Impacts of climate change adaptation strategies need to be evaluated using principled methods spanning sectors and longer time frames. We propose machine-learning approaches to study the long-term impacts of flood protection in Bangladesh. Available data include socio-economic survey and events data (death, migration, etc.) from 1983–2014. These multidecadal data, rare in their extent and quality, provide a basis for using machine-learning approaches even though the data were not collected or designed to assess the impact of the flood control investments. We test whether the embankment has affected the welfare of people over time, benefiting those living inside more than those living outside. Machine-learning approaches enable learning patterns in data to help discriminate between two groups: here households living inside vs. outside. They also help identify the most informative indicators of discrimination and provide robust metrics to evaluate the quality of the model. Overall, we find no significant difference between inside/outside populations based on welfare, migration, or mortality indicators. However, we note a significant difference in inward/outward movement with respect to the embankment. While certain data gaps and spatial heterogeneity in sampled populations suggest caution in any conclusive interpretation of the flood protection infrastructure, we do not see higher benefits accruing to those living with higher levels of protection. This has implications for Bangladesh’s planning for future and more extreme climate futures, including the national Delta Plan, and global investments in climate resilient infrastructure to create positive social impacts.

In the context of developing countries, effective groundwater resource management is often hindered by a lack of data integration between resource availability, water demand, and the welfare of water users. As a consequence, drinking water-related policies and investments, while broadly beneficial, are unlikely to be able to target the most in need. To find the households in need, we need to estimate their welfare status first. However, the current practices for estimating welfare need a detailed questionnaire in the form of a survey which is time-consuming and resource-intensive. In this work, we propose an alternate solution to this problem by performing a small set of cost-effective household surveys, which can be collected over a short amount of time. We try to compensate for the loss of information by using other modalities of data. By combining different modalities of data, this work aims to characterize the welfare status of people with respect to their local drinking water resource. This work employs deep learning-based methods to model welfare using multi-modal data from household surveys, community handpump abstraction, and groundwater levels. We employ a multi-input multi-output deep learning framework, where different types of deep learning models are used for different modalities of data. Experimental results in this work have demonstrated that the multi-modal data in the form of a small set of survey questions, handpump abstraction data, and groundwater level can be used to estimate the welfare status of households. In addition, the results show that different modalities of data have complementary information, which, when combined, improves the overall performance of our ability to predict welfare.