Print PageQuantification of uncertainty in ecosystem input-output budgets is needed for determining the significance of observed differences, analyzing trends over time, and guiding research investments by identifying which components contribute the most to the overall uncertainty. However, quantification of natural variability and model uncertainty is missing from most estimates of ecosystem budgets. The mission of QUEST (Quantifying Uncertainty in Ecosystem Studies) is to improve the quality and frequency of uncertainty analyses in ecosystem studies. Our goals are to raise consciousness about the value of uncertainty analysis, provide guidance to researchers interested in uncertainty analysis, and provide support to both developers and users of uncertainty analyses.
We are conducting a cross-site comparison of small watersheds to assess whether uncertainties in hydrologic element fluxes in precipitation and streamflow volume and chemistry arise from fundamental differences in ecosystem function across sites or in sampling or computational methods. Sources of uncertainty include the precision and accuracy of measurements, natural variation in space and time, and errors in conceptualizing or modeling the system. We are propagating the uncertainty through streamwater input-output budgets and comparing a variety of models to estimate the model uncertainty for a 10-year period at several long-term study sites throughout the US.
Results/Conclusions
In a preliminary analysis of phosphorus inputs and outputs at Hubbard Brook, we used a Monte Carlo simulation with 10000 iterations and estimated an average precipitation input of 0.063 kg P ha-1 yr-1 with a 95% CI of 0.041-0.107 kg P ha-1 yr-1. We estimated an average streamflow output of 0.0067 kg P ha-1 yr-1 with a 95% CI of 0.0035-0.0102 kg P ha-1 yr-1. The difference between precipitation inputs and streamflow outputs was 0.056 kg P ha-1 yr-1 with a 95% CI of 0.034-0.098 kg P ha-1 yr-1. We are using a similar approach at six other watershed study sites (Andrews LTER, Biscuit Brook Watershed, Coweeta LTER, Fernow Experimental Forest, Luquillo LTER, Marcell Experimental Forest, and Niwot Ridge LTER) to estimate uncertainty over a ten-year period (1995-2005).
In addition to the improvement to ecosystem budgets, the results of this work have important implications for monitoring programs and the assessment of trends with environmental change. In future work, we also plan to expand this analysis to include other ecosystem measurements such as vegetation, soils, and atmospheric deposition.
