Three parameters must be known to use the thin boundary-layer model (or other bulk transfer models) for CO, flux between water and air: the concentration of dissolved CO*, CO,,,,,, the concentration of CO, in the air immediately above the water, COZ(atm), and the wind velocity, which is used to determine the appropriate transfer coefficient. These parameters change hourly and from day to day in a nonlinear fashion, so the frequency of measurements is an important factor in determining the accuracy of flux estimates for any period. To achieve a high frequency measurement, we developed a self-contained, solar-powered, in situ sampling system that continuously measures and records CO,,,,,, COZ(atm), and windspeed. Unique to this technique is an underwater in situ equilibration chamber (ISEC). The ISEC was tested in a shallow wetland pond in which changes in both CO,,,,, and COZ(ntm) were large. The data obtained showed that large errors may result from extrapolating flux calculations made from short-term data (e.g. daily) to longer time periods.The most commonly used method for estimating gas flux across the air-water interface is the thin boundarylayer (TBL) model, which requires measurements of windspeed and of the concentrations of the gas in air and water. Gas concentration data are usually acquired by periodic sampling, and the frequency of sampling is usually limited by accessibility to the site and the time available for analyses.When measurements are made for the purpose of calculating CO2 flux in lakes where diel changes in water and air arc small, limitations on sampling frequency may not be serious. However, there are two common situations in which very frequent measurements in both air and water may be necessary: eutrophic lakes and shallow ponds-both of which have strong diel changes due to high rates of photosynthesis and respiration. Even where diel changes are not so strong (as in oligotrophic lakes), other considerations, such as large changes in windspeed, may make sampling frequency important.Because CO2 flux is proportional to the dzjkwzce bcAcknowledgments