Effect of Sampling Depth on Air-Sea CO2 Flux Estimates in River-Stratified Arctic Coastal Waters

Abstract:

In summer-time Arctic coastal waters that are strongly influenced by river run-off, extreme stratification severely limits wind mixing, making it difficult to effectively sample the surface ‘mixed layer’, which can be as shallow as 1 m, from a ship. During two expeditions in southwestern Hudson Bay, off the Nelson, Hayes, and Churchill River estuaries, we confirmed that sampling depth has a strong impact on estimates of ‘surface’ pCO₂ and calculated air-sea CO₂ fluxes. We determined pCO₂ in samples collected from 5 m, using a typical underway system on the ship’s seawater supply; from the ‘surface’ rosette bottle, which was generally between 1 and 3 m; and using a niskin bottle deployed at 1 m and just below the surface from a small boat away from the ship. Our samples confirmed that the error in pCO₂ derived from typical ship-board versus small-boat sampling at a single station could be nearly 90 μatm, leading to errors in the calculated air-sea CO₂ flux of more than 0.1 mmol/(m²s). Attempting to extrapolate such fluxes over the 6,000,000 km² area of the Arctic shelves would generate an error approaching a gigamol CO₂/s. Averaging the station data over a cruise still resulted in an error of nearly 50% in the total flux estimate. Our results have implications not only for the design and execution of expedition-based sampling, but also for placement of in-situ sensors. Particularly in polar waters, sensors are usually deployed on moorings, well below the surface, to avoid damage and destruction from drifting ice. However, to obtain accurate information on air-sea fluxes in these areas, it is necessary to deploy sensors on ice-capable buoys that can position the sensors in true ‘surface’ waters.