Trans-compartment transport of energy and matter during precipitation and flood extremes
Importance and motivation
Floods have a large variety of societal impacts that span across space and time and cause economic and environmental consequences (IPCC, 2012). The acceleration in population growth and changes in land-use patterns have increased human vulnerability to floods. Floods affect more people worldwide than any other natural hazard, and their global expected annual loss in the built environment is estimated at US$ 104 billion (GAR, 2015). For Germany, flooding has caused 50% of all economic losses due to natural hazards during the last 6 decades. Moreover, the flood in the Elbe and Danube river basins in 2002 was so far the most costly natural hazard for Germany. According to recent projections, climate change will be associated with a significant increase in frequency and intensity of floods due to extreme rainfall with so far unforeseeable consequences and feedbacks (Alfieri et al., 2016). These changes need to be considered when assessing future flooding risks, and respective prognostic tools need to be developed.
Currently, extreme precipitation and related river flooding is often investigated within the respective compartments, only considering compartment-related measurements, although flooding is a cross-compartment phenomenon. This focus on compartments causes significant knowledge gaps with respect to process understanding and prognoses. During flood events cascading effects lead to complex, and little understood, event-chains that affect multiple compartments in an unknown time-delay manner. As an example, floods triggered by synoptic storms have a strong potential to mobilize nutrients with direct consequences on the environmental health status of lakes and coastal environments due to the pulse-like supply of massive amounts of nutrients and carbon but also through harmful substances that are leached from soils into the terrestrial environment. Flooding increases the volume and velocity of the moving water and thus significantly decreases the retention time of these substances in the tributary waters and reduces their degradation and sedimentation prior to entering the coastal zones. Thus, flood induced matter transport from terrestrial to coastal environments is considered as a relevant process influencing the aquatic coastal and marine environment.
The propagation of extreme rainfall events through the catchment and the river system to the coastal zone exerts a decisive influence on the structure and long-term system functioning of the different environments along the flood path. Floods are drivers, for example, of landscape evolution as they have the potential to massively mobilize fertile top soils, are crucial for bedload transport in rivers and drive the morphological dynamics of running waters at the river reach and catchment scales (Nguyen et al., 2015, Voynova at al. 2017). Highly dynamic flood events drain large water masses to the estuary and coastal areas creating a bypass for many water constituents, like nitrate and phosphate, but also for contaminants. Under normal conditions, these constituents are filtered out by sedimentation or biological transformation and degradation before entering the coastal area. During a flood event, these constituents reach the coastal areas almost unfiltered and in extremely high concentrations the ecosystems are not adapted to. However, the long-term effects of flood-induced matter transport through the catchment-river-estuary system are not well understood.
Challenges and approach
The overall scientific aim of Show Case B is to better understand and quantify the coupled processes during and after flood events, including cascading short-term and long-term effects on the terrestrial and coastal systems. We are specifically interested in developing and integrating SMART Monitoring tools in/from WP1 into our investigation strategies, which will allow to access data from a wide range of sources, to use opportunistic sensors and to guide monitoring campaigns for optimal deployment of sensors in near real-time. For example, the accurate spatio-temporal quantification of extreme precipitation stays a scientific challenge which limits the skill of hydrological modelling (Moulin et al., 2009) and hence results in a significant uncertainty at the very basis of all research questions related to flooding. To overcome this shortcoming, we will include new ways of near-real time precipitation quantification by using Commercial Microwave Links (CMLs) as operated by commercial cellphone providers (Chwala et al., 2012, Chwala et al, 2016). The potential for streamflow prediction using CML was recently shown by Smiatek et al. (2017) and including such opportunistic sensors will strengthen SMART Monitoring efforts.
Furthermore, Data Exploration tools supplied by WP2 will allow integrating a large variety of data to obtain a comprehensive situational picture from different user perspectives. An important aspect is to enable and establish iterative ‘feedback loops’ between data acquisition and data analyses. In particular, this is necessary to adaptively decide how mobile sensor networks should be deployed and monitoring strategies adapted in the most effective and efficient way. We foresee that SMART Monitoring tools will help to guide data acquisition of specific field data (Tasks 1.1 to 1.3), harvest data from a variety of sources (Task 2.2), and feed them into Data Exploration Tools which provide information about data/knowledge gaps either by visual (Task 2.3) or computational data inspection (Task 2.4). We explicitly focus on a close feed-back between real-time data and an update of the knowledge/data bases to obtain a dynamic situational picture of processes, prognoses and warnings immediately before, during and after flood events.
Near real-time data integration is required as floods represent highly dynamic situations, and forecast and warning times of heavy precipitation events and flooding in the according mountainous regions are relatively short (max. 5-6 days for the upper Elbe River Catchment). Therefore, work flows need to account for real-time data ingestion of rapidly quality controlled and assessed data (Task 2.2) linked to a fast (automated) data analysis (Task 2.4). Another demanding challenge is to comprehensibly combine cross-compartment and cross-discipline process and property data with a wide range of space-time scales, from cm to km and from seconds to years. Being able to integrate data from different compartments (atmosphere, vegetation, soil, surface waters, groundwater, coastal waters), is a challenge for the visual methods and workflow concepts when aiming for identifying relevant coupling effects within the wide range of water and matter transport processes (sediment, contaminants, nutrients) in and in-between catchments, river system and coastal waters.
