Achievement

Students from our first workshop (2009/2010) continued to work on their model of pond depths for the Archbold Biological Research Station using a long-term series of depth measurements from sample ponds. The students used these data and ancillary environmental data to develop a time-series model of depths for all ponds at the field station. They developed a novel statistical technique to overcome challenges in data availability and irregularity of pond depth measurements by developing a recursive model that incorporated the previous estimated pond depth across time. They also applied a creative data transformation to use standard linear techniques to efficiently run thousands of potential combinations of covariates. The resulting model effectively predicted pond depths during the data collection period. It also was able to hindcast and forecast depths using available ancillary data and/or simulated data. Students recently presented results to the clients.