Modelling the terrestrial hydrological cycle at ‘hyper-resolution’, i.e., with a grid cell size of 1 km or below) was and still is a major quest in hydrological sciences. With an increase in computational power and the number of readily available and open datasets at useful spatial resolutions increasing as well, hyper-resolution modelling efforts have grown in number as well. We here present a first continental-scale application of the global hydrological model PCR-GLOBWB over Europe at 1 km spatial resolution, and offset it against runs with traditional resolutions of 10 km and 50 km, respectively. Model output was validated for more than 200 water provinces against observed discharge and the following remotely sensed data products: ESA-CCI soil moisture data and GRACE/GRACE-FO terrestrial water storage anomalies. Evaluation metrics indicate good model performance over Europe and increased accuracy with finer spatial resolutions, particularly for discharge simulations. While the used validation products have the advantage of global coverage and long observational records, their spatial resolution is actually too coarse to fully assess the accuracy of models at hyper-resolution. At that scale, more recent satellite products can be of more use but at the cost of only short observation record. We thus additionally validated 1 km model output against Sentinel-1 surface soil moisture and compared it against results obtained for ESA-CCI soil moisture data. Besides challenges related to global-scale fine-resolution observational data, we also acknowledge that additional work needs to focus on model parameterization for hyper-resolution as well model improvements such as routing schemes better utilizing the available spatial detail. Another challenge we identified is required run time and computational power to analyze continental-scale 1km data, even when using the state-of-the-art Dutch supercomputer. Here, efficient programming and use of latest parallelization techniques will become even more crucial. Despite these solvable challenges, our research shows that large-scale hyper-resolution modelling is now feasible and that further pursuing these efforts can eventually lead to more locally-relevant hydrological information and process understanding.

In terrestrial systems research, technologies related to observations, modeling, and data assimilation and analyses underwent considerable improvement in the past two decades. Simultaneously, the development of digital infrastructures related to cloud and high-performance computing, data storage, big data analytics, and observational technologies continue to advance at an extremely high speed. Thus, the potential is huge of fusing available technologies to monitor, reconstruct, forecast, and project the terrestrial water, energy, and biogeochemical cycles up to continental and global scales for the benefit of societies. Here, we discuss the main aspects of technology fusion performed at the Research Centre Jülich and the Centre for High-Performance Scientific Computing in Terrestrial System, Geoverbund ABC/J. The focus is placed on high-resolution groundwater-to-atmosphere simulations utilizing the Terrestrial Systems Modeling Platform (TSMP) up to the continental scale, and also on merging terrestrial models with observations using parallel data assimilation technologies in supercomputing environments to arrive at an experimental monitoring system. Results from proof-of-concept global hydrologic simulations will be presented including coupled 3D groundwater and surface water flow. These frontier simulations are commensurate with high-resolution global atmospheric models that are coming online, thus, providing opportunity for scale-consistent closure of the terrestrial water and energy balance at the global scale. We will also show examples of model-data fusion like the assimilation of remotely sensed soil moisture or in situ groundwater level measurements into TSMP. These examples illustrate the importance of very high spatial resolution models, and scientific and applied challenges we still face in model-data fusion. Based on our experience with an existing experimental system, we conclude that many technologies are currently available to construct a terrestrial monitoring system and include novel satellite observational data sets in reconstructions at high spatial resolution. In this context, we will also discuss new concepts for performance portability that afford to port geoscientific software efficiently to the next-generation accelerated supercomputing hardware infrastructures.

The hydrological cycle is intensifying in West Africa due to both climate change and the anthropisation of land surfaces associated with high population rates. In this context, predicting the evolution of water resources over the next few decades is both a major societal and economic issue and a scientific challenge. The scientific difficulties include, for example, the paradoxical rise in groundwater levels over the past 30 years in the Sahel while rainfall has been falling (known as the Niamey paradox), the increase in the number of temporary pools, the recent change in the flow dynamics of the Niger and the groundwater-river interactions, which are significant in the south and low or non-existent in the north of West Africa.
To accurately represent these various hydrological behaviours over West Africa, it is necessary to use a model capable of simulating the evolution of soil water, river and groundwater dynamics and their interaction with the soil, rivers and atmosphere. In addition, it is necessary to use a fine spatial resolution in order to correctly represent the hydrological flows.
In this study, we used the parflow-CLM model over West Africa at high spatial resolution (1 km²). Over more than 3.5 million km², the model simulates the water dynamics over the 100 m depth during 2015 year. This simulation required 400,000 hours of CPU computing on the CEA's Irene-Rome computer as well as a spinup of 1700 years to reach the groundwater equilibrium (see figure).
The evaluation of such a simulation requires both in situ and satellite measurements. The latter allow to evaluate the model over very large areas compared to in situ measurements. In this presentation, we will show results on river water levels using altimetry measurements (Sentinel 3, Topex-Poseidon, Sarral-AltiKa), results on groundwater and soil water dynamics using gravity measurements (GRACE), and results on soil moisture dynamics using microwave measurements (SMOS).

