Correlation lengths for estimating the large-scale carbon and heat content of the Southern Ocean

Mazloff, MR, Cornuelle BD, Gille ST, Verdy A.  2018.  Correlation lengths for estimating the large-scale carbon and heat content of the Southern Ocean. Journal of Geophysical Research-Oceans. 123:883-901.

Date Published:



Algorithms, budget, carbon, heat, model, Oceanography, pco(2), sink, Southern Ocean, space, Spatial correlation lengths, variability


The spatial correlation scales of oceanic dissolved inorganic carbon, heat content, and carbon and heat exchanges with the atmosphere are estimated from a realistic numerical simulation of the Southern Ocean. Biases in the model are assessed by comparing the simulated sea surface height and temperature scales to those derived from optimally interpolated satellite measurements. While these products do not resolve all ocean scales, they are representative of the climate scale variability we aim to estimate. Results show that constraining the carbon and heat inventory between 35 degrees S and 70 degrees S on time-scales longer than 90 days requires approximately 100 optimally spaced measurement platforms: approximately one platform every 20 degrees longitude by 6 degrees latitude. Carbon flux has slightly longer zonal scales, and requires a coverage of approximately 30 degrees by 6 degrees. Heat flux has much longer scales, and thus a platform distribution of approximately 90 degrees by 10 degrees would be sufficient. Fluxes, however, have significant subseasonal variability. For all fields, and especially fluxes, sustained measurements in time are required to prevent aliasing of the eddy signals into the longer climate scale signals. Our results imply a minimum of 100 biogeochemical-Argo floats are required to monitor the Southern Ocean carbon and heat content and air-sea exchanges on time-scales longer than 90 days. However, an estimate of formal mapping error using the current Argo array implies that in practice even an array of 600 floats (a nominal float density of about 1 every 7 degrees longitude by 3 degrees latitude) will result in nonnegligible uncertainty in estimating climate signals.