Quantifying the Representation Error of Land Biosphere Models using High Resolution Footprint Analyses and UAS Observations

Abstract:

The validity of land biosphere model outputs rely on accurate representations of ecosystem processes within the model. Typically, a vegetation or land cover type for a given area (several Km squared or larger resolution), is assumed to have uniform properties. The limited spacial and temporal resolution of models prevents resolving finer scale heterogeneous flux patterns that arise from variations in vegetation. This representation error must be quantified carefully if models are informed through data assimilation in order to assign appropriate weighting of model outputs and measurement data. The representation error is usually only estimated or ignored entirely due to the difficulty in determining reasonable values. UAS based gas sensors allow measurements of atmospheric CO₂ concentrations with unprecedented spacial resolution, providing a means of determining the representation error for CO2 fluxes empirically. In this study we use three dimensional CO₂ concentration data in combination with high resolution footprint analyses in order to quantify the representation error for modelled CO₂ fluxes for typical resolutions of regional land biosphere models. CO₂ concentration data were collected using an Atlatl X6A hexa-copter, carrying a highly calibrated closed path infra-red gas analyzer based sampling system with an uncertainty of ≤ ±0.2 ppm CO₂. Gas concentration data was mapped in three dimensions using the UAS on-board position data and compared to footprints generated using WRF 3.61.

Chad Hanson, Oregon State University, Corvallis, OR

Andres Schmidt, Oregon State University, Corvallis, OR

Bev Law, Oregon State University, Corvallis, OR