| Popis: |
Floods are among the most frequently occurring and most damaging natural hazards in Europe, impacting heavily on economies and communities through a loss of life, property and livelihood. Moreover, an increasing trend in economic flood damages has been observed in the past decades, mainly driven by socioeconomic developments, such as a rising standard of living and urbanization in flood-prone areas. This trend prompted a shift in flood management from flood prevention, relying on structural defense measures, to a more integrated, system-wide approach focused on managing flood risk, which is defined as a combination of the flood hazard or probability and its potential damages. As such, sustainable flood risk management aims to efficiently reduce the societal, environmental and economic impacts of floods, for instance by looking for synergies with land use systems mitigating downstream flood risk. Land use systems, and their spatial configuration in the catchment, can impact flood risk by influencing the hydrological processes. Vegetated surfaces, such as forest and arable cover crops, have the capacity to mitigate flood hazard by increasing water retention and infiltration, thereby reducing the fraction of rapid surface runoff downstream. Conversely, sealing pervious soil surfaces exacerbates flood risk by inhibiting retention and infiltration, thereby increasing surface runoff. This capacity of land use systems to reduce or increase flood risk can be interpreted as a positive or negative flood insurance value. This research project aimed to support nature-based flood risk management. For this purpose, a spatial optimization and comparative flood risk assessment framework were developed to identify the most suitable locations in a catchment for land use changes mitigating flood hazards downstream, and to quantify the flood insurance value of these land use changes. The developed frameworks were illustrated for study areas in Flanders, the northern region of Belgium, which has a high flood risk across its entire territory, as it is both flood-prone and characterized by a high degree of urbanization. First, the impact of soil sealing on flood severity was assessed in data-driven analyses based on the Flemish spatial flood archive, containing records of flood extents dating back to 1988. Flooded area extents and corresponding flood volumes from this archive were analyzed along with time series of rainfall and land use for three middle-sized river subcatchments using linear regression and two machine learning methods, Support Vector Regression and Boosted Regression Trees. The machine learning methods were found suitable for this type of study, since their flexibility allows for spatially explicit models with larger sample sizes. However, the relationship between soil sealing and flood volume and extent could not be confirmed by the empirical analyses. The analyses were mainly limited by the length of the time series, limiting the number of observations. Additionally, uncertainties and possible inaccuracies associated with the recorded historical flood extents and inconsistencies in the land use classifications also impeded the data-driven analyses. It is therefore stressed that continued consistent monitoring of floods and land use changes is required. As land use changes impact rainfall-runoff interactions in the catchment, spatially distributed hydrological models are needed to assess the hydrological impact of land use changes. However, due to their large computational burden, such hydrological models are typically applied in scenario-analyses, assessing 'what-if' problems requiring a limited number of model simulations. To address 'where should' questions, iterative spatial optimization analyses are required with a high number of model simulations to identify the spatial configuration of certain land use interventions promoting water retention and infiltration. The computational demand of such analyses can be reduced by implementing heuristic algorithms, limiting the solution space and approximating the optimal solution or by integrating a more computationally efficient and sufficiently accurate hydrological model. The latter approach was adhered to in this research whereby a computationally efficient and spatially-explicit RR-model was developed, taking into account the spatial interactions between surface runoff generation, propagation and re-infiltration along the flow paths. The widely used Soil Conservation Service Curve Number (SCS-CN) method was used as a basis to formulate eighteen raster-based model configurations, testing different values for three model parameters in combination with two methods considering antecedent soil moisture conditions (AMC) and three re-infiltration algorithms. These model configurations were evaluated for three catchments, resulting in NSE values of 0.57, 0.56 and 0.64 for the most performant model configuration, implementing a λ parameter value of 0.05, the AMC correction method of the Soil and Water Assessment Tool (SWAT) and the re-infiltration algorithm devised by Van Loo (2018) with a hydraulic radius of 3 mm and a seasonally adjusted Manning's roughness coefficient. This model was judged sufficiently accurate to assess and compare the off-site hydrological impacts of land use alternatives. Consequently, the developed rainfall-runoff model was integrated in an iterative optimization framework to address the question of where to implement certain land use interventions in order to most effectively minimize runoff accumulation, and thus flood hazard, in a downstream point of interest. This optimization framework iteratively ranks the performance of all alternative locations, while taking into account spatial interactions. The framework was tested for two medium-sized catchments for three land use change scenarios: afforestation, soil sealing and the implementation of winter cover crops. Results show the considerable impact of these land use changes and their locations on runoff accumulation at the downstream point of interest, with the priority locations having the greatest impact on downstream runoff volume and providing an indication on how to achieve a maximum impact on flood hazard with a minimum extent of the land use change under consideration. The priority locations for afforestation are characterized by high flow accumulation, highlighting the importance of enhancing the infiltration capacity in river valleys. Conversely, soil sealing is to be avoided in these locations and confined to locations upstream in the catchments. Finally, the impact of land use changes at locations determined by the optimization framework were evaluated in a comparative flood risk assessment framework. In this framework, the relative economic impact of land use changes on flood damages and corresponding flood risk is determined, thus allowing for an explorative assessment of the efficiency of the proposed land use changes as flood mitigation measures. This flood risk assessment was illustrated for one study area, with the results showing that afforestation in river valleys corresponds to a large flood insurance value, while soil sealing in the upstream areas only results in a limited increase of flood risk. In conclusion, the iterative optimization framework allows for the identification of the most effective locations for flood hazard mitigation through a particular land use change in catchments, while the comparative flood risk assessment allows for the calculation of the flood insurance value associated with land use changes mitigating flood risk. The generic frameworks can be applied to small to medium-sized, hilly catchments; and was illustrated for such catchments in Flanders, Belgium. The results consistently show the importance of river valleys in mitigating downstream flood hazards, as these areas are to be avoided for soil sealing and prioritized for afforestation, leading to a high potential flood insurance value. Consequently, the developed frameworks can be applied to build support for spatial planning initiatives integrating sustainable flood risk management, a pertinent policy topic, both in Flanders and in wider Europe. status: published |
