The determination of the latent and sensible heat fluxes at the sea surface viewed as an inverse problem

Abstract

The determination of the latent and sensible heat fluxes at the air‐sea interface is formulated here as an inverse problem: given an upper ocean model, one seeks the successive values of the “nonsolar” surface heat flux Fnsol (sum of the latent, sensible, and net infrared fluxes) that yield a nearly exact simulation of the observed evolution of the sea surface temperature (SST). The model is otherwise forced by surface insolation and wind stress data. Provided that an estimate of the infrared flux is available, the sum of the latent and sensible heat fluxes is directly obtained as the difference between Fnsol and the infrared flux. A major interest of this approach is that all surface inputs (stress, SST, long‐wave and shortwave radiation) can be measured from a satellite. In this preliminary exploration of the inverse flux estimation technique, attention is restricted to situations where the oceanic advection is weak so that a one‐dimensional ocean model can be used. The inverse problem, in a rudimentary form, is solved using a simple iterative shooting method. The inverse flux estimation technique is tested using a 2‐year data set from station P, in the Gulf of Alaska, and a 2‐week data set from the Long‐Term Upper Ocean Study (LOTUS) experiment, in the Sargasso Sea. The first tests are performed using regularly sampled in situ measurements of the stress, SST, and insolation, rather than actual satellite data. At both test sites the inverse‐derived estimates of Fnsol prove to be consistent with the observed changes of the upper ocean heat content over periods of a few days. On the other hand, the inverse‐derived heat fluxes exhibit a large diurnal variability that is not present in classical bulk‐derived estimates. This difference in the short‐term variability of the two flux estimates is as yet unexplained. The final tests at LOTUS involve actual satellite data for the isolation and SST. Using the few available satellite measurements, the inverse technique still provides a realistic mean flux estimate. The simulated evolution of the SST in between the satellite observations agrees well with in situ observations, and a fair simulation of the deeper temperatures is obtained.