4

Mechanisms and Pathways for Modeling the Impacts of Catastrophic Chronic and Slow-Onset Events on Human Migration

In addition to rapid-onset catastrophic events, Session 2 focused on understanding the drivers and impacts of chronic, gradual-onset, and sustained events such as droughts. Speakers examined the conditions under which these long-term shifts influence migration and examined some of the modeling and data challenges and the knowledge gaps in studying the complex interactions involved.

MODELING SLOW-ONSET CLIMATE EVENTS

Kevin Anchukaitis (University of Arizona) discussed the challenges of modeling a large system of interactions that link slow-onset climate events with food systems, socioeconomic factors, and the decision to migrate. His research is also examining how perceptions of these trends or changes may or may not lead to action.

Anchukaitis presented his work based in Central America in an area known as the dry corridor that is characterized by low rainfall and high levels of food insecurity and is one of the spots predicted for future drying. Various models and observational data suggest that in areas of expected climate signals, the change will emerge after four decades. Therefore, how chronic long-term changes and natural climate variability are managed, specifically average precipitation in this case and its subsequent effect on agriculture and food systems, will be important to consider, he said (Anderson et al. 2019).

One of the challenges in modeling is the insufficiency of data. Although there are weather stations in the region, precipitation datasets are not

equally distributed or fully represented in time, Anchukaitis said. The data available for researchers may therefore come from only a few stations. Variability in data quality or what datasets are used in which models may result in different trends in precipitation. He noted that there is a lack of observations not just in Central America but also large parts of the tropics in the Global South and other regions where chronic climate conditions are expected to be observed. Climate models are useful tools, but it is also important to consider the potential limitations and uncertainties of the models as part of a larger system of downstream factors such as food security, said Anchukaitis.

Anchukaitis also discussed how people perceive changes in climate and how that perception can influence decisions about livelihood or migration beyond measurements and model simulations. For example, surveys of farmers in two Guatemalan communities on how they perceived change in rainfall over the last decade revealed that there was a wide range of responses, indicating that they perceived no change, less rain, or more rain. Perceptions also differed when comparing the two communities and led to different adaptive actions such as their agricultural practices. The clearest predictor of this difference was off-farm income; in a community heavily dependent on farming income, farmers were more apt to adjust their practices in response to perceived changes, whereas people in a community with more options for off-farm income perceived the same changes but were less likely to adjust their farming practices.¹

In closing, Anchukaitis noted that climate models consistently suggest areas such as Central America and the Mediterranean will see significant drying in the coming century, and this is likely to impact vulnerable populations in these areas. Even as natural climate variability continues to be a dominant influence on agriculture, livelihoods, and food security, it is important to look ahead at these longer trends, how people in these regions perceive the climate changes they experience, and how they respond, he said.

CHALLENGES IN MODELING CLIMATE-INDUCED MIGRATION

Cristina Cattaneo (Foundation Euro-Mediterranean Center on Climate Change and RFF-CMCC European Institute on Economics and the Environment) examined some of the challenges in modeling relationships between climate change, weather shocks, and migration. She discussed how

___________________

¹ Presentation at the Workshop on Climate Change and Human Migration: An Earth Systems Science Perspective, March 18–19, 2024, see https://www.nationalacademies.org/event/41814_03-2024_workshop-on-climate-change-and-human-migration-an-earth-systems-science-perspective#sectionEventMaterials (accessed June 24, 2024).

key gaps in global migration data, limited knowledge of the interplay between alternative adaptation mechanisms and migration, and incomplete information about moderating factors or heterogeneous responses could hinder efforts to project future climate-induced migration flows.

One factor constraining the modeling of the relationships between climate change and international migration is the availability of high-quality migration data, she noted. Censuses and registries of destination countries are often used to track the nationality or place of birth of foreigners, and for many years these were the only data on international migration. Cattaneo said that while this approach allows large temporal coverage and provides some global perspective on origin and destination, it also has limitations. These data lack social demographic details, typically limited to only gender and education level and do not provide exact location within origin country. Detailed geographic and demographic data are important, as climatic conditions may widely vary within the same country.

Another source of data from origin countries is from surveys originally intended to assess conditions such as poverty and health. These surveys may provide geo-localized information on the place of origin and additional social demographic details. Cattaneo added that it can also be difficult to extract generalized messages and outcomes from data that come from a single country.

Newer sources of data, known as big data or digitally traced data, are available from social media with user-provided location data, Cattaneo noted. Information from mobile call records that capture cell tower location, or from remote sensing technology aboard satellites or drones can capture fine geographic detail, but Cattaneo noted that they tend to lack social demographic details and represent a limited time frame.

She emphasized that in addition to limitations in the data used to measure migration, there are also knowledge gaps that hinder efforts to model the links between climatic shocks and migration. Migration is one possible strategy for adapting to climate change, and it is often the last resort. However, it is not known what other options households and individuals consider in response to a climate shock and what makes migration more or less likely compared to other strategies. Cattaneo said that researchers lack a full understanding of the interplay between different adaptation strategies and climate shocks. Although there is a tendency to view climatic shocks as driving people to move, there are many contexts and circumstances in which people do not move, for example, because of lack of resources. For example, even if people have a high incentive to move, they may be less able to do so in situations where climate shocks reduce available resources. Consequently, it can be challenging to model migration responses to climate changes due to difficulty in determining whether immobility is due to lack of resources or to alternative strategies having been implemented.