At the base of all investigations lies the question of "where and how to find and get the data"? The data are widely distributed throughout agencies and various scientific institutions and address a variety of disciplines, have different data formats, trustworthiness, and licensing regulations. This diversity in combination with the all-encompassing underlying data-diversity itself (e.g. satellite/airborne, maps, observation, simulation, social media, …), makes the data situation of Show Case B extremely heterogeneous and challenging. These complications, more than in Show Case A, link directly into Task 3.3 where stakeholder involvement is planned to foster innovations for societal benefits.
Data and Methods
Show Base B strongly links to E&E infrastructure project MOSES which was launched in 2017. The project aims at a synoptic understanding of flood events within the atmospheric, terrestrial, limnic and marine environments. In MOSES, the so-called event chain group ‘Hydrological Extremes’ comprises AWI, GEOMAR, HZG, UFZ, GFZ, HGMU, KIT, FZJ and DLR. Two main scientific questions are put forward here that drive this Show Case: A) To what extent do flood events occur in central Europe, which are "individually not extreme", but in the context of climate change happen in ever faster succession (for example, several times per year) and thus accumulate into a "cumulative event" with cumulative effects? B) How long do the different ecosystems (upper and lower river reaches, lakes, terrestrial sites and coastal areas) need to return to their "normal" steady-state mode after a medium "flood" event, and what happens if a minimum time needed for ‘recovery’ cannot be reached due to a rapid succession of minor and medium events? The Elbe catchment area has been identified as good case study region including the impact on the North Sea and thus being an ideal as a Show Case for Digital Earth. The Elbe River has been a hotspot of recent flood events with disastrous consequences in the past (2002, 2013). With 1,100 km length and a catchment size of about 140,000 km2, the Elbe is a large river in Europe. The German part of the river basin alone is inhabited by 20 million people (De Kok et al. 2000) and the average discharge near the estuary is about 700 m3s-1 (e.g. Krysanova et al. 2004) causing a significant effect on the coastal ecosystems.
The very close link to MOSES will allow to iteratively test feedback loops between the SMART Monitoring Design and the Data Exploration Framework under (near) real-time conditions. A reliable recognition of in-time series patterns from sensor measurements that employ user-tuneable algorithms for (pre)warnings building on constant checks, if sensor values are within an expected range (based on sensor specific properties) is fundamental. To achieve the scientific objectives, a set of data exploration methods and tools is needed including data integration, data selection and transformation, as well as pattern evaluation methods before data are visualized in different perspectives or assimilated in simulation models. Data integration involves combining heterogeneous data residing in different sources and providing the scientist with a unified access to the data. Data sources will include long-term observational data, e.g. online streamflow data from different agencies or from the TERENO observatories; forecast products from international and national centres (ECMWF, DWD, state flood forecasting centres) such as precipitation, temperature, streamflow, soil moisture. Further outputs from simulation models run by the Helmholtz centres are needed (e.g. TerrSysMP at FZJ, Drought Monitor at UFZ, Regional Flood Model RFM at GFZ), as well as satellite data whereby the most relevant ongoing missions are SMAP, MODIS, TerraSAR-X, Sentinel I and II. With regards to satellite data, the next suite of missions like Sentinel III-VI, EnMAP, GRACE-FO, or the planned TANDEM-L and ATMO-SAT missions already need to be integrated. Data selection and transformation implicates mainly the extrapolation of local fluxes and states in space and time, e.g. generation of dynamic maps including the entire catchment and receiving coastal waters (Up- and downscaling of information). Pattern evaluation mainly addresses the identification of patterns, relationships and causalities in high-dimensional surface data using advanced exploratory methods from data mining, machine learning, visual analytics etc.
As indicated by the data sets mentioned above, it becomes clear that flood research requires a close cooperation with external partners like state offices (Sachsen, Sachsen-Anhalt, Niedersachsen, HH, SH), federal institutes (BfG, BSH, BAW), forecasting centres (DWD, Flood forecasting centres of federal states) and others. Thus, this Show Case is a very good example to advance stakeholder involvement (=> WP3) and test data exchange and digestion of different data formats within the workflows and software tools developed in WP2. It links towards disaster management and the societal demand for science-based information about the causes and consequences of ongoing floods. This includes support to mitigate current disasters and to reduce negative consequences (soil removal, spread of contaminants, coastal ecosystem impacts, flooding of cities) in the future. The compilation of real-time information about ongoing floods will provide science-based facts about the flood event for areas that will/might be inundated and on consequences that have to be expected. We plan to provide such information to society and specific end-users. This information will complement the available information, as it will go beyond the sectoral information provided by forecasting institutions, such as the German Weather Service, Federal Institute of Hydrology or the flood forecasting and warning centres of federal states. To maximize outreach and to bundle the Helmholtz’s communication channels, the provision of data and information will be embedded in ESKP.