Using Digital Camera and Eddy Covariance Data to Track Vegetation Phenology and Carbon Dioxide Fluxes in the Badain Jaran Desert

Abstract

Understanding on relationships between seasonality of vegetation phenology and photosynthesis is lacking for desert ecosystems. We used digital camera (i.e., PhenoCam) to monitor the phenology of forest (i.e., 2 sites with one being closer to a lake) and grassland (i.e., 1 site) ecosystems in the Badain Jaran Desert, China. The vegetation phenology was quantified using vegetation indices calculated from the red, green, and blue digital numbers in images obtained by the PhenoCams. Additionally, various meteorological variables were continuously measured, and gross primary production (GPP) was obtained using the eddy covariance technique at the grassland site. The difference between the phenological periods extracted from the PhenoCam images and the artificial visual method was small (≤6 days), indicating that the digital camera can effectively obtain desert vegetation phenology. The key meteorological factors affecting changes in the vegetation indices were identified, with temperature being the most important factor (i.e., correlation coefficients = 0.4–0.8 and p-value < 0.001 for all three study sites). Although precipitation showed weak correlation with the vegetation index (correlation coefficient = 0.18–0.14, p-value < 0.01), rapid increases in the vegetation index were observed in response to precipitation events. Vegetation indices were strongly correlated with GPP variations at the grassland, and the strongest correlation was observed in the green-up stage (correlation coefficient = 0.67 to 0.85, p-value < 0.001). The highest GPP lagged about 1 month behind the peak in the vegetation indices in summer (June–August). Our results can markedly improve the knowledge of desert ecosystem processes and aid in assessing the influence of future climate changes in drylands.