Current estimates of China's ammonia (NH3) volatilization from paddy rice differ by more than twofold, mainly due to inappropriate application of chamber-based measurements and improper assumptions within process-based models. Here, we improved the Jayaweera-Mikkelsen (JM) model through multiplying the concentration of aqueous NH3 in ponded water by an activity coefficient that was determined based on high-frequency flux observations at Jingzhou station in Central China. We found that the improved JM model could reproduce the dynamics of observed NH3 flux (R2 = 0.83, n = 228, P < 0.001), while the original JM model without the consideration of activity of aqueous NH3 overstated NH3 flux by 54% during the periods of fertilization and pesticide application. The validity of the improved JM model was supported by a mass-balance-based indirect estimate at Jingzhou station and the independent flux observations from the other five stations across China. The NH3 volatilization losses that were further simulated by the improved JM model forced by actual wind speed were in general a half less than previous chamber-based estimates at six stations. Difference in wind speed between the inside and outside of the chamber and insufficient sampling frequency were identified as the primary and secondary causes for the overestimation in chamber-based estimations, respectively. Together, our findings suggest that an in-depth understanding of NH3 transfer process and its robust representation in models are critical for developing regional emission inventories and practical mitigation strategies of NH3.
Keywords: Dynamic chamber; Improved Jayaweera-Mikkelsen model; Model simulation; NH3; Paddy field.