Blended Winds with QSCAT and ADEOS-II
The "SeaWinds on ADEOS-II Scatterometer" mission was operational from April 10, 2003, to October 24, 2003 (when the ADEOS-II satellite lost power).
For these 6.5 months, blended winds were generated based on QSCAT, ADEOS-II, and NCEP. With two scatterometers the
blending window can be reduced from 12 hr to 7.5 hr; i.e. every 6-hourly NCEP global field is blended with 7.5 hr each
of QSCAT and ADEOS-II, centered on the time of the NCEP data.
-
Comparison of satellite swaths used for blending QSCAT & ADEOS-II vs. QSCAT-only
-
The propagation of fronts and storms is better resolved by two Scatterometers
Comparison of satellite swaths used for blending QSCAT & ADEOS-II vs. QSCAT-only
The following figures show the number of satellite revs used for blending and their geographic distribution.
When two scatterometers are available the blending window can be reduced to 7.5 hours , while still covering
the entire globe. This contrasted with single sactterometer blending (QSCAT only), when a 12 hour blending
window is necessary for global coverage.
Click on the images for enlarged versions of plots.
Fig. 1. QSCAT (red swaths) and ADEOS-II (blue) revs used for 7.5hr blending window on day 105 (4/15/2003),
(a) at 6hr, and (b) at 12hr.
Fig. 2. Time of rev data with respect to analysis time, i.e. center of blending time, (a) at 6hr,
and (b) at 12hr.
Each satellite rev traverses the globe from South to North (ascending swath) and then North to South (descending swath).
In Fig. 1.a, the first ADEOS-II swath used for blending is 1739, of which only the descending part is used (from
the East China Sea to the Southwest of Australia in the Indian Ocean). The second ADEOS-II rev is 1740, which ascends
from south of South America to the Gulf of Mexico, and then descends from the Bay of Bengal to the ice edge around
Antarctica.
In Fig. 1, QSCAT revs are marked in blue, and ADEOS-II revs in red. The white numbers refer to the last two digits of the
rev numbers.
In Fig. 2, revs are colored according to the time difference of the swath data with respect to the analysis time,
i.e. the center of the blending window. The white numbers printed on top of the swaths refer to the the maximum
time difference (in hours) of the swath. For example, the ADEOS revs (1739 to 1743), start within 3.75hr before
day 105.25 (the descending swath of rev 1739), reach day 105.25 just to the west of the southern tip of India in the
descending swath of rev 1741, and end almost 3hr after the analysis time in the short ascending segment of rev
1743 at (140° W, 30° S).
The two triangular areas in Fig. 1.b and 2.b mark those swath data that are used for blending at day 105.25 and at day
105.50: in the North Atlantic just east of Cape Hatteras (70° W, 40° N; QSCAt rev 19896), and in the South Pacific
(140° W, 50° S; ADEOS-II rev 1743).
When using only one sactterometer (QSCAT only), the blending window needs to expanded to 12 hours in order to
achieve global coverage. The next two figures illustrate the distribution of revs from only one scatterometer for
the same two consecutive times as in figures 1 and 2.
Fig. 3. QSCAT-only revs used for 12hr blending window on day 105 (4/15/2003), (a) at 6hr, and (b) at 12hr.
Fig. 4. Time of rev data with respect to analysis time, i.e. center of blending time, (a) at 6hr,
and (b) at 12hr.
When using a 12hr blending window, about four revs are used in two consecutive blending fields. In Fig. 3, QSCAT revs
19893 - 19896 are used both at 6hr (a) and 12hr (b). In Fig. 4, the re-used revs are marked in red in Fig. 4.a (0 to 6hr
past analysis time), and in blue in Fig. 4.b (0 to 6hr before analysis time).
The propagation of fronts and storms is better resolved by two Scatterometers
To demonstrate the effect of two scatterometer (and the smaller blending window of 7.5hr), the propagation
of fronts and storms is illustrated below.
As an example, the global distribution of wind stress curl on day 199.75 (7/18/2003) is shown below in NCEP data,
QSCAT + NCEP blended winds, and QSCAT & ADEOS-II + NCEP blended winds. Attention is focused on a storm feature
in the area of 75° -105° E x 40° -60° S. The fine-scale structure of this storm is very well resolved by
the two scatterometer blended winds, but the single scatterometer blending puts this feature too far to
the East. The reason for this misplacement is that the QSCAT data used for blending in this area come from
a satellite swath that is almost 6hr ahead of the analysis field. In the two-scatterometer blending, the
satellite data are within 1hr of the analysis time.
The same small area is also shown in Fig. 8 below.
Click on the images for enlarged versions of plots.
Fig. 1. NCEP curl on day 199.75 (7/18/2003 18:00)
Fig. 2. Blended winds curl on day 199.75, for (a) QSCAT only (one scatterometer), and (b) QSCAT and ADEOS-II (two scatterometers).
The following figures are time-longitude diagrams at 49.75° S, from 0° to 360° E, for 18 days (starting at day 193).
The first figure includes negative curl values (blue colors), as well as positive curl values (red colors). For clarity,
the following figures only include the negative features. The storm feature at 90° E, 50° S from figures 1. and 2.
is indicated by the 3 boxes in figures 3., 4. and 5.
Fig. 3. QSCAT & ADEOS-II + NCEP blended winds, curl for 18 days at 49.75° S (negative and positive curls).
Fig. 4. NCEP, curl from day 193-211 at 49.75° S (negative curls only).
Fig. 5. Blended winds curl from day 193-211, for (a) QSCAT only, and (b) QSCAT & ADEOS-II (same as in Fig. 3, but only negative curls plotted).
The propagation is much better resolved with two scatterometers (Fig. 5.b) than with only one scatterometer (Fig. 5.a).
In the latter case, the propagating storm "stalls" very often and then jumps to the next location, due to the fact that 12hr
of data are used for blending every 6-hourly field, so that 6 hours of scatterometer data are always "re-used" in two
consecutive 6-hourly fields.
Fig. 6. NCEP, curl from day 218-236 at 49.75° S.
Fig. 7. Blended winds curl from day 218-236, for (a) QSCAT only, and (b) QSCAT & ADEOS-II.
The 3 small boxes in Fig. 4 & 5, and 4 small boxes in Fig. 6 & 7 indicate several days for which the time sequence
of propagating curl features is shown in the following figures.
Each figure shows 4 time panels (every 6 hours), and 3 different datasets: NCEP, QSCAT blended with NCEP, and QSCAT &
ADEOS-II blended with NCEP. In each panel the zero-countour line from the NCEP dataset is repeated for reference.
Fig. 8. 6-hourly time sequence at 40° - 60° S, 75° - 135° E, for (a) day 199, (b) day 200, and (c) day 201.
On day 199 12hr single and dual blended winds agree very well, and the sharp front of negative curl is within the
area of negative curl in the NCEP data (although much narrower and with higher amplitude). At the next time step, day 199 18hr, the dual blended winds agree very well with the location in NCEP, whereas the single blended winds are already 6hr ahead, placing
the curl front too far to the East. 6 hours later, on day 200 0hr, single and dual blended winds agree again.
The global map of curl for day 199 18hr is shown in Fig. 1 & 2: the area of interest is indicated by the small box in the lower left part of the figures.
Fig. 9. 6-hourly time sequence at 35° - 55° S, 35° - 120° E, for (a) day 228, (b) day 229, (c) day 230, and (d) day 231.
last modified on March 19, 2004
Go
Back
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)