Info

Table 1, the data coverage for ERS-1 is about 500 km, with a footprint of 50 km. Before the ERS-1 wind data were implemented operationally, two data assimilation experiments — one with the use of ERS-1 scatterometer wind data (the test run), and one without (the control run) — were conducted for a period of about 40 days, starting on November 21, 1995, and ending on December 31, 1995. The results shown in Table 3 are based on 40 cases of forecasts. As one can see from the table, inclusion of ERS-1 scatterometer winds in the NCEP GDAS leads to a small positive impact on the forecast of 10 m winds during the five days of forecasts. Note that the improvement in the forecasts is very small and may not be significant from the overall statistics. However, significant improvements are most often seen in some special synoptic situation where the inclusion of scatte-rometer winds can make a large improvement in identifying the storm intensity and circulation center, as will be discussed in the next section. Based on results of these parallel experiments, the ERS-1 wind data were implemented operationally in NCEP's global data assimilation experiment in the spring of 1997.

QuikSCAT, a NASA oceanographic satellite, was launched in July 1999. The main mission of the satellite was to provide global coverage of ocean surface winds for NAVY and NOAA applications. The QuikSCAT scatterometer single cell measures a footprint of 25 km, and with a total of 72 cells it can cover a data swath of 1800 km, and can provide more than 100000 data points in a 6-hour window during a synoptic analysis cycle (see Table 1). Early impact experiments with the full resolution (25 km) QuikSCAT wind data in the NCEP GDAS and NASA GDAS did not lead to any positive impact. One of the reasons for the lack of early impact investigations was the incompatibility of the fine data resolution (25 km) and the much coarser model and analysis resolution (of greater than 100 km) configured in the NCEP and NASA global data assimilation systems.

Follow-up experiments using coarser resolution QuikSCAT wind data by a super-averaging approach subsequently led to a positive impact on the global data assimilation systems of NCEP (Yu, 2003) and NASA (Atlas et al., 2001).

The super-averaging approach, sometimes also called the supper-ob approach, is designed mainly to thin out large numbers of satellite data. In order to use thinned-out QuikSCAT data in NCEP GDAS experiments, the data are averaged over a 1° x 1° longitude-latitude grid box, reducing the total data numbers to nearly one-fourth of the original, and resulting in a 100 km effective resolution for the data. Table 4 shows mean sea level pressure forecast errors with respect to midlatitude deep ocean buoys during data assimilation experiments conducted from October 2, 2001, to November 10, 2001, to investigate the impact of QuikCAT

winds (100 km resolution) on the NCEP operational global data assimilation system (T170, L42). The results, based on about 40 forecast cases, clearly show that use of QuikSCAT winds significantly improves the sea level pressure forecasts when compared to a buoy's observations. Similar conclusions are found when comparing the forecasts of model 10 m winds with in situ observations from midlatitude deep ocean buoys (Table 5) and tropical TOGA buoys (Table 6). Based on these statistics, QuikSCAT winds of 100 km resolution were implemented operationally in the NCEP's global data assimilation system on January 15, 2002. About one week later, ECMWF also implemented the QuikSCAT wind data of a coarser resolution in its global data assimilation, on January 22, 2002.

After the operational implementation of the coarse resolution (100 km) of QuikSCAT wind

Table 4. Mean sea level pressure forecast errors (mb) with respect to about 45 midlatitude deep ocean buoys during data assimilation experiments (from October 2 to November 10, 2001) conducted to investigate the impact of QuikSCAT scattero-meter winds (100 km resolution) on the NCEP operational global data assimilation system (T170, L42).

Table 4. Mean sea level pressure forecast errors (mb) with respect to about 45 midlatitude deep ocean buoys during data assimilation experiments (from October 2 to November 10, 2001) conducted to investigate the impact of QuikSCAT scattero-meter winds (100 km resolution) on the NCEP operational global data assimilation system (T170, L42).

Forecast Hours (hours)

Without Use of QuikSCAT

(%)

Improvement (hours)

0 0

Post a comment