VALIDATION OF ATMOSPHERIC HEAT BUDGET PARAMETERS IN THE WESTERN INDIAN SECTOR OF THE SOUTHERN OCEAN

• Main Authors: Alvarinho J, M. Sudhakar, R. Ravindra
• Other Authors: The Pennsylvania State University CiteSeerX Archives
• Format: Text
• Language: English
• Subjects: Remote sensing; Atmosphere; Databases; Marine; Statistics; Analysis; Comparison; Satellite; India Bay; Indian; Prydz Bay; Southern Ocean; International Polar Year
• Online Access: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.642.8789; http://www.isprs.org/proceedings/XXXVII/congress/8_pdf/3_WG-VIII-3/17.pdf

The work reports validation of numerical weather forecasts (NWF), QuikSCAT-based wind speed and AMSR-E sea surface temperature (SST) for the surface atmospheric boundary layer in the Indian sector of the Southern Ocean under the International Polar Year. The database consists of wind speed, air temperature, relative humidity, sea level pressure collected along the ship route from Durban (32.4°S, 30.4°E) to India Bay (62.4°S, 18°E) and from Prydz Bay (69.35°S, 76.17°E) to Mauritius (20.15°S, 57.48°E) during February- March 2007. The NCEP-NCAR and ECMWF meteorological data have been validated. Both the NWFs overestimate mean sea level pressure by-5 to-12 mb; the root mean square difference (RMSD) ranges from 1.7 to 2.3 mb. The validation of air temperature yields a RSMD of 0.8°C and the correlation exceeds 0.9. The relative humidity from NWFs is underestimated by 3%. The validation of QuikSCAT-based wind speed yields a root mean square difference (rmsd) of 1.4 m/s. The validation of AMSR-E SST is found to be highly significant with high correlation of 0.9 and yields a RMSD of 0.7°C. The comparison of sensible and latent heat fluxes estimated from in situ data and that from NCEP-NCAR and ECMWF yields RMSD in the range 17-19 and 38-42 Wm-2, respectively, with ECMWF data being poorly correlated to the in situ data than the NCEP-NCAR product. There is no significant difference in the turbulent heat flux estimated by Liu-Katsaros-Bussinger and Kondo’s algorithm.