Cloud Remote Sensing: Major Projects
Cloud Detection: Geostationary
Geostationary cloud detection example
|
|
| |
|
GOES-7 satellite image pair over the southeast U.S. and Mexico, valid at 2000 and 2100 UTC in September 1994. In these visible images dark tones denote low reflectivity and bright tones denote high reflectivity. Black pixels denote open ocean in the Gulf of Mexico and eastern Central Pacific.
Geostationary example with cloud mask
SERCAA
employs a temporal differencing algorithm that operates on pairs
of geostationary satellite images taken over the same location
from sequential scans, usually one hour apