Title: Predicting the future of the Amazon rainforests using regression analysis
Presenting Author: Kamalika Das
Organization: UCSC/NASA Ames Research Center
Co-Author(s):
Anuradha Kodali Marcin Szubert Josh Bongard Sangram Ganguly

Abstract:
Both short-term (weather) and long-term (climate) variations in the atmosphere directly impact various ecosystems on the earth. Forest ecosystems, especially tropical forests, are crucial as they are the largest reserves of terrestrial carbon sink. For example, the Amazon rainforests are a critical component of global carbon cycle storing about 100 billion tons of carbon in its woody biomass. There is a growing concern that these forests could succumb to precipitation reduction in a progressively warming climate, leading to release of significant amount of carbon in the atmosphere. Therefore, there is a need to accurately quantify the dependence of vegetation growth on different climate variables in order to obtain better estimates of drought-induced changes to atmospheric CO2. The availability of globally consistent climate and earth observation datasets have allowed global scale monitoring of various climate and vegetation variables such as precipitation, radiation, surface greenness, etc. Using these diverse datasets, we aim to quantify the magnitude and extent of ecosystem exposure, sensitivity and resilience to droughts in forests. The Amazon rainforests have undergone severe droughts twice in last decade (2005 and 2010), which makes them an ideal candidate for the regional scale analysis. Current studies on vegetation and climate relationships have mostly explored linear dependence due to computational and domain knowledge constraints. Also, a significant number of studies on the Amazon resort to time series trend analysis in order to make significant claims about the resilience (or the lack of it) of the Amazon rainforests. Our goal is to build a statistically robust and comprehensive machine learning model from all the observed historical data that can be used to predict the fate of the Amazon rainforests, given climate variables such as rainfall, land surface temperature, soil moisture, and radiation. We will use both regularized linear regression and nonlinear regression models such as Support Vector Machines that will explore nonlinearities in the dependency relationships. We also explore a modeling technique called symbolic regression based on evolutionary computation that allows discovery of the dependency structure without any prior assumptions. This method allows us to find unknown relationships that are not captured by existing modeling techniques due to lack of domain knowledge. Some of the climate variables used in our study are temperature, and rainfall. We also incorporate elevation and slope information in our modeling. Initial results on portions of Amazon reveal interesting dynamics between the different regressors and vegetation during the normal (no drought) period.