ALBERT

Description: Increasing amounts of data, together with more computing power and better machine learning algorithms to analyse the data, are causing changes in almost every aspect of our lives. This trend is expected to continue as more data keep becoming available, computing power keeps improving and machine learning algorithms keep improving as well. Flood risk and impact assessments are also being influenced by this trend, particularly in areas such as the development of mitigation measures, emergency response preparation and flood recovery planning. Machine learning methods have the potential to improve accuracy as well as reduce calculating time and model development cost. It is expected that in the future more applications will become feasible and many process models and traditional observation methods will be replaced by machine learning. Examples of this include the use of machine learning on remote sensing data to estimate exposure and on social media data to improve flood response. Some improvements may require new data collection efforts, such as for the modelling of flood damages or defence failures. In other components, machine learning may not always be suitable or should be applied complementary to process models, for example in hydrodynamic applications. Overall, machine learning is likely to drastically improve future flood risk and impact assessments, but issues such as applicability, bias and ethics must be considered carefully to avoid misuse. This paper presents some of the current developments on the application of machine learning in this field and highlights some key needs and challenges.

Description: Water authorities responsible for water quantity and water quality management may strongly influence the magnitude of greenhouse gas emissions from the surface waters and the adjacent peat areas within their territories. Climate smart water management (reducing influx of organic matter and improving water quality) is therefore a potentially strong mitigation tool. We hypothesize that climate smart water management has a stronger mitigation potential than reducing emissions from the operational management of a Water Authority. Based on literature data on greenhouse gas emissions from ditches and agricultural peatlands, we present a case study of a Dutch Water Authority – Amstel, Gooi and Vecht (operated by Waternet). We estimate that greenhouse gas emissions from the 195 km2 large peat area within its territory are 470 kt CO2-eq per year. An additional 231 kt CO2-eq yr−1 is emitted from the water bodies within the 102 km2 large water area territory. Both emissions are considerably higher than the estimated climate footprint of the operational management of the water board (∼62 kt CO2-eq per year in 2017). While Waternet strives to have a net zero emission of greenhouse gases related to its operational management by 2020, we postulate that measures (to be taken before 2030) such as the prevention of organic matter and nutrients entering surface waters, the removal of organic carbon from ditches and higher groundwater levels in agricultural peatlands, may reduce greenhouse gas emissions in ditches and agricultural peat meadows with 26 and 27 kt CO2-eq per year, respectively. Measures that are taken to reduce greenhouse gas emissions in water bodies are expected to have a positive impact on water quality as well.

Description: Waternet is the executive agency of the regional water authority Amstel, Gooi and Vecht. Water authority Amstel, Gooi and Vecht manages the water levels (ditches) for 19 400 ha of peat meadows around the Netherlands capital Amsterdam. At present the ditches levels at about 40–60 cm beneath the peat meadow surface, resulting in a groundwater level between from 30 until 80 cm below peat surface and a subsidence of about 9 mm each year. A study was carried out on peat soil subsidence in the Amstel, Gooi and Vecht water authority water management area towards 2100: for short term effects (until 2027), midterm effects (until 2050) and longer term effects (until 2100). This study explores 4 scenarios: (1) present policy (maintain ditch waterlevel at maximum 60 cm below surface); (2) active rewetting, groundwater level at surface; (3) passive rewetting, subsidence is not compensated by lowering of water levels; (4) subsurface irrigation by submerged drains (infiltration in summer, drainage in winter). The scenarios are compared on farming, houses, public infrastructure, greenhouse gases and water management. At present, the total net benefit for farmers are EUR 7 million per year for the whole area, while the costs for the water authority are EUR 37 million per year for managing ditches, dikes and pumps. Costs for greenhouse gases are EUR 18 million (at a price of EUR 40 per ton CO2-eq). Active rewetting would reduce soil subsidence maximally from 2 to 0.5 m towards 2100 but reduces the benefits for farming, whilst the costs for water management stay alike. The costs for greenhouse gases however drops with EUR 3 million per year immediately because CO2-eq emissions drops. Best (financial) results (with respect to all stakeholders) on the long term are booked by passive rewetting with lower costs for water management, houses, public works and greenhouse gases. This scenario will eventually take away the farming possibilities, but not before 2050 and could be too slow to contribute strongly to Paris agreement goals. Best result with respect to climate for short and long term is active rewetting, which will drop the greenhouse gas emissions strongly (equivalent of EUR 2.3 million per year), reduce soil subsidence, but makes farming harder (drop from 7.1 up to EUR 2.5 million per year benefit) and brings no direct reduction of costs for the water authority. Best result on short term for farmers is submerged infiltration drains. However, the effect of this scenario on GHG emission is limited in this study.

