 |
NARSTO Workshop on Innovative Methods for Emission Inventory Development and Evaluation University of Texas, Austin October 14-17, 2003 |
 |
NH3 Emission Inventories for Agroecosystems: Role of GIS and Process-based Models in Developing Site Specific Emission Factors, Assessing Variability, and Providing Uncertainty Estimates
William Salas1 and Changsheng Li2
1 Applied Geosolutions, LLC, Durham, NH 03824
2 Institute for the Study of Earth, Oceans and Space, University of New Hampshire, Durham, NH 03824
Biogeochemical processes that control GHG and Ammonia emissions from agroecosystems, like denitrification, nitrification, and ammonia volatilization, are non-linearly coupled with anthropogenic and ecological drivers that are highly variable in space and time. As a result, static emission factors (EFs) cannot capture this variability without development of detailed site and management specific Efs. In addition, EFs typically do not provide estimates of uncertainties particular to a set of conditions. Therefore, assessment of impact of agricultural management alternatives on trace gas emissions and nutrient fluxes needs to be done in a modeling context that incorporates “mass balance” constraints. GIS process-based models can simulate spatially heterogeneous conditions that control temporal and spatial patterns of GHG and ammonia emissions.
This paper will present an overview of a suite of GHG (CO2, CH4 , NO, and N2O) and NH3 emissions inventories using GIS data and a process-based biogeochemical model called Denitrification-Decompostion, or DNDC. Coupled with the inventories are uncertainty analyses that focuses on quantifying how biophysical factors (e.g. soil properties), environmental conditions (e.g. precipitation, temperature) and management alternatives (e.g. amount and timing of fertilizer/manure application, types of manure application, tillage, irrigation) impact the production of GHG, changes in terrestrial C stocks, and NH3 volatilization. Uncertainties generated from the modeled processes as well as from the input data sets are assessed through two levels of sensitivity analyses. First, individual variables are varied across their range of expected values while all other variables are held constant to identify those variables that cause the majority of variance in modeled trace gas emissions, C stocks, and nutrient availability. Secondly, to examine the interaction between variables, Latin Hypercube Sampling (LHS) is utilized. LHS is based on a stratified sampling approach that creates statistically significant results with appreciably fewer model runs. The LHS technique ensures that the entire range of each variable is sampled. Statistical summaries of the model results produce indices of uncertainty that relate the effects of heterogeneity of input variables to model predictions of GHG and NH3 emissions.