Decision Support Process for CO2 Monitoring Network Design, as applied to the North American Permafrost Region

The Northern regions of Canada and the United States hold vast reserves of soil organic matter (carbon), once protected from decomposition because of cool wet soils, and/or permafrost. Unfortunately, these Northern areas are changing rapidly because the rate of climate warming in the North has significantly outstripped the global average. As a result of permafrost melting and soil processes stimulated by higher temperatures, soil CO2 emissions in these high latitude areas are expected to accelerate. On an annual basis, it is conceivable that these natural Northern CO2 emissions could match or begin to exceed human industrial emissions. From a risk management perspective, the Northern region represents an important climate risk of international interest that should be monitored in a systematic long-term effort. Currently, there is almost no high latitude monitoring done due to limitation of instrumental techniques, remoteness, and other factors. While technology improvements will make this monitoring physically possible, it is critical that the conceptual framework for monitoring also be planned carefully. The monitoring network will need to make efficient use of funding to target high-risk areas, balance many possible emission predictions that vary spatially and temporally, allow for incremental network growth and take into account transportation costs and accessibility. To address these needs, we use a Simulated Annealing based algorithm that determines optimal sampling densities and distributions according to various risk factors. This network optimization makes use of several different potential CO2 emission estimates generated using the Canadian Regional Climate model forecasts and other spatial datasets. These estimates represent a wide range of potential evolutions of soil emissions based on factors such as expected climate change, the distribution of carbon storage, and carbon sensitivity to temperature and moisture changes. The calculated optimal network demonstrates a need for broad spatial coverage, but also suggests that several sub-networks, centered in locations such as Anchorage, Regina and Iqaluit, could potentially serve as the basis for an adequate network. It also highlights the urgency associated with permafrost decay, with substantial respiration changes (more than 10%) occurring in some areas within the next decade. Due to the reality of conflicting predictive datasets, a detailed sensitivity analysis of this optimization is also presented. We hope that this research will result in discussions that will lead to Long-Term monitoring initiatives, and future work is focused on applying this approach to other regions.