Description: Soil subsidence is one of the major issues in the management area of the water authority Amstel, Gooi and Vecht, including emissions of greenhouse gases. This paper describes four different methods to calculate these emissions in agricultural peat meadows, based on (1) the mean lowest groundwater level, (2) the mean groundwater level, (3) the subsidence rates and (4) general numbers. The emissions were calculated in two polders (about 2600 ha peat meadow), these were comparable for all methods, ranging from 42 up to 50 kton CO2-eq yr−1 (based on data of 2015), which is about 14.5 up to 19 t CO2-eq ha−1 yr−1. Besides, the greenhouse gas emissions were compared for different policy scenario's in one polder subunit (283 ha): (1) standard policy (lowering surface water level at the same rate as soil subsidence taking place), (2) passive rewetting (surface water level fixation), (3) subsurface irrigation by submerged drains, and (4) a maximum surface water level decrease of 6 mm yr−1. Comparing the four policy scenario's in one polder subunit, greenhouse gas emissions were lowest in case of subsurface irrigation, decreasing greenhouse gas emissions by about 35 %–50 % in this polder compared to standard policy, meaning a decrease of about 5.5–9.3 t CO2-eq ha−1 yr−1. This represents a value of about 550–930 EUR ha−1 yr−1 (at a price of EUR 100 per ton CO2-eq). The scenario passive rewetting leads to a decrease of about 12 %–21 %, or 2–3 t CO2-eq ha−1 yr−1 compared to standard policy. The estimation of the decrease in GHG emissions depends on the assumptions made. In this study it was assumed that subsurface irrigation halves soil subsidence. The water board will use the described procedures to estimate greenhouse gas emissions in the future to support water level management in areas with peat soils.

Description: Volcanic eruptions comprise an important airborne hazard for aviation. Although significant events are rare, e.g. compared to the threat of thunderstorms, they have a very high impact. The current state of tools and abilities to mitigate aviation hazards associated with an assumed volcanic cloud was tested within an international demonstration exercise. Experts in the field assembled at the Schwarzenberg barracks in Salzburg, Austria, in order to simulate the sequence of procedures for the volcanic case scenario of an artificial eruption of the Etna volcano in Italy. The scope of the exercise ranged from the detection (based on artificial observations) of the assumed event to the issuance of early warnings. Volcanic-emission-concentration charts were generated applying modern ensemble techniques. The exercise products provided an important basis for decision-making for aviation traffic management during a volcanic-eruption crisis. By integrating the available wealth of data, observations and modelling results directly into widely used flight-planning software, it was demonstrated that route optimization measures could be implemented effectively. With timely and rather precise warnings available, the new tools and processes tested during the exercise demonstrated vividly that a vast majority of flights could be conducted despite a volcanic plume being widely dispersed within a high-traffic airspace over Europe. The resulting number of flight cancellations was minimal.

Description: Flood damage assessment is usually done with damage curves only dependent on the water depth. Recent studies have shown that data-mining techniques applied to a multi-dimensional dataset can produce significantly better flood damage estimates. However, creating and applying a multi-variable flood damage model requires an extensive dataset, which is rarely available and this can limit the application of these new techniques. In this paper we enrich a dataset of residential building and content damages from the Meuse flood of 1993 in the Netherlands, to make it suitable for multi-variable flood damage assessment. Results from 2D flood simulations are used to add information on flow velocity, flood duration and the return period to the dataset, and cadastre data is used to add information on building characteristics. Next, several statistical approaches are used to create multi-variable flood damage models, including regression trees, bagging regression trees, random forest, and a Bayesian network. Validation on data points from a test set shows that the enriched dataset in combination with the data-mining techniques delivers a significant improvement over a simple model only based on the water depth. We find that with our dataset, the trees based methods perform better than the Bayesian Network.

