2018 Hellman Fellow
Assistant Professor, ESPM
Project Title: Big Data, Big Uncertainty: Ecological Decision Theory for the 21st Century
Summary: We live in an era of pressing ecological questions, unprecedented access to ecological data, and unprecedented uncertainty. How can we make informed ecological management decisions using new data sources without being misled by uncertainty? Ecological research is experiencing a data revolution thanks to satellites and micro-sensors, national observatories and citizen science, but we lack approaches to make good use of this data. With big data comes big uncertainty, including new sources of measurement errors, hidden assumptions and gaps in data series. Today’s machine learning approaches may prove ill-suited to the long timescales, heterogeneous data and shifting baselines typical of ecological problems. A dramatic illustration of this problem comes in the form of ecological regime shifts: sudden transitions from kelp forest to urchin barrens, the sudden loss of a top predator or crucial habitat such as coral reefs or mangroves. Predictions and forecasts based on data during one regime will be of little use after the shift. To harness this data revolution, we will need new approaches that can be informed by our scientific understanding & reflect inherent uncertainty to make use of such data without being misled by it.