J. Marotzke et al., Construction of the adjoint MIT ocean general circulation model and application to Atlantic heat transport sensitivity, J GEO RES-O, 104(C12), 1999, pp. 29529-29547
We first describe the principles and practical considerations behind the co
mputer generation of the adjoint to the Massachusetts Institute of Technolo
gy ocean general circulation model (GCM) using R. Giering's software tool T
angent-Linear and Adjoint Model Compiler (TAMC). The TAMC's recipe for (FOR
TRAN-) line-by-line generation of adjoint code is explained by interpreting
an adjoint model strictly as the operator that gives the sensitivity of th
e output of a model to its input. Then, the sensitivity of 1993 annual mean
heat transport across 29 degrees N in the Atlantic, to the hydrography on
January 1, 1993, is calculated from a global solution of the GCM. The "kine
matic sensitivity" to initial temperature variations is isolated, showing h
ow the latter would influence heat transport if they did not affect the den
sity and hence the flow. Over 1 year the heat transport at 29 degrees N is
influenced kinematically from regions up to 20 degrees upstream in the west
ern boundary current and up to 5 degrees upstream in the interior. In contr
ast, the dynamical influences of initial temperature (and salinity) perturb
ations spread from as far as the rim of the Labrador Sea to the 29 degrees
N section along the western boundary. The sensitivities calculated with the
adjoint compare excellently to those from a perturbation calculation with
the dynamical model. Perturbations in initial interior salinity influence m
eridional overturning and heat transport when they have propagated to the w
estern boundary and can thus influence the integrated east-west density dif
ference. Our results support the notion that boundary monitoring of meridio
nal mass and heat transports is feasible.