Data processing uncertainties may lead to an overestimation of the land carbon sink of the Tibetan Plateau.

Wei et al. (1) report that the Tibetan Plateau is a net CO2 sink that is 4 times greater than previously estimated. Model simulations in the same study suggest that the CO2 sink will increase further as the climate becomes warmer and wetter in the future. These results have significant implications for understanding the responses of the carbon cycle to climate warming. However, Piao et al. (2) used different methods to estimate the size of China’s terrestrial ecosystem carbon sink and proposed that it ranges from 170 Tg C⋅y−1 to 350 Tg C⋅y−1. This result from Piao et al. (2) shows that the estimated extent of the Tibetan Plateau carbon sink in Wei et al. (1) is an overestimate. Using net ecosystem production (NEP) to estimate the size of CO2 sinks may neglect the effects of several key processes, for example, aquatic carbon export (3) and multiple anthropogenic activities (4). In addition, uncertainties associated with eddy covariance (EC) data processing may also lead to the overestimation of the carbon sink.

The results from some of the 10 self-processed sites used in Wei et al. (1) differ from findings presented in other comparative research. For example, Wei et al. (1) calculated NEP for Arou to be 31.7 g C⋅m−2⋅y−1, while two other studies have calculated NEP for the same year to be 144 and 137 g C⋅m−2⋅y−1 (5, 6). For alpine steppe sites, the average NEP for Nam Co was 17.1 g C⋅m−2⋅y−1 in Wei et al. (1), which is much higher than the findings in Wang et al. (6). The site where the estimates in Wei et al. (1) differ most from those in other studies is Ali. Wei et al. (1) estimate that the Ali alpine steppe is a strong carbon sink (206.9 g C⋅m−2⋅y−1), while Wang et al. (6) estimate that it is only a weak carbon sink. Where other studies have used methods developed by the Max Planck Institute for Biogeochemistry (7, 8), Wei et al. (1) do not. We also checked the daily NEP recorded for several sites used in Wei et al. (1) and found that there were unusual spikes in the daily variations of NEP (Fig. 1). We suggest that Wei et al. (1) may not have implemented outlier diagnosis and rejection prior to data gap filling, meaning that outliers may have significantly affected the quality of the gap-filled data (9).

Fig. 1.

Comparison of NEP for different sites used in Wei et al. (1) and in Wang et al. (6) for the same year. (A) Arou in 2015, (B) Muztag in 2016, and (C) Nam Co in 2008.

The data processing methods used in Wei et al. (1) may have led to an overestimation of the carbon sink of the Tibetan Plateau, particularly for alpine steppe areas. There is also an error in the vegetation type classification at the Mount Everest site, where alpine steppe (10) is incorrectly classified as alpine shrub in Wei et al. (1). Observational EC data are critical for driving model calculations and for regional upscaling of carbon sink estimates. Widely used data-spike detection and gap-filling methods may result in a more reasonable estimation of the carbon sink of the Tibetan Plateau.