Projections of future climate-induced flows are crucial modeling tasks, said Cattaneo, but there is a large knowledge gap in the available historical parameters that are needed for projections. Currently, most of the existing studies provide short-term dynamics, making it hard to predict migration patterns further in the future. Cattaneo noted two specific efforts to project future migration: the World Bank’s Groundswell report (Rigaud et al. 2018), focused on internal migration, and research from Missirian and Schlenker (2017), exploring historical relationships and future flows of migration in specific destination countries.

DEFINING SLOW-ONSET CLIMATE IMPACT

Cascade Tuholske (Montana State University) discussed how nuances in defining slow-onset, or chronic, climate impacts can elucidate insights about climate migration mobility by connecting these impacts to socioecological systems. He offered that not having preconceived notions as to what might drive migration, a human action, in a complex socioecological system is important, pointing to a recent paper highlighting the inherent risk in climate or environmental determinism (Horton et al. 2021). The authors of the study suggested that ignoring human agency may oversimplify what is occurring on the ground in local communities. To understand the various dynamics at play for informed policy decision making, top-down modeling that can identify hotspots for decreased habitability and slow-onset events could be combined with local and social contextual information (Horton et al. 2021). Tuholske said that with the increasing availability of data, it will be important to work with transdisciplinary and cross-cultural teams to consider the context in any given location as to what may or may not be driving people to move.

“There is a serious risk of ecological fallacy whereby we observe movement of people often through secondary data, and then we associate it spatially with some sort of climate hazard or climate change event,” Tuholske said. “Those two things may be correlated in space and time, but climate change is not, per se, the driver of movement of people. I really fear that we will over-attribute movement of people to climate change rather than to the social, political, or cultural contexts that might actually be creating underlying conditions where people don’t feel safe or comfortable or are seeking new opportunity elsewhere.”

Data limitations are another key issue, Tuholske said, pointing to a paper that examined the use of Earth observation data in areas that lack weather stations (Zaitchik and Tuholske 2021). Researchers estimated that 4 billion people live more than 25 kilometers away from a weather station with a reliable reporting record. In India, where there are approximately 3,000 urban settlements, there are only 111 weather stations. This makes

it challenging to measure slow-onset climate changes in a given location and also influences modeling precision and accuracy. For example, data from the Coupled Model Intercomparison Project outputs offer resolutions only up to 100 by 100 kilometers, which does not capture the spatial variability in local climate, and most likely does not match up with the scale of location data for migration, Tuholske noted. However, researchers recently were able to achieve 5-by-5-kilometer outputs that can be applied to high-resolution climatology models (Williams et al. 2024). This improved resolution may then be used in a range of climate-related projections, such as for hot and dry extremes relevant to agricultural productivity and projecting the impacts of extreme heat on human health and labor productivity by approximated shaded wet globe temperature.

Moving forward, Tuholske emphasized the importance of trend-preserving higher-resolution climate projections and other analytical products that utilize multiple datasets. He added that it is important to elucidate the actual linkages among slow-onset events, human impacts, and decisions to migrate that are substantiated by actual experiences of people. “Conveying the complexity of the human decision to migrate is really paramount so we can really understand how climate change may be creating slow-onset events that drive people to leave their homes,” he said.

DISCUSSION ON MODELING SLOW-ONSET EVENTS

Shanna McClain (National Aeronautics and Space Administration) moderated a panel discussion with the speakers that examined additional facets and limitations of data in modeling and understanding the impact of slow-onset climate events; perspectives on adaptation strategies and other factors that may influence the decision to migrate or stay; and the challenges of separating the influence of climate versus other factors in understanding why people move.

On resiliency, Tuholske noted that discussions on sustaining rural livelihoods and climate adaptation often consider agricultural technologies, particularly large-scale mechanized farming, as a solution to building resilience in farming communities and discourage migration. However, experiences from Europe and the United States suggest that increased reliance on machines can lead to decreased demand for agricultural labor over time and raise questions about whether technology truly enhances resiliency in rural communities facing climate change, if the aim is to provide alternative strategies to migration. While there is no definitive answer, he said that it is important to examine how new labor economic theories and historical development perspectives inform decisions regarding capital allocation to these communities, considering both their desires and the potential outcomes of such initiatives. Cattaneo added that migration is a last-resort

choice; if climatic shocks can be addressed through adaptation, then people can choose to stay.

Alexander de Sherbinin (Columbia University) highlighted additional sources of data, including sentinel sites in Bangladesh, Mexico, and elsewhere, where researchers collect data across multiple years to try to understand migration and other demographic patterns. In addition, he pointed to long-term demographic surveys that collect information on household composition changes over time; have been conducted in some countries in Africa and other regions and can provide some information on youth mobility. Cattaneo added that relying only on country-level data requires estimating the specific challenges on an individual level. Ideally, she said, more data on both climate and information about where people are located would provide important information.

In response to a question about the impact of the general trend toward urbanization, Anchukaitis commented that urbanization trends also add to the complexity of modeling how climate changes might ripple through various communities. What are the consequences of increased urbanization and the climate change impacts in urban environments?

Cattaneo pointed out that people may sometimes be labeled as climate migrants even though other social and economic factors may be present. Climate has a direct role when, for example, a flood or sea-level rise occurs, and people are forced to move because their home is destroyed. However, it is more complicated for indirect relationships, which are more likely during slow-onset events. For example, climate impacts may affect economic factors, which in turn leads to relocation. Tuholske commented that the impact of an event within the current globalized economic system is location dependent. He emphasized that it is important to think about the interplay of various processes, what it means to be a climate migrant, and the consequences of formalizing that label.