Evaluation of the U.S. Navy’s Modular Ocean Data Assimilation System (MODAS) Using South China Sea Monsoon Experiment (SCSMEX) Data

Peter C. Chu,Wang Guihua,Chenwu Fan

DOI: https://doi.org/10.1007/s10872-005-0009-3

IF: 2

2004-01-01

Journal of Oceanography

Abstract:The Navy’s Modular Ocean Data Assimilation System (MODAS) is an oceanographic tool to create high-resolution temperature and salinity on three-dimensional grids, by assimilating a wide range of ocean observations into a starting field. The MODAS products are used to generate the sound speed for ocean acoustic modeling applications. Hydrographic data acquired from the South China Sea Monsoon Experiment (SCSMEX) from April through June 1998 are used to verify the MODAS model. MODAS has the capability to provide reasonably good temperature and salinity nowcast fields. The errors have a Gaussian-type distribution with mean temperature nearly zero and mean salinity of −0.2 ppt. The standard deviations of temperature and salinity errors are 0.98°C and 0.22 ppt, respectively. The skill score of the temperature nowcast is positive, except at depth between 1750 and 2250 m. The skill score of the salinity nowcast is less than that of the temperature nowcast, especially at depth between 300 and 400, where the skill score is negative. Thermocline and halocline identified from the MODAS temperature and salinity fields are weaker than those based on SCSMEX data. The maximum discrepancy between the two is in the thermocline and halocline. The thermocline depth estimated from the MODAS temperature field is 10–40 m shallower than that from the SCSMEX data. The vertical temperature gradient across the thermocline computed from the MODAS field is around 0.14°C/m, weaker than that calculated from the SCSMEX data (0.19°–0.27 °C/m). The thermocline thickness computed from the MODAS field has less temporal variation than that calculated from the SCSMEX data (40–100 m). The halocline depth estimated from the MODAS salinity field is always deeper than that from the SCSMEX data. Its thickness computed from the MODAS field varies slowly around 30 m, which is generally thinner than that calculated from the SCSMEX data (28–46 m).