Premium
Sensitivity of Analytical Flux Footprint Models in Diverse Source‐Receptor Configurations: A Field Experimental Study
Author(s) -
Kumari Shweta,
Kambhammettu B. V. N. P.,
Niyogi Dev
Publication year - 2020
Publication title -
journal of geophysical research: biogeosciences
Language(s) - English
Resource type - Journals
eISSN - 2169-8961
pISSN - 2169-8953
DOI - 10.1029/2020jg005694
Subject(s) - mean squared error , footprint , point source , offset (computer science) , flux (metallurgy) , wind speed , sensitivity (control systems) , eddy covariance , mathematics , physics , meteorology , environmental science , geodesy , statistics , geology , optics , paleontology , materials science , computer science , metallurgy , programming language , ecology , engineering , electronic engineering , ecosystem , biology
Flux footprint models simulate the source area of scalar fluxes from a measurement site but are hindered by the assumptions that are rarely satisfied in real field conditions. We conducted artificial tracer (CO 2 ) experiments in unstable conditions over an open field, to evaluate three analytical footprint models (Kormann and Meixner‐KM, Hsieh‐HS, and Schuepp‐SP) under variable source‐receptor settings. Experimental configurations include (i) source length (point to semi‐infinite), (ii) emission height (below and above zero‐plane displacement), (iii) measurement height (1.5 and 2.8 m above ground level), and (iv) offset from the principal wind direction (±60°). Eddy covariance (EC) fluxes were measured using one source‐multiple sampling system approach. Results indicate that KM model is in good agreement with the EC measurements under ideal conditions ( R 2 = 0.81, root‐mean‐square error (RMSE) = 0.008 m −1 ) compared to HS ( R 2 = 0.60, RMSE = 0.014 m −1 ) and SP ( R 2 = 0.72, RMSE = 0.009 m −1 ) models. The KM model captured both footprint maximum and its location. Uncertainties in the model estimates were ascertained by considering random ensembles of wind speed, friction velocity, and Obukhov length. The KM model has resulted in the least uncertainty band containing all the flux observations. Error propagation into model simulations resulting from a progressive deviation of the underlying assumptions was assessed. Model‐to‐measurement discrepancies were positive (flux overestimation) for offset deviations up to ±30° and negative beyond. Modeled fluxes were sensitive to source length and emission height in comparison to the offset deviation. Overall, the KM model showed the least error (9.6 ± 8.12%) for different source‐receptor deviations. Study findings can help in evaluating and improving the analytical models for use under nonideal source‐receptor configurations and scaling measurements for different applications.