:30 14:00
      Tour of supercomputer hall (optional)
    • 14:00 17:30
      Additional activity: Demonstration sessions
      Conveners: David Fairbairn (ECMWF), Fredrik Wetterhall (ECMWF)