Photo of a presenter at Tracking Climate Change Adaptation: Methodological Challenges in the Age of Big Data.

Despite the urgency of climate change, keeping track of global adaptation efforts has proven tricky. Adaptation means different things in different places: in Rotterdam, the creation of parks that can absorb larger, more frequent floods; in Los Angeles, retrofitting homes to withstand scorching summer heat.

In many places, adaptation goes by another name. Efforts to diversify agricultural communities’ income or reroute roads away from vulnerable coastlines might be classified as development or infrastructure activities, though they also brace communities against more volatile weather. Adaptation activities span the jurisdiction of different agencies and the risks of different regions. It can be tough to identify, let alone track, how a country, state, or city government is reacting to a changing climate.

On May 11 and 12, Yale-NUS College, in collaboration with the Adaptation Tracking Collaborative (ATC), hosted Tracking Climate Change Adaptation: Methodological Challenges in the Age of Big Data. The workshop explored ways to develop the data needed to measure climate adaptation policies.

The ATC, a network of adaptation researchers, works to improve the understanding of adaptation around the world.  They’ve piloted a monitoring framework to capture the policy goals and instruments that governments have adopted to adapt to climate change. But to apply it, they first need to track down adaptation data.  

Over the course of two days, workshop participants brainstormed ways alternative forms of data collection, such machine learning, web scraping, analyzing text as big data, and social media analysis, could gather information about adaptation policies. They took cues and inspiration from presentations outlining the wide range of these tools’ applications. Text analysis, for instance, has traced the migration of political ideas through different bills and laws. Machine learning techniques can uncover the sociology behind political affiliations. The discussion was also grounded by lessons from the International Institute for Environment and Development (IIED), who shared lessons from their experience piloting the ATC framework in Bangladesh.  

Participants drew on these examples to produce a list of tools that might help jumpstart the ATC framework. A mix of web scraping and object recognition, for instance, could potentially trace how often, and in what context, adaptation enters high-level political discussions. Network analysis could better capture levels and types of collaboration within and across governments implementing adaptation policies. Text analysis could dramatically shorten and streamline time spent combing through dense policy documents.

The workshop also generated a list of questions to troubleshoot. Some questions resist a technical solution. For instance, the best way to identify how adaptation is “mainstreamed” or integrated into other activities, like development, water management, or agriculture, remains an open question. Taking stock of the full scope of a country’s adaptation activities often means piecing together data spread across many different documents. And while new tools can help parse and search through documents, they are less adept at identifying which ones matter. Applying the ATC framework in Bangladesh relied on IIED’s ability to piece together the different moving parts within the country’s adaptation strategy. Researchers, in other words, knew where to look, while automated tools might struggle.  

As the ATC works to capture the state of adaptation efforts, it may also provide insight into the potential and limitations of the use big data to address these questions.

For workshop participants, presentations videos and materials are available below. (E-mail to request access.)



Image: Trevor Martin discusses text parsing during his presentation at Tracking Climate Change Adaptation: Methodological Challenges in the Age of Big Data.