importance of reducing the systematic error due to non-linearity in N₂O flux measurements by static chambers

Closed (non-steady state) chamber measurements are often used to determine the gas exchange of N₂O. Many researchers have addressed the underestimation of the emission estimates obtained from closed chamber measurements when using linear regression methods. However, the linear regression method is s...

Full description

Saved in:
Bibliographic Details
Published in:Nutrient cycling in agroecosystems 2008-10, Vol.82 (2), p.175-186
Main Authors: Kroon, P. S, Hensen, A, van den Bulk, W. C. M, Jongejan, P. A. C, Vermeulen, A. T
Format: Article
Language:English
Subjects:
accuracy
Agriculture
Biomedical and Life Sciences
closed-chamber data
data analysis
Emission
Emission measurements
Estimates
ethane
experimental design
fertilizer rates
fertilizers
Fluctuations
Flux
Gas exchange
Goodness of fit
Life Sciences
linear models
Linearity
measurement
new methods
Nitrous oxide
Nonlinearity
Regression analysis
Research Article
Systematic errors
Temporal variations
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Closed (non-steady state) chamber measurements are often used to determine the gas exchange of N₂O. Many researchers have addressed the underestimation of the emission estimates obtained from closed chamber measurements when using linear regression methods. However, the linear regression method is still usually applied to derive the flux. The importance of using non-linear regression methods is demonstrated with data from four fertilizing events each consisting of 1 month of automatic chamber measurements at Cabauw in the Netherlands in the period from July 2005 to July 2006. It is presented that the cumulative emission estimates with the exponential regression method are close to the cumulative emissions estimates with the intercept method. The linear estimates differ by up to 60% of the estimates with the exponential method. The performance of each method is validated using a C₂H₆ tracer and a goodness-of-fit analysis. The goodness-of-fit is much better for the exponential than the linear regression method. The systematic error due to linear regression is of the same order as the estimated uncertainty due to temporal variation. Therefore, closed-chamber data should be tested for non-linearity and an appropriate method should be used to calculate the flux.
ISSN:1385-1314
1573-0867
DOI:10.1007/s10705-008-9179-x