Climate change has led to an increased frequency of extreme rainfall events, creating significant challenges for polder regions where water has to be pumped out actively. Tractor pumps stand out as a quick, flexible measure to improve polder discharge capacities. However, in extreme rainfall events, where the need for tractor pumps exceeds available supply, strategic decisions must be made on where to deploy them. This requires a data-driven methodology to support decision making. To achieve this, flood modelling, damage assessment and pump allocation optimization are combined in this study. The study aimed to create a framework that combines these in one integrated model and allocates a limited set of tractor pumps to the polders where they reduce total economic losses most. The study focused on the 18-20 June 2021 flood event, using 20 tractor pumps and 48 selected polders in Hoogheemraadschap Hollands Noorderkwartier. Polder flood damages were quantified through Depth Damage Curves (DDCs), incorporating terrain and land use data from the WaterSchadeSchatter. A two-stage Mixed Integer Linear Program was then formulated to determine optimal placement of tractor pumps, where the First-Stage selected polders and the Second-Stage assinged pumps. The model systematically evaluated all possible allocation options, identifying which placements minimized total damage.

Key findings were that DDCs are not suited for assessing the impact of tractor pumps on polders in linear programming models, as the relation between volume and water levels in a polder are nonlinear. Instead, Volume Damage Curves (VDCs) are more appropriate, as they are able to quantify damage per cubic meter, the variable that pumps directly influence. VDC derivatives were used to classify polders into three types: Type 1 (always relevant), Type 2 (tipping point dependent), and Type 3 (low priority). Of the 48 polders included in the study, 25 were classified as Type 3. Of these, only 2 received pumps, and in both cases the prevented damage was minimal. In contrast, Type 1 and tipping point exceeding Type 2 polders accounted for nearly all significant damage reduction. This suggests that Type 2 polders below their tipping point, and Type 3 polders can be used for polder deselection and as an alternative for the First-Stage model.

A shortcoming was that with the current VDC use, the maximum water volume is the primary driver of damage, as the flood duration is assumed fixed. For agricultural and infrastructural areas, flood duration strongly influences economic losses, suggesting that both the VDC construction and use in the optimization model must be altered to incorporate duration as an influencing variable. VDCs that do so require the direct damage term of every polder to be corrected for the flood duration. This can be done for specific polder increments, where each increment is multiplied with a duration factor. After every model run the accumulated duration for each increment should be stored and passed to the next run, allowing the model to account for ongoing flooding.

Modeling duration in linear programming greatly increases the number of variables and constraints. To keep the enlarged formulation solvable, the pump placement variable should be aggregated by counting pumps only per type and time step instead of individual pump tracking. This change removes the distinction between the First- and Second-Stage, rendering polder subset selection by the First-Stage infeasible. Instead, the polder classification types can be used for subset selection before the solver starts, so the enlarged single stage model still finishes in time for operational use.

To support real-time decisions, HHNK should develop a short horizon model that integrates improved VDCs, forecasted rainfall, current water levels converted to polder volumes, and current pump placements. As new data becomes available, the model should update pump allocation accordingly. The current optimisation model can serve as a starting point for this operational tool.

The research question addressed in this study is ”To what extent does the presence of a cloud forest canopy impact the Mestelá River catchment and how will this affect the various involved stakeholders?”. The study aims to investigate the importance of cloud forests in the Mestelá River catchment, Alta Verapaz, Guatemala, related to water security and the social impact of cloud forest conservation and management. The research methods used in this study were a combination of quantitative and qualitative methods.

Cloud forests play a vital role in regulating water flow in catchments. The Mestelá River catchment, where the NGO Community Cloud Forest Conservation (CCFC) is situated, is the focus of this research. The project’s primary aim was to establish a long-term canopy setup, ensuring future data collection. The project’s scope encompasses a range of methodologies, including the installation of a long-term measurement station in the canopy, computation of the Mestelá River discharge, the development of a rating curve, and the utilisation of a FLEX-Topo model to simulate the hydrological cycle in the catchment. Additionally, a stakeholder management analysis was conducted to understand the complex impact of cloud forests (conservation) on various stakeholders.

The study did not explicitly formulate any hypotheses, but the findings provide evidence for the impact of cloud forest canopies on river catchments and discharge. The study also has limitations, including the small sample size and the lack of long-term data. However, the study provides valuable insights into the importance of cloud forest ecosystems for water security and the social impact of cloud forest conservation and management. The stakeholder analysis reveals that for CCFC two methods of advocacy can be used. Whilst the CCFC is effective in bottom-up engagement with the community, in addition, a strip for small children was constructed. For top-down advocacy, using the FLEX-Topo
model for visualising water security in combination with cloud forest protection holds promise.

The implications of this work are substantial for cloud forest conservation and associated ecosystems. The findings offer valuable insights for developing effective conservation strategies that consider the canopy’s impact on the catchment and its stakeholders. It is important to note that the FLEX-Topo model is currently conceptual and requires further refinement and detail for the Mestelá River catchment. Nevertheless, this study contributes significantly to the understanding of cloud forest ecosystems and offers practical and theoretical applications for future research and conservation efforts.