A Survey of Current Practices in Uncertainty Analysis in Ecosystem Ecology.

Ecosystem budgets have not traditionally reported uncertainty in hydrologic inputs and outputs or pools of vegetation and soils, although quantifying uncertainty is important to establishing the significance of comparisons, to making predictions with known confidence, and to identifying priorities for investment.  One problem is that a small watershed ecosystem is defined by one stream, and thus replication cannot be used to assess uncertainty in stream loads.  Temporal interpolation is a source of uncertainty for stream chemistry, while spatial interpolation is needed to estimate precipitation at the ecosystem scale.  Sampling error is commonly reported for soils and vegetation, but it is challenging to propagate the uncertainty in the measurements and models involved in the calculations.  We conducted a survey to learn about current practices in reporting uncertainties in precipitation, streamflow, soils, and vegetation.

We obtained responses from 77 researchers representing 49 research sites around the world.  Sampling error was the most important source of uncertainty in calculations of biomass and soil pools, according to respondents in these fields, and this was the source they said they most often reported.  Uncertainty in the chemical analysis of precipitation and stream water was the source most commonly reported by hydrologists, although this was one of the least important sources of uncertainty to calculations of hydrologic flux. Measurement error in watershed area, model error in rain gage interpolation, and the sampling error in estimates of soil coarse fraction ranked among the least reported relative to their perceived importance. Some sources of uncertainty are easy to quantify and should be routinely reported (e.g., rain gage interpolation), whereas others lack standard methods and may require establishment of best methods (e.g., watershed area, biomass model selection).