A Portfolio of Trees in a Changing Climate: Using Signal Processing and Individual Tree Growth Simulations to Develop Mean-Variance Tradeoff Frontiers for Forest Establishment in the Southern United States

According the the IPCC's 2023 Synthesis Report on Climate Change, global temperatures have risen approximately 1C since the the pre-industrial period, and there is significant uncertainty around future climate projections. Additionally, IPCC and related scientific literature find that the forestry sector is both vulnerable to and already feeling the effects of climate change. This work sets out to accomplish two goals. The first is contribute a new modeling approach that accounts of intra-annual changes in the variability of weather patterns on tree growth using signal processing and statistical modeling techniques. The second uses these models, in conjunction with climate projections, to develop a portfolio view of the forest through the lens of a changing and uncertain climate future. We leverage publicly available data from the USFS's Forest Inventory and Analysis Database, ORNL's DAYMET, and NASA's NEX-GDDP-CMIP6 to train models based on past observation and then simulate future growth based on 88 projections of future climate. Our models consider species-level reactions to site characteristics and weather patterns across the southeastern United States. 

Finally, we compare the performance of roughly 4.6 million forest compositions, across four species and two management scenarios, to explore the trade-off between expected return and the variance of said return in a Markowitz Portfolio Selection framework when optimiizing financial returns to timber and carbon production, respectively. Special attention is paid to the performance of different species and their relative prevalence in portfolios along the efficient frontier.