 |
NRL Monterey, Marine Meteorology Division
|
|---|
|
NRL Monterey, Marine Meteorology Division
|
|---|
| The quest for quantitative and qualitative wind information over oceanic
regions lacking in conventional observations was the prime motivation for
the development of methods to track clouds in sequential satellite
images. These data are increasingly assimilated into numerical models,
like COAMPS and NOGAPS. For forecasters, the applications are many,
ranging from analysis of tropical cyclones, to the estimation of wind
shear, to the location of jet streams, to the analysis of coastal gap
winds. Perhaps the outstanding advantage of these winds is their
geostationary coverage. Large and complex wind systems can be followed
easily over huge portions of the globe. FAQ on geostationary winds, including explanation of wind barbs:
|
| The automated wind-extraction algorithms are from the University of
Wisconsin-Madison/Cooperative Institute for Meteorological Satellite
Studies (UW-CIMSS). "Cloud drift" winds are derived using infrared and
visible channels, while "water vapor" winds are derived using water vapor
imager and sounder channels. Utilization of different channels allows for
the derivation of wind vectors at numerous levels within the atmosphere,
from near the surface to the top of the troposphere. Winds are derived using a sequence of three images. Features targeted in the first image (cirrus cloud edges, gradients in water vapor, small cumulus clouds, etc.) are tracked within the second and third images yielding two displacement vectors, which are averaged to give the final wind vector. Vectors are height assigned in a two-step process. The first utilizes the measured radiances of the target and is based upon the spectral response function of the particular satellite channel utilized. Once determined, the brightness temperature is compared with a collocated numerical model guess temperature profile (Navy NOGAPS model forecast fields), from which an initial height is estimated. The final vector height is derived in the post-processing of the vector field. The raw winds are run through an auto- editing process, a two-stage, three- dimensional objective analysis of the wind fields. This scheme utilizes conventional data assimilation, neighboring wind "buddy" checks, and numerical model analysis for wind vector editing and final height assignment. Currently, water vapor and cloud drift infrared winds are derived using a 30-minute interval, four kilometer horizontal resolution data every three to six hours, depending on the satellite. Visible cloud drift winds are derived every three or six hours, using one kilometer horizontal resolution data at varying time intervals (either 3, 7, 15, or 30 minute intervals) depending on satellite used. Theoretically, a new wind set can be derived with the availability of every new image if computer resources are available. The visible cloud drift winds represent conditions at low levels from 950 to 600 mb. The infrared cloud drift winds represent winds from low to mid levels, from 950 to 400 mb. Rawinsonde validation of the infrared winds yielded a root-mean-square vector error of 7.07, a standard deviation of the vector difference of 4.26, and a bias of -0.40 (all values in m/s). Water vapor winds represent the mid and upper troposphere, with winds from 500 to 100 mb. Rawinsonde validation of the vapor winds yielded a root- mean-square vector error of 7.47, a standard deviation of the vector difference of 4.45, and a bias of -0.26 (all values in m/s). Source of error characteristics: Upper-Tropospheric Winds Derived from Geostationary Satellite Water Vapor Observations, 1997. Velden et. al., Bulletin of the AMS. Vol. 78, No. 2, pp. 173-195.
