Latvian Region Charts Flexible Defense Against Climate-Driven Floods

In the flat lowlands of central Latvia, water has always been both a blessing and a menace. The Zemgale region, crisscrossed by the Daugava and Lielupe rivers and blanketed with some of the country’s most fertile farmland, faces a future in which its flood patterns are changing faster than its infrastructure can adapt. A new study published in Frontiers in Environmental Economics lays out one of the most detailed attempts yet to plan for that future, combining participatory planning with computational optimization to sequence flood defenses across eight decades.

The research, conducted within the European Union’s Horizon 2020 IMPETUS project, tackles a deceptively simple question: how should a region with limited budgets and uncertain climate futures decide which flood protection measures to build, and when to build them? The answer, according to the team led by Kristina Veidemane of the Baltic Environmental Forum in Riga, is not a single master plan but a set of thirty-seven alternative adaptation pathways, each a different ordered sequence of interventions designed to be adjusted as conditions evolve.

Zemgale is unusually exposed. Covering 10,742 square kilometers and home to roughly 222,000 people, the region consists of flat terrain, a dense river network, and extensive floodplains. Forty-three percent of its land is devoted to intensive agriculture, dominated by grain and rapeseed monocultures, while forests cover another forty percent. Spring floods, driven by melting snow and ice, regularly swell rivers across the entire region. Increasingly, however, the greater threat to its urban centers comes from intense summer rainfall that overwhelms drainage systems and triggers flash floods, a hazard amplified by impervious surfaces and aging drainage networks.

The study’s methodological framework rests on the concept of adaptation pathways, a planning approach that treats climate adaptation not as a one-time investment but as a dynamic sequence of decisions. At its core are adaptation tipping points, the conditions under which a given measure no longer meets its objectives. Rather than fixing those points in advance, the researchers let an optimization process determine the timing of interventions based on continuous risk indicators, intervention characteristics, and stakeholder criteria. The approach explicitly leaves options open for further adaptation depending on how the future unfolds, echoing decision frameworks pioneered in projects such as the Thames Estuary 2100 plan.

Central to the work was a co-design and co-creation process involving stakeholders from every level of governance, including all six municipal authorities, regional institutions, farmers, environmental NGOs, and national flood management bodies. Between 2023 and 2025, participants gathered in a series of workshops where moderated discussions and World Café sessions were used to identify and refine adaptation options. Eleven measures ultimately made the shortlist, ranging from soft, non-structural interventions like improved early warning systems and construction restrictions in flood-prone areas to structural works such as drainage upgrades, riverbed cleaning, urban rainwater collection capacity, and nature-based solutions for water retention in both rural and urban settings.

Each option was scored against eight criteria, including effectiveness, lifetime, technical and social readiness, costs, co-benefits, hidden risks, and implementation time. The results revealed a telling tension. Long-established measures such as maintaining open ditches, drainage systems, and riverbed dredging, practices implemented in Latvia for decades, scored highest across most criteria and enjoyed strong stakeholder acceptance. By contrast, restricting development in flood-risk zones received the lowest scores for technical and social readiness, reflecting stakeholder resistance and limited political appetite for restrictive regulation. Stakeholders also expressed skepticism toward permeable pavements and urban nature-based solutions, citing maintenance burdens and space constraints in historically developed town centers.

The quantitative core of the study involved two linked modeling tools developed within IMPETUS. The Risk and Emerging Problems Inference Engine, or REPIE, computes climate risk as the geometric mean of hazard, vulnerability, and exposure components, using alarm thresholds to classify threat levels from least concern to critical. Its output feeds the Dynamic Contingency Response Tool, or DCRT, which applies the NSGA-II multi-objective genetic algorithm to identify Pareto-optimal sequences of interventions, balancing risk reduction against costs, feasibility, and externalities. The models ran under three climate and development scenarios, SSP1, SSP2, and SSP5, with risk projections drawn from the CMCC Earth System Model.

The scenario results were stark. Under the mildest pathway, SSP1, risk rises only about six percent above the 2021 baseline, suggesting existing measures may suffice. Under SSP2, risk climbs roughly seventeen percent, while under the fossil-fueled SSP5 scenario it surges by thirty-one percent by the end of the century. In the baseline scenario, the region’s flood risk index, estimated at a moderate 4.5 on a normative one-to-ten scale, is projected to reach 6.2 by 2100 if no additional action is taken. Under SSP5, eleven of the modeled pathways, each comprising only two or three interventions, fail to keep risk at or below its initial value; only the most comprehensive twelve- and thirteen-intervention pathways cut risk below half of the 2021 level by 2050.

Costs varied dramatically across the thirty-seven pathways. The most ambitious, P_1, delivers a cumulative risk reduction benefit of 235.2 units at an estimated total cost of 73.1 million euros, while the leanest options achieve modest benefits for under seven million euros. Comparing benefits per million euros invested revealed four distinct clusters, from low-benefit, low-efficiency combinations that the authors recommend rejecting, to a sweet spot of fourteen pathways pairing five to ten measures with above-average performance on both counts. Notably, cost-efficiency did not scale simply with the number of interventions, underscoring that the type and combination of measures matter as much as their quantity. A five-measure pathway, P_16, achieved the greatest short-term reduction by 2030 and the best benefit per million euros, and stakeholders ultimately preferred it over the more powerful but costlier P_1.

The study’s findings fed directly into the Zemgale Regional Climate Change Adaptation Plan, adopted in March 2026 as the region’s first such document. The authors caution that their simplified modeling approach trades hydrological precision for computational speed, and that sparse data on the long-term effectiveness of nature-based solutions remains a limitation. Yet the broader lesson travels well beyond Latvia: the pathway with the highest modeled benefit is not always the one communities will choose, because adaptation decisions weigh social readiness, feasibility, and local priorities alongside technical performance. For flood-prone regions across Northern Europe and beyond, the message is to start feasible, cost-effective actions now, while preserving the flexibility to escalate as the waters rise.

Subject of Research: Climate adaptation pathways for flood risk management in the Zemgale region of Latvia

Article Title: Adaptation pathways to address climate change-induced flood risk in the Zemgale region, Latvia

Article References: Adaptation pathways to address climate change-induced flood risk in the Zemgale region, Latvia. (n.d.). https://doi.org/10.3389/frevc.2026.1931037

Image Credits: AI Generated

DOI: 10.3389/frevc.2026.1931037

Keywords: climate change adaptation, flood risk, adaptation pathways, Zemgale, Latvia, nature-based solutions, stakeholder engagement, SSP scenarios, multi-criteria analysis, NSGA-II optimization, flood modeling, regional planning