B33A-0628
Optimizing Collocation of Instrument Measurements and Field Sampling Activities

Wednesday, 16 December 2015
Poster Hall (Moscone South)
Gabriel T Bromley, University of Wisconsin Madison, Atmospheric and Oceanic Science, Madison, WI, United States
Abstract:
The National Ecological Observatory Network (NEON) will provide data from automated instrument measurements and manual sampling activities. To reliably infer ecosystem driver-response relationships, two contradicting requirements need to be considered: Both types of observations should be representative of the same target area while minimally impacting each other. For this purpose, a simple model was created that determines an optimal area for collocating plot-based manual field sampling activities with respect to the automated measurements.

The maximum and minimum distances of the collocation areas were determined from the instrument source area distribution function in combination with sampling densities and a threshold, respectively. Specifically, the maximum distance was taken as the extent from within which 90% of the value observed by an instrument is sourced. Sampling densities were then generated through virtually distributing activity-specific impact estimates across the instrument source area. The minimum distance was determined as the position closest to the instrument location where the sampling density falls below a threshold that ensures <10% impact on the source area informing the instrument measurements. At most sites, a 30m minimum distance ensured minimal impact of manual field sampling on instrument measurements, however, sensitive sites (e.g., tundra) required a larger minimum distance.

To determine how the model responds to uncertainties in its inputs, a numerical sensitivity analysis was conducted based on multivariate error distributions that retain the covariance structure. In 90% of all cases, the model was shown to be robust against 10% (1 σ) deviations in its inputs, continuing to yield a minimum distance of 30 m. For the remaining 10% of all cases, preliminary results suggest a prominent dependence of the minimum distance on climate decomposition index, which we use here as a proxy for the sensitivity of an environment to disturbance.