FAQ on geostationary winds, including explanation of wind barbs:
|
| Satellite-derived wind vectors can be generated over the entire globe
using the same processing algorithm. These winds can provide data to
weather forecasters in regions devoid of conventional data sources, such
as radiosonde data, aircraft and ship reports, and other atmospheric wind
observations. The information provided by these winds can help define
weather features and systems, such as the extent and magnitude of winds
around tropical cyclones and regions of strong shear/divergence that can
help/hinder the development of severe weather events. These winds are
truly "three dimensional," giving not only a high density of winds in the
horizontal but also multiple winds in the vertical. The winds can also be used to view mesoscale events such as low-level jets, coastal gap winds, and sheltering in the lee of headlands. In this regard, it is useful to compare them to other kinds of surface remotely sensed winds, e.g., from SSM/I or scatterometers. In addition, derived wind vectors can provide numerical modelers another data set for the verification of their models. By comparing the analyzed wind fields to current analysis or forecast fields, modelers can determine how individual models "handle" certain weather phenomena. |
| Height assignment of the individual wind vectors can be a problem. The
heights depend the particular satellite channel employed and the
processing method. The observed brightness temperatures are a measurement
of a layer of the atmosphere, not of a single level. The assignment of a
height the layer depends on the wavelength being used and atmospheric
conditions. Since specific levels are assigned in the vector height
derivation processes (to comply with the standards of conventional data
and NWP assimilation), errors are inherent in the height assignment
process. In addition, wind vectors tracked within multiple cloud layer regions can also lead to errors in the assigned heights. Numerous quality control checks are performed in the wind vector derivation process, but due to the volume of wind vectors produced, not all situations can be corrected. Cloud-drift winds cannot be derived in cloud-free areas because of the absence of targets. Also, winds will not be derived when clouds do not lend themselves to tracking because of the absence of well- defined embedded features to follow. Thus, there are large "blank" regions on the products without winds. Water vapor winds are also impossible to derive where there are no gradients to follow, so that there are "blank" regions on images without wind retrievals. Winds are not available at the surface of the earth because of the difficulties inherent in tracking very low clouds and fog. Geostationary winds are limited to tropical and mid-latitude regions. Visible cloud drift winds are only available during the daytime. This coverage shifts pole-ward or equator- ward seasonally with changing solar illumination. |
| Low-level Winds Rushing down Yellow Sea in Visible Drift Winds | IR Winds similar to Visible Winds, but extend to higher altitudes | Water Vapor Winds Appear over upper-level water vapor gradients |
|---|---|---|
) |
) |
) |
| The image on the left shows winds streaming (orange) down the Yellow Sea
based on tracking cloud features in the GMS visible channel. These show
northerly low level winds from 950 to 800 mb. Don't be misled into
thinking that it is calm over land on either side of the Yellow Sea simply
because there are no vectors plotted. There are no vectors because there
are too few clouds to track. Notice that the orange vectors continue just
south of 40 N. However, green vectors in this vicinity give the wind
speeds based on a higher band of clouds (600 to 800 mb). This example
illustrates the capability of the product to show wind shear. The IR winds image (center) shows a similar distribution of winds as the visible winds image, except that they extend higher into the atmosphere. For example, the blue winds in Northern Japan (400-599 mb) trace middle clouds. The water vapor winds panel (right) shows very few winds in the region of Japan, Korea, or the Yellow Sea. This is because there are few water vapor gradients to track there. It does not mean that the winds over these areas are light or nonexistent! In the tropics, where there are gradients to track, these upper-level winds are much more abundant. The blue winds (100 - 250 mb) are near the top of the troposphere. FAQ on geostationary winds, including explanation of wind barbs:
| ||
| IR Drift Winds show Gap Winds in the Pacific off Northern Japan | Visible Winds Show Elevated Speeds in Gap Wind Region | Vapor Winds show no low-level Detail |
|---|---|---|
) |
) |
) |
| The IR cloud vectors (left) show winds at low-levels through an island gap
in Northern Japan. Winds in the vicinity of the gap appear to be about 25
knots. Blue winds over Asia show flow at roughly the 500 mb level. The visible wind vectors (center) show similar low-level conditions near the gap between the islands. It analyzes a single wind there with speed of 40 knots (blue). The water vapor winds (right) show no evidence of a gap wind because the product only shows winds from about 500 mb to the top of the troposphere. For example, it shows winds near the tropopause just north of Borneo. | ||
| Image I: Visible Cloud Drift Winds | Image II: IR Cloud Drift Winds | Image III: Water Vapor Winds |
|---|---|---|
) |
) |
) |
| Quiz Image I: You are a heliocopter pilot operating at low elevation (801 -950 mb) over the west coast of Japan. Based on this image, what is the approximate direction and speed of the winds you would encounter?
a) South at 70 knots. Quiz Image II: a) There are
insufficient clouds for the algorithm to track. Quiz Image
III: a) From the south at low latitudes, from the north
at midlatitudes.
See Answers Below. | ||
| Answers: I: d) Orange vectors indicate low level flow off Western Japan. FAQ on geostationary winds, including explanation of wind barbs: University of Wisconsin CIMSS II: a) Insufficient clouds to track. Dark shading indicates the cloud- free sea surface. III: b) easterly at low latitudes, westerly at midlatitudes. |
Author: Tom Lee Last Updated: Wed Dec 18 07:26:12 2002 Produced by: The Composer (Ver: 1.1.2 ) |
|