Model Based Clustering to Capture Climate Variability

Research about social behavioral adaptation to climate change struggles with assessing covariation because the empirical requirements of matching social and environmental evidence necessitates both temporal depth and spatial variability. Most assessments are limited in terms of environment measures, social measures, and temporal depth that might capture the long term shifts in climate-related weather patterns. Our study employs longitudinal data from Thailand about migration over a 16-year period and matches it with remote sensing satellite information about vegetative health, to capture exposure to drought stress. We measure the environmental variability across the landscape and across time using Fraley and Raftery's (2002) clustering approach. Our initial results are very promising and we are seeking insights about how best to improve our measures of climate variability, especially after having just completed some ground truthing fieldwork in March 013.