Page Banner

United States Department of Agriculture

Agricultural Research Service

Research Project: USING REMOTE SENSING & MODELING FOR EVALUATING HYDROLOGIC FLUXES, STATES, & CONSTITUENT TRANSPORT PROCESSES WITHIN AGRICULTURAL LANDSCAPES Title: Intercomparison of Nine Micrometeorological Stations During the BEAREX08 Field Campaign

Authors
item Alfieri, Joseph
item Kustas, William
item Prueger, John
item Hipps, Lawrence -
item Chavez, Jose -
item French, Andrew
item Evett, Steven

Submitted to: Journal of Atmospheric and Ocean Technology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: July 19, 2011
Publication Date: November 1, 2011
Citation: Alfieri, J.G., Kustas, W.P., Prueger, J.H., Hipps, L.E., Chavez, J.L., French, A.N., Evett, S.R. 2011. Intercomparison of nine micrometeorological stations during the BEAREX08 Field Campaign. Journal of Atmospheric and Ocean Technology. 28:390-406.

Interpretive Summary: Land surface conditions, such as the type and amount of vegetation, play a critical role in controlling the exchange of heat and moisture between the land surface and the atmosphere. In turn, the exchange of heat and moisture regulates a broad range of meteorological, hydrological and environmental processes such as water loss from agricultural fields and its influence on downwind precipitation and impacts on water resources. In order to model the impacts of land surface conditions on land-atmosphere exchange and subsequent effects on the environment, it is first necessary to quantify the uncertainty in the field measurements from different sensor systems used for model validation. Using data collected as a part of the 2008 Bushland Evapotranspiration and Agricultural Remote Sensing Experiment (BEAREX08), measurements from nine co-located eddy covariance (EC) systems measuring heat, water and energy flux exchange across the land-atmosphere interface were compared with the two-fold objective of (1) characterizing the inter-instrument variation in the measurements, and (2) quantifying the error in the measurements from each system. The results show that magnitude and cause of measurement error varied significantly among the different EC systems. It also showed that the error estimates derived from the commonly applied Ordinary Least Squares linear regression (OLS) tended to be 10% to 20% greater than those derived from Errors-in-Variables linear regression (EIV). This is largely due to the ability of EIV linear regression to account for the error in the predictor variable. As a result, the use of EIV provides a more realistic way of determining measurement uncertainty and its impact on model calibration and validation.

Technical Abstract: Land-atmosphere interactions play a critical role in regulating numerous meteorological, hydrological and environmental processes. Investigating these processes often requires multiple measurement sites representing a range of surface conditions. Before these measurements can be compared, however, it is imperative that the differences between the instrumentation are fully understood. Using micrometeorological data collected as a part of the 2008 Bushland Evapotranspiration and Agricultural Remote Sensing Experiment (BEAREX08), measurements from nine co-located eddy covariance (EC) systems were compared with the two-fold objective of (1) characterizing the inter-instrument variation in the measurements, and (2) quantifying the error in the measurements from each system. Focusing on four quantities – the sensible and latent heat flux, Carbon Dioxide density, and net radiation – this study also evaluates two well-established regression methods to determine the appropriateness for use with micrometeorological data. These methods are Ordinary Least Squares (OLS) and Errors-In-Variable (EIV) linear regression. The results show that magnitude and cause of measurement error varied significantly among the different EC systems. It also showed that the error estimates derived from OLS linear regression tended to be 10% to 20% greater than those derived from EIV linear regression. This is largely due to the ability of EIV linear regression to account for the error in the predictor variable. Also, EIV linear regression allows for reliable estimates of the actual value that a highly consistent from system-to-system. As a result, the relationship between the measured and actual quantities derived from EIV linear regression could be used to adjust BEAREX08 data collected after the intercomparison period to account for measurement errors.

Last Modified: 9/10/2014
Footer Content Back to Top of Page