The complex underlying surface underestimates the carbon sink capacity of reed salt marsh wetlands: A case study from the Liaohe River Estuary, China.

When applying the eddy covariance (EC) technique to measure CO2 exchange at the ecosystem-atmosphere interface, we found that small-scale heterogeneous underlying surfaces can significantly impact measurements. However, this influence is often overlooked in previous studies. In this study, we selected the reed salt marsh wetland at the Liaohe River Estuary, China, as our study area. Focusing on the Suaeda salsa patches and tidal channels, which are heterogeneous surfaces within the EC system's observation area, we employed the footprint model and wind direction distribution to classify the general underlying surface (AUS) into two categories: homogeneous (T1) and heterogeneous (T2). Our results revealed that the contribution of heterogeneous underlying surfaces to the overall flux is consistently controlled by wind direction. Although the diurnal, seasonal and cumulative trends of net ecosystem exchange (NEE) of carbon dioxide were similar across the three underlying surface types, T1 exhibited significantly greater amplitude in these variations. During the growing season, the linear slope fitted to the cumulative change of NEE showed T1 (-128.11) < AUS (-96.87) < T2 (-54.28), indicating the highest carbon sequestration efficiency in T1. Consequently, T1 had a substantially higher cumulative net CO2 absorption (824.53 g C·m-2) compared to AUS (518.66 g C·m-2) and T2 (199.98 g C·m-2), suggesting that the carbon sequestration capacity of the original reed wetland was underestimated by approximately 37 %. Nonlinear fitting analysis revealed that during the day, heterogeneous underlying surfaces significantly affected the fitting parameters of the double M model, whereas the impact on the V model fitting parameters at night was negligible, highlighting that the influence of heterogeneous surfaces is primarily diurnal. Tidal channels were identified as the primary factor contributing to the underestimation of carbon sequestration capacity. Therefore, our study emphasizes that when applying EC observation systems in carbon cycle research, the presence of small-scale heterogeneous surfaces should not be overlooked. Data screening can improve the accuracy of these observations.