Accounting for Parameter Uncertainties in Model Verification: An Illustration with Tropical Sea Surface Temperature

M. Benno Blumenthal Lamont-Doherty Geological Observatory of Columbia University, Palisades, New York

Search for other papers by M. Benno Blumenthal in
Current site
Google Scholar
PubMed
Close
and
Mark A. Cane Lamont-Doherty Geological Observatory of Columbia University, Palisades, New York

Search for other papers by Mark A. Cane in
Current site
Google Scholar
PubMed
Close
Restricted access

Abstract

A simplified model originally developed to simulate sea surface temperature (SST) in the tropical Pacific is extended in two directions. First, the model is formulated and run for the tropical Atlantic. Second, optimal values are determined for a number of model parameters, especially those determining surface heat flux. In the Pacific, the difference between model SST and data SST is the size of the probable error due to uncertainties in the heat flux, 0 (35 W m−2). in the equatorial Atlantic the discrepancies may also be attributed to heat flux errors but model-data differences in the Atlantic near coastal regions north of 10°N and south of 10°S are too large to be explained by heat flux errors: model problems involving other physical processes must be involved. By giving the best result possible in the face of our uncertain knowledge of the parameters and observations, the optimal-fit procedure unambiguously indicates which oceanic features require model changes in order to be adequately characterized. Only in these cases can potential improvements be evaluated without improving the database.

Abstract

A simplified model originally developed to simulate sea surface temperature (SST) in the tropical Pacific is extended in two directions. First, the model is formulated and run for the tropical Atlantic. Second, optimal values are determined for a number of model parameters, especially those determining surface heat flux. In the Pacific, the difference between model SST and data SST is the size of the probable error due to uncertainties in the heat flux, 0 (35 W m−2). in the equatorial Atlantic the discrepancies may also be attributed to heat flux errors but model-data differences in the Atlantic near coastal regions north of 10°N and south of 10°S are too large to be explained by heat flux errors: model problems involving other physical processes must be involved. By giving the best result possible in the face of our uncertain knowledge of the parameters and observations, the optimal-fit procedure unambiguously indicates which oceanic features require model changes in order to be adequately characterized. Only in these cases can potential improvements be evaluated without improving the database.

Save