Description: This paper addresses the large differences that are found between damage estimates of different flood damage models. It explains how implicit assumptions in flood damage functions and maximum damages can have large effects on flood damage estimates. This explanation is then used to quantify the uncertainty in the damage estimates with a Monte Carlo analysis. The Monte Carlo analysis uses a damage function library with 272 functions from seven different flood damage models. The paper shows that the resulting uncertainties in estimated damages are in the order of magnitude of a factor of 2 to 5. The uncertainty is typically larger for flood events with small water depths and for smaller flood events. The implications of the uncertainty in damage estimates for flood risk management are illustrated by a case study in which the economic optimal investment strategy for a dike segment in the Netherlands is determined. The case study shows that the uncertainty in flood damage estimates can lead to significant over- or under-investments.

Description: This paper addresses the large differences that are found between damage estimates of different flood damage models. It explains how implicit assumptions in flood damage models can lead to large uncertainties in flood damage estimates. This explanation is used to quantify this uncertainty with a Monte Carlo Analysis. As input the Monte Carlo analysis uses a damage function library with 272 functions from 7 different flood damage models. This results in uncertainties in the order of magnitude of a factor 2 to 5. The resulting uncertainty is typically larger for small water depths and for smaller flood events. The implications of the uncertainty in damage estimates for flood risk management are illustrated by a case study in which the economic optimal investment strategy for a dike segment in the Netherlands is determined. The case study shows that the uncertainty in flood damage estimates can lead to significant over- or under-investments.

Description: Volcanic eruptions comprise one of the most important airborne hazards for aviation. Although significant events are rare, they have a very high impact. The current state of tools and abilities to mitigate aviation hazards associated with an assumed volcanic cloud was tested within an international demonstration exercise. Experts in the field assembled at the Schwarzenberg barracks in Salzburg, Austria, in order to simulate the sequence of procedures for the volcanic case scenario of an artificial eruption of Etna volcano in Italy. The scope of the exercise ranged from the detection of the assumed event to the issuance of early warnings. Volcanic emission concentration charts were generated applying modern ensemble techniques. The exercise products provided an important basis for decision making for aviation traffic management during a volcanic eruption crisis. By integrating the available wealth of data, observations and modelling results directly into a widely used flight planning software, it was demonstrated that route optimization measures could be implemented effectively. With timely and rather precise warnings available, the new tools and processes tested during the exercise demonstrated vividly that a vast majority of flights could be conducted despite a volcanic plume widely dispersed within a high-traffic airspace over Europe. The resulting number of flight cancellations was minimal.

Description: Flood damage assessment is usually done with damage curves only dependent on the water depth. Several recent studies have shown that supervised learning techniques applied to a multi-variable data set can produce significantly better flood damage estimates. However, creating and applying a multi-variable flood damage model requires an extensive data set, which is rarely available, and this is currently holding back the widespread application of these techniques. In this paper we enrich a data set of residential building and contents damage from the Meuse flood of 1993 in the Netherlands, to make it suitable for multi-variable flood damage assessment. Results from 2-D flood simulations are used to add information on flow velocity, flood duration and the return period to the data set, and cadastre data are used to add information on building characteristics. Next, several statistical approaches are used to create multi-variable flood damage models, including regression trees, bagging regression trees, random forest, and a Bayesian network. Validation on data points from a test set shows that the enriched data set in combination with the supervised learning techniques delivers a 20 % reduction in the mean absolute error, compared to a simple model only based on the water depth, despite several limitations of the enriched data set. We find that with our data set, the tree-based methods perform better than the Bayesian network.