COS 154-4 - Land use change, carbon dynamics, and emission mitigation potential of California natural and working lands

Thursday, August 10, 2017: 2:30 PM
B110-111, Oregon Convention Center
David C. Marvin1, Benjamin M. Sleeter2, Dick Cameron1, Tamara Wilson2 and Jinxun Liu2, (1)The Nature Conservancy - California, (2)Western Geographic Science Center, USGS, Menlo Park, CA
Background/Question/Methods

California’s natural and working lands have been a net source of carbon dioxide emissions in recent years. Managing ecosystems differently could reverse this source of carbon into significant net annual sink. Existing land use and carbon change assessments for CA have relied upon national land cover data products with the potential for high rates of local errors in land use classification, or which did not differentiate important variation in the underlying carbon stock of specific land cover classes (e.g., orchard vs annual crop within agricultural areas). Improved data on land use classification, land use change, and carbon stocks are needed to produce a more accurate understanding of the historical carbon dynamics across California’s ecosystems, allowing for enhanced predictions of future source-sink dynamics. We used a combination of satellite optical and radar imagery to train machine learning classification algorithms for improved land use class predictions across the state of California. The land use classifications were combined with carbon flux parameters from a dynamic global vegetation model and run through the LUCAS model - a stochastic space-time simulation model built to represent land change feedbacks on carbon cycling in terrestrial ecosystems.

Results/Conclusions

We find California was a small net carbon source over the period 2001-2015, with high interannual variability largely driven by climate. Trends in agriculture and developed land use are swamped by large pulses of carbon due to wildfire, especially during the periods from 2006-2008 and 2013-2015. Future updates to the model include producing projections of carbon dynamics due to land use and climate change scenarios. The ability of the model to produce spatially explicit outputs allows for visualization of various land management policy interventions and their resulting effect on ecosystem carbon dynamics. Once assessed, the performance of each intervention and activity will inform recommendations to local and state agencies on how to best direct funding, resources, and policy decisions toward natural and working lands for carbon emissions mitigation.