Sea-level rise (SLR) caused by climate change is projected to significantly impact global migration patterns, primarily through increased internal migration within countries. The intensity and spatial distribution of this migration will depend on the magnitude of SLR, future socioeconomic development, and the adaptation strategies implemented to reduce coastal vulnerability. While most current internal migration is rural-to-urban, climate change is expected to intensify this trend, particularly in coastal areas. Previous studies have often focused on estimating the total number of climate migrants, and lack a detailed examination of the spatial patterns of climate change-induced internal migration in general and SLR-related migration specifically. This study addresses this gap by using a spatially explicit model, CONCLUDE, in the Mediterranean region to analyze the interplay between SLR, adaptation policies (including protection, accommodation, and managed retreat), and the resulting migration patterns. This region is highly vulnerable due to its high population density and urbanization along the coast. The model is calibrated and validated for the Mediterranean to account for the differences in socioeconomic development and adaptive capacity between the northern and southern/eastern parts of the region. The study leverages existing scenario frameworks in climate change research (RCPs, SSPs) and introduces specific shared policy assumptions (SPAs) for coastal adaptation to generate a range of integrated scenarios to explore the potential range of outcomes.

The existing literature acknowledges the strong link between climate change and migration, anticipating increased migration flows globally. The impact of climate change extends beyond environmental factors like extreme weather events, impacting all drivers of migration (economic, political, social, demographic). While international migration is anticipated, the majority of climate migration is likely to be internal, over short distances, and from rural to urban areas, trends expected to worsen with climate change. Several reports project substantial numbers of climate migrants globally. While SLR-related migration is receiving increased attention, due to impacts like land submergence, saltwater intrusion, and coastal flooding, most studies assume autonomous migration of the affected population. Fewer studies analyze how different adaptation strategies influence SLR-related internal migration, particularly at supra-national scales. A key gap is the lack of research exploring the combined effects of protection, accommodation, and managed retreat policies on SLR-induced internal migration. Moreover, previous studies have typically focused on the total number of migrants, instead of the spatial distribution. Analyzing spatial patterns is crucial for identifying migration hotspots and informing decision-making related to anticipating and managing these flows. The lack of supra-national scale studies analyzing spatial patterns necessitates further research into the potential for a "safe development paradox," where protected areas become more attractive, leading to increased exposure despite mitigation efforts.

This study employs the spatially explicit model CONCLUDE, an extension of INCLUDE, to project internal migration patterns due to SLR in the Mediterranean region. CONCLUDE is a gravity-based population downscaling model that produces spatial population projections at a resolution of 30 arc seconds. It accounts for inland-coastal migration alongside rural-urban migration and considers factors like population potential, distance decay, and local attractiveness. The model integrates several data sources: Sea-level rise projections from the IPCC, based on representative concentration pathways (RCPs) 2.6, 4.5, and 8.5. Socioeconomic projections are based on Shared Socioeconomic Pathways (SSPs) 1, 3, and 5, reflecting a range of socioeconomic conditions and adaptive capacities. Shared Policy Assumptions (SPAs) for coastal adaptation were specifically developed for this study, including scenarios for protection, accommodation, and managed retreat. Three integrated scenarios were selected to cover a range of SLR, socioeconomic conditions, and adaptation strategies. A "bathtub" approach was used to model submergence due to SLR, using digital elevation models (MERIT DEM) and considering land in hydrological connection to the sea. Adaptation policies were modeled by generating spatially explicit raster layers for each policy type (hard protection, accommodation with setback zones and wetland restoration, and managed retreat). The effects of adaptation are quantified by comparing the "with adaptation policies" scenarios against the "no adaptation policies" reference scenarios. Migration hotspots were identified by analyzing the upper and lower 10% of in- and out-migration cells across all scenarios to assess robustness. The study acknowledges uncertainties related to scenario assumptions, the modeling approach, and input data; the results are presented as plausible trends rather than precise predictions.

