Direction accuracy

No directions

+/- 20°


measurement from DMSP and scatterometer measurements from ERS-1/2 and QuikSCAT, are shown in Table 1. It is important to note that the accuracy of satellite ocean surface winds is about 2m/s or 10% for wind speed, and about 20° for wind direction for both the scat-terometer and radiometer measurements. These wind error statistics are borne out by validation studies which compare satellite-derived surface winds with in situ buoy reports (e.g. Gemmill et al., 1999). From Table 1, one can also see that the resolution of satellite ocean surface winds ranges from 25 km to 50 km, and the coverage of the satellite swath varies, ranging from about 500 km for ERS-1/2, 600 km for NS CAT, to 900 km for QuikSCAT on both sides of the spacecraft. Figure 3 shows typical satellite swaths for these three satellites. As such, during a 6-hour window within a synoptic analysis cycle, the number of satellite ocean surface wind data can range from several thousands to hundreds of thousands of data points, depending on the resolution of the footprint design for any particular satellite. When compared with the conventional marine observations of ships and buoys, which are typically less than 1000 for a synoptic analysis cycle, these satellite ocean surface winds constitute the most important source of wind data to fill the void of the global oceans.

Computers And Sattleites Ocean Analysis
Figure 3. Scatterometer data coverage in six hours during a synoptic analysis cycle for ERS-2 (top panel), NSCAT (middle panel), and QuikSCAT (bottom panel).

2.2. Atmospheric analysis and data assimilation systems

In the early days of NWP operations during the late 1950's and the 1960's, with limited computer speed and storage space, the Cressman analysis (Cressman, 1958) scheme was the operational scheme at NMC (National Meteorological Center, the predecessor of NCEP) and other NWP operational centers. The Cressman analysis scheme is a basically two-dimensional spatial interpolation scheme, and therefore it is a very efficient scheme. However, it suffers a deficiency in its arbitrariness of weights assigned to each observation as a function of the distance between observations and grid points to be analyzed. With the rapid advance of computer power and storage during the 1970's and 1980's, the optimum interpolation (OI) scheme (Gandin, 1965) replaced the Cressman scheme and became the main operational atmospheric analysis scheme at most of the operational NWP operational centers in the world. Then, during the early 1990's, owing to the vast improvement in computer power, it became feasible to apply the variational analysis principles of Sasaki (1970) to a viable atmospheric analysis scheme (Le DiMet and Telegrand, 1986), including some of the attributes of the OI scheme for treating the observations from various satellites. The main advantage of OI and variational analysis schemes lies in their ability to handle a variety of satellite data with various error attributes, including ocean surface wind data retrieved from scatterometers and radiometers.

The NCEP global operational forecasting model was changed from a grid point model to a spectral model in 1980, starting with R30 resolution and 12 vertical levels (Sela, 1980), hereafter referred to as (R30, L12). With the increase of computer power, the NCEP operational spectral model has improved both the horizontal and vertical resolutions throughout the last two decades, with the most current operational spectral model having a configuration of T354 and L64, respectively, for its

Table 2. Evolution of NCEP global models and atmospheric analysis schemes from the 1980's to the present time.




Date of





0 0

Post a comment