New generation of satellite-derived ocean thermal structure for the western north pacific typhoon intensity forecasting

• A new generation method of satellite-derived ocean thermal structure is developed.
• The new method derives high resolution profiles with 26 subsurface layers.
• Derived ocean thermal structure has higher accuracy.
• Ocean thermal structure over the west North Pacific can be derived in near-real time, suitable for operational purposes.
• The error in SST cooling and air-sea flux estimations is significantly reduced.

Ocean thermal structure is critical for the intensity change of tropical cyclones (TCs). It has been operationally derived from satellite altimetry for TC forecasting and research. The existing derivation is, however, based on a simple two-layer method; as a result, only two isotherms can be obtained to coarsely characterize the subsurface ocean thermal structure. Improvement on the vertical resolution to enhance ocean characterization is desirable for more accurately estimating ocean’s energy supply for TC intensity change.In this study, we present a new generation of derivation to improve ocean’s subsurface characterization for the Western North Pacific Ocean (WNPO) because this region has the highest TC occurrence on the Earth. In addition to the presently used two isotherms for the depths of 20 °C and 26 °C isotherms (D20 and D26), we derive continuous isotherms from D4 up to D29 (maximum 26 subsurface layers) to characterize the subsurface ocean thermal structure in detail. This is achieved through applying a large set (>38,000) of in situ Argo thermal profiles to regression development. A smaller set of in situ Argo profiles (>7000), independent of those used for regression, is utilized for validation, to assess the accuracy of the new derivation. The root-mean-square differences (RMSDs) between the derived and the in situ isotherms are found to be within ∼10–20 m for the upper isotherms (D20 to D29) and within ∼40–60 m for the lower isotherms (D4 to D19). No significant biases of derived isotherms are found.In addition to increasing the vertical resolution from two layers to 26 layers, higher accuracy for the two isotherms of D20 and D26 is also achieved. For example, in the existing two-layer method, D20 in the northern WNPO is grossly overestimated with a high RMSD of ∼80–100 m; the new method significantly reduces the RMSD to ∼20 m. Better subsurface characterization leads to improvement in important TC-intensity related parameters, like during-cyclone sea surface temperature (SST) cooling and air–sea enthalpy flux supply. Based on a series of ocean mixed layer numerical experiments using 16 randomly-selected profiles, we find that the error in SST cooling (air–sea flux supply) can be reduced from 36% (120%) to 13% (20%).