Determining biophysical, carbon and climate feedbacks of tropical deforestation: Do the pathways matter in both science and policy?
Abstract
Research on climate and land surface coupling has made considerable progress in recent years, and this study contributes to our understanding of key feedbacks and to discussion in the science-policy nexus by exploring the question, "Do the deforestation pathways matter?" Deforestation perturbs both biophysical and carbon feedbacks on climate. However, biophysical feedbacks operate at temporally-immediate and spatially-focused scales, and thus may be sensitive to the rate of deforestation rather than just to the total forest cover loss. Transient shocks or non-linearities may exist but are not detected in equilibrium deforestation simulations. Described here is a method for implementing annual tree-to-grass conversion in the tropics using a fully-coupled global climate model (GCM) with interactive vegetation for the purpose of testing biophysical, carbon and climate response to "deforestation pathways". Deforestation pathways are characterized here as the rates of tree loss paired with a preservation target (10% per tropical tree type). Code modifications are tested, and methods of analysis are initially explored using Community Climate System Model 3.0 (CCSM3) with the Dynamic Global Vegetation Model (DGVM) simulating a 10% vs. 1% rate. A newer version, CCSM4 with prognostic carbon and nitrogen cycles and dynamic vegetation (CNDV), which corrects crucial biases in the previous model, is used to simulate a range of deforestation rates—5%, 2%, 1% and 0.5%—again paired with a 10% preservation target. Sensitivity analyses are conducted on two levels—firstly, trends in biophysical, carbon and climate variables during the period of active deforestation are quantified; secondly, these trends are compared across deforestation pathways to determine the effect of the rate. These analyses are applied over tropical land, the Amazon Basin, Central Africa, and Southeast Asia. On a local to regional scale, the biophysical impacts of tree cover loss on climate were found to be comparable (as much as 55% in magnitude) to the combined biophysics and carbon impacts. The results further show that the biophysical, total ecosystem carbon and climate sensitivities to transient deforestation can be approximated as linear, and vary more across regions than across rates. Post-deforestation climate means are also similar across the different pathways. Only when the trends and means are averaged over Southeast Asia and over global land do the rates consistently produce statistically different results for the climate variables (as determined by an analysis of variance). However, there are no clear trends or patterns that are consistent across rates and regions, and in the post-deforestation period, the means are still within each other's standard deviation. These suggest that the earth system response in CCSM4-CNDV to this particular range of rates may be less dependent on the rate than originally hypothesized. It is unclear whether a slow but prolonged deforestation is any better or worse than one that is rapid but short given the same preservation target. However, the deforestation rate is still an important consideration in that it hastens or delays significant warming on the order of decades, which matters from a mitigation, adaptation and development planning perspective. Implications on climate policy, particularly on reducing emissions from deforestation and forest degradation in developing countries (REDD), are discussed. The linearity of the earth system response allows for flexibility in project design and supports proposals to streamline monitoring methodologies by assigning fixed amounts of carbon per unit forest area. The results also imply that the use of projected rather than historical baselines would be more appropriate. However, the significance of biophysical impacts on local temperature and precipitation also suggests developing countries (especially in the Amazon Basin) have climate benefits to gain beyond the income from REDD+ carbon credits, and that carbon-based metrics are inadequate for the forestry sector, as well as for land cover/land use change in general. The modeling methodology shows it is possible to develop a direct "climate sensitivity" metric normalized per unit forest area, although it is unclear how this can be operationalized alongside the current carbon metric of fossil fuel-based sectors. Thus, composite indices of carbon credits which are enhanced or discounted by net biophysical impacts remain a viable alternative to be explored.
Degree
Ph.D.
Advisors
Gurney, Purdue University.
Subject Area
Ecology|Climate Change|Biogeochemistry|Biophysics
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.