The study's key findings are as follows: 1. **Total Migrant Numbers:** Without adaptation, over 20 million internal migrants could result from SLR by 2100 in the Mediterranean, largely regardless of the specific SSP-RCP combination, although the timing of the increase varies. Southern and eastern Mediterranean countries experience far higher projected migration compared to the north (roughly three times higher). Adaptation policies significantly reduce the total migrant numbers, by factors ranging from 1.4 to 9, depending on the policy scenario. The 'Hold the Line' scenario (SSP5-RCP8.5), focusing on large-scale hard protection, dramatically reduces migration (to ~2 million), while 'Build with Nature' (SSP1-RCP2.6), emphasizing diverse strategies, shows a more gradual increase in migrants. 2. **Spatial Migration Patterns:** Without adaptation, the model projects out-migration from a narrow coastal strip and widespread in-migration into inland urban areas. These spatial patterns remain consistent across scenarios, even with adaptation, although less pronounced in scenarios with effective adaptation strategies. The 'Hold the Line' scenario, with extensive hard protection, shows less pronounced migration overall. 3. **The Levee Effect:** The implementation of adaptation policies reverses some of the spatial patterns. Adaptation mainly leads to increased population concentration in protected coastal areas ("levee effect"). This effect is most significant under "Save Yourself" (SSP3-RCP4.5) in the South and East, and under "Hold the Line" in the North. Migration towards protected coastlines mostly occurs from nearby areas and larger inland cities. 4. **Robustness:** Analysis of migration hotspots shows remarkable robustness in overall spatial migration patterns across all scenarios, both with and without adaptation policies. Consistent out-migration from a narrow coastal strip and widespread in-migration into urban settings are observed in most scenarios. 5. **Urban Migration:** A significant portion of migrants are projected to move to urban areas. Without adaptation, 70-86% of migrants are projected to be urban migrants, while with adaptation, this share is somewhat lower (56-68%). This is due to the assumption that most protection strategies are implemented in urban settings.

The study's findings underscore the significant impact of SLR on internal migration in the Mediterranean and highlight the crucial role of adaptation policies in shaping both the number and spatial distribution of migrants. The robustness of spatial patterns, even under diverse adaptation strategies, emphasizes the importance of proactive planning. The "levee effect" highlights a potential risk of maladaptation, where protected areas attract further settlement, increasing future exposure and risk. The study's results demonstrate that socioeconomic conditions significantly influence migration outcomes, exceeding the impact of SLR alone, highlighting the need to consider broader social and economic factors in planning. The results are presented as plausible trends rather than precise predictions due to inherent uncertainties in the data and models used, but they provide valuable insights into potential future migration scenarios.

This study demonstrates the importance of spatially explicit modeling in understanding the complex interactions between SLR, adaptation policies, and internal migration. The findings highlight the significant potential for large-scale migration due to SLR in the Mediterranean, which can be substantially mitigated through well-designed adaptation policies. However, the "levee effect" poses a considerable risk and emphasizes the need for strategies beyond hard protection. Future research should focus on refining adaptation policy scenarios, incorporating individual characteristics of vulnerability, and combining top-down models with agent-based models for a more nuanced understanding of migration decisions. A systematic sensitivity analysis of uncertainties is also needed to strengthen the robustness of the model's projections.

The study acknowledges several limitations. The integrated scenarios might not cover the full range of future socioeconomic and climatic conditions. The adaptation policy assumptions assume immediate and consistent policy implementation until 2100, which is not always realistic. The gravity-based migration model simplifies migration decisions and might not perfectly capture all aspects of human behavior, especially complex responses to SLR like temporary displacement and subsequent relocation. The input data, including population projections, urbanization data, and SLR projections, are subject to uncertainties. The model does not capture migration distance, which is a crucial aspect for this research question. Finally, the lack of empirical data on SLR-related migration currently limits the validation of the model’s findings.