Models of instantaneous soil-surface CO 2 efflux (SCE ins) are critical for understanding the potential drivers of soil C loss. Several simple SCE ins models have been reported in the literature. Our objective was to compare and validate selected soil temperature ( Ts)- and water content (theta v)-based equations for modeling SCE ins among a variety of cropping systems and land management practices. Soil-surface CO 2 effluxes were measured and modeled for grain-harvested corn ( Zea mays L.)-soybean [ Glycine max (L.) Merr.] rotations, grain- and stover-harvested continuous corn systems with and without a cover crop, and reconstructed prairies with and without N fertilization on soils with subsurface drainage. Soil-surface CO 2 effluxes, Ts, and theta v were measured from 2008 to 2011. Models calibrated with weekly measured SCE ins, Ts, and theta v throughout the growing season produced lower root mean squared error (RMSE) than models calibrated with several weeks of hourly measured data. Model selection significantly affected SCE ins estimations, with models that use only Ts parameters having lower RMSE than models that use both Ts and theta v. However, the model that produced the lowest RMSE during validation estimated growing-season SCE that did not significantly differ from numerical integration of weekly measured SCE ins. All models had similar residual errors with autocorrelated trends at monthly, weekly, and hourly scales. Autoregressive moving average functions were able to precisely describe the temporal errors. To accurately model SCE ins and scale across time, improvement of temporal errors in Ts- and theta v-based SCE ins models is needed to obtain accurate and precise closure of C balances for managed and natural ecosystems.
