In tall vegetation canopies, calm weather conditions may result in the formation of an isolated below-canopy air layer. Especially during night, below-canopy carbon dioxide (CO2) fluxes may be concealed from above-canopy eddy covariance (EC) measurements, masking the magnitude of CO2 net ecosystem exchange (NEE). Particular attention to this “nighttime problem” needs to be paid in oil palm (Elaeis guineensis Jacq.) plantations, which can form a dense top canopy and which are grown in tropical regions where wind speeds are often low. Here, we investigated EC-based NEE measurements, wind and micrometeorological patterns, and LiDAR-derived terrain and canopy structure in a mature commercial oil palm plantation in the tropical lowlands of Jambi province, Sumatra, Indonesia. Over 3 years, we assessed the strength of turbulent and thermal mixing and tested five different methods to determine decoupling: (i) single-level (above-canopy) and (ii) two-level (above- and below-canopy) friction velocity (u*), (iii) above- and below-canopy standard deviation of the vertical wind (σw), (iv) bulk Richardson number (Rib), and (v) vertical turbulence intensity (ITurb). Our results show that decoupling is a common phenomenon in this plantation, with the two-level filters of σw and u* being most sensitive in its determination, showing ∼93% nocturnal decoupling. Decoupling is driven by the oil palm canopy structure, which develops a marked blocking layer at ∼10 m above the ground, high frequency of calm weather conditions, inverse temperature gradients and a terrain slope of 3° which was enough to create a thermally-induced drainage flow. Filtering NEE data, especially the two-level filters of σw and u*, challenged estimations of the plantation's CO2 balance due to high uncertainties in nocturnal respiration rates which could not be reduced by adding storage components to the measured fluxes highlighting the importance of additional below-canopy CO2 flux measurements, especially in such a weak-wind and dense-canopy tropical environment.
Read full abstract