IRI@AGU: Can We Predict “Climate Migrations”?

IRI scientists Ángel Muñoz and Diego Pons are interested in better understanding the root causes of migration in Central America. With their Columbia colleagues, Alex de Sherbinin and Susana Adamo–from the Center for International Earth Science Information Network (CEISIN)–and Diana Giraldo from the University of Reading, they’ve developed a prototype model that considers climate and socioeconomic factors to see if displacements of people can be predicted and better explained. Muñoz elaborates in the Q&A below and also discusses a new forecasting system he’s helping implement as part of the Columbia World Project, ACToday.

You’ve been developing a model that tries to predict climate migrations, using the 2018 Guatemalan migration as a test case. What are the factors that go into such a model? What are its limitations?

Migrations are caused by multiple and entangled factors, making them in general virtually impossible to predict. Our analysis found that a combination of an increasing infant mortality rate since 2012, a high food-price inflation rate (the fourth highest since 1996), and an increase in the unemployment rate set the stage for the migration that occurred in 2018. In addition, the region endured a multi-year drought during the previous three years. This acted as the final stressor because it drastically increased household debts via reduced staple crop harvests, and it limited access to unskilled employment in the agricultural sector.

Over the last year, there have been numerous stories (e.g., here, here and here) that imply climate change was the chief culprit behind the mass migration out of Guatemala and other Central American countries to the U.S.? Is this accurate?

It is not accurate. Our research shows that socio-economic factors are the most important drivers of the 2018 Guatemalan migration to the U.S., although the recent multi-year drought in the region played a role as a trigger of the population displacement. Nonetheless, this multi-year drought is mostly related to natural climate variability. Climate change typically explains about 1% of the total annual rainfall variation in Guatemala, Honduras and El Salvador.

You’ll also be showcasing the latest developments in the NextGen forecasting system that ACToday has helped implement so far in Guatemala and Colombia. Briefly, what is NextGen, and why is the forecasting community becoming so excited about it?

NextGen is a systematic general approach for designing, implementing, producing and verifying objective regional-scale climate forecasts. It helps staff at national meteorological institutions select the best set of dynamical models for their regions of interest. It allows them to create forecasts based on these models for seasonal and sub-seasonal time scales and at a regional, national or sub-national level. NextGen also automates the process of generating and verifying these forecasts.

The forecast and climate services communities in Colombia and Guatemala are excited about NextGen because it helps provide more skillful predictions at multiple timescales for total rainfall, consecutive dry days and other variables and thresholds of interest to their users. It’s an approach known as flexible format forecasting, pioneered at IRI.

Thousands of Guatemalan farmers will now have access to forecasts and other climate information to help them increase crop yields and earn more, thanks to five new regional collaborative networks launched by the ACToday Columbia World Project and its international partners.