Regression models for calculating gas fluxes measured with a closed chamber

Portable closed chambers provide a valuable tool for measuring crop photosynthesis and evapotranspiration. Typically, the rates of change of CO2 and water vapor concentration are assumed to be constant in the short time required to make the closed-chamber measurement, and a linear regression model is used to estimate the CO2 and H2O fluxes. However, due to the physical and physiological effects the measurement system has on the measured process, assuming a constant rate and using a linear model may underestimate the flux. Our objective was to provide a model that estimates the CO2 and H20 exchange rates at the time of chamber closure. We compared the linear regression model with a quadratic regression model using field measurements from two studies. Generally, 60 to 100% of all chamber measurement data sets were significantly nonlinear, causing the quadratic model to yield fluxes 10 to 40% greater than those calculated with the linear regression model. The frequency and degree of nonlinearity were related to the measured rate and chamber volume. Closed-chamber data should be tested for nonlinearity and an appropriate model used to calculate flux. The quadratic model provides users of well-mixed closed chambers an alternative to a simple linear model for data sets with significant nonlinearity.