Satellite Earth observations are an accurate and reliable data source for use in atmospheric and environmental science. Their increasing spatial and temporal resolution, as well as the seamless availability over ungauged regions, make them appealing for hydrological modeling. This work shows the results of the ESA DTE Hydrology project as a contribution to the development of a Digital Twin Earth focused on the water cycle and hydrological processes. It aims to highlight the potential of high-resolution satellite products in describing the water cycle and monitoring hydrological extremes and water resources. Through various dedicated experiments, we test the influence of five innovative high-resolution satellite-derived datasets on the performance of the distributed hydrological model Continuum set up for the entire Po River Basin, in northern Italy. The distributed hydrological model is forced, by high resolution satellite precipitation and evaporation, while high resolution satellite-derived soil moisture and snow water equivalent are ingested through a data-assimilation scheme. Further, satellite-based estimates of precipitation, evaporation and river discharge are used for hydrological model calibration and results compared with those based on conventional data. Despite the high density of conventional measurements and the strong human influence in the focus region, all satellite products are found to have strong potential in operational hydrology, with skillful estimates of river discharge throughout the model domain. Satellite based evaporation and snow water equivalent marginally improve (by 2% and 4%) the mean Kling-Gupta efficiency (KGE) at 27 river gauges, compared to a baseline simulation (KGE_mean=0.51) forced by high quality conventional data. Precipitation has the largest impact on model performance. Further, we take substantial steps towards a distributed hydrological model fully relying on satellite data as dynamic input, by calibrating and subsequently running the model using satellite precipitation and evaporation as forcing, and satellite-based estimates of river discharge as benchmark data to calibrate the model.

Terrestrial evaporation (E) is an essential component of the climate system that links the water and energy cycle. Conventional field-scale E formulations driven by remote sensing data of radiation, temperature and environmental properties have been used to develop global E datasets. However, these global datasets cannot be easily downscaled to the resolution required for local or regional-scale studies, since the meteorological input variables they use are often very coarse. Given these constraints, the following research questions arise: Can current global remote sensing-based E models be used to derive high-resolution retrievals at continental scales? Can those retrievals inform on the variability of root-depth soil moisture and E to local stakeholders? Here we present activities making use of the high-resolution observations of several geophysical variables provided through the Sentinel satellites and incorporating these into a state-of-the art global evaporation and soil moisture model, namely the Global Land Evaporation Amsterdam Model (GLEAM). By adapting the modelling framework to make use of the wide palette of Sentinel observations we produce E but also root-zone soil moisture data across Europe at 1 km.

In detail we present (a) the assimilation of Sentinel-1 backscatter through an Ensemble Kalman Filter using Machine-Learning as a forward operator, (b) the merging of geostationary radiation and land surface temperature retrievals at high temporal resolution with Sentinel-3 SLSTR LST retrievals to create a 1 km temperature and net radiation forcing. Initial results using Support-Vector-Regression as a forward operator show a higher degree of agreement between backscatter simulations and observations when compared to a semi-empirical implementation of the Water Cloud Model (WCM). Furthermore, correlations between E and soil moisture simulations against in situ data improve when using the 1 km modelling setup forced by the SLSTR based radiation and temperature data. The experiments are carried out with a modified version of GLEAM using a high-resolution land surface characterisation, e.g., soil texture, vegetation fractions, and the substitution of Vegetation Optical Depth (VOD) from the global model with Leaf Area Index (LAI) data for vegetation phenology.

The presented work contributes to the roadmap of providing E and soil moisture at 1km or beyond operationally across Europe. This allows us to bring recent research developments from the hydrology and climate community towards the actual water management and agricultural needs.

GloFAS Next: Towards hyper-resolution flood forecasting at global scales

Authors: Stefania Grimaldi(1), Carlo Russo(3), Ervin Zsoter(2), Peter Salamon(1), Cinzia Mazzetti(2), Christel Prudhomme(2), Corentin Canton De Wiart(2), Margarita Choulga(2), Damien Decremer(2), Juliana Disperati(1), Francesca Moschini(2)

(1) European Commission, Joint Research Centre, Ispra, Italy,
(2) European Centre for Medium-Range Weather Forecasts, Reading, UK
(3) Unisystems Luxembourg Sàrl

The Global Flood Awareness System (GloFAS) is a freely available flood forecasting service that is running fully operational as part of the Copernicus Emergency Management Service since April 2018. GloFAS offers a range of forecast products which are tailored to give a quick overview of current and future hydro-meteorological situation, and of the ongoing and upcoming flood events.

The aim of GloFAS is twofold: on the one hand to provide countries, i.e. national and regional authorities and services that have no or only limited flood forecast capacity with complementary, up-to-date flood forecast information for large (trans-national) river basins, and on the other hand to support international organizations such as e.g. the UN or the International Federation of Red Cross and Red Crescent Societies with forecast-based financing in decision making and preparatory measures before major flood events.

GloFAS products are generated using the open source hydrological model called LISFLOOD. LISFLOOD is an open-source, distributed, physically based rainfall-runoff model able to represent all the main hydrological processes: partition of precipitation into rain and snow, snow melting, canopy interception of rain, water infiltration into the soil, groundwater storage, surface runoff, lakes, dams, irrigation and other human water uses, flow in the rivers and in the floodplains. A set of maps showing the terrain slope, soil properties, land cover and land use features, allows to account for the morphological, physical, and land use characteristics of catchments and thus provide a more accurate representation of the rainfall-runoff processes in different climates and socio-economic contexts.

This presentation focuses on the upcoming major upgrade of the currently operational GloFAS set-up (v 3.1). Specifically, the spatial resolution of the entire system is doubled from 0.1 degrees (~10 km) to 0.05 degrees (~5km) which will allow a more detailed representation of the river network and of the spatial variability of catchments’ characteristics. A crucial feature of the upgrade is the use of the latest research findings and remote sensing datasets to prepare a set of completely revamped maps that are required as input for LISFLOOD. Examples include the latest land cover information and leaf area index layers from the Copernicus Global Land Service, global hydrology datasets based on MERIT Hydro (Yamazaki et al. 2019) or global soil data from ISRIC. The new set of maps together with a completely new hydrologic model calibration incorporating more than 1800 gauging stations will lead to more physically based parameters and to more accurate flood forecasts. The pre-operational release of the high resolution GloFAS set-up is planned for the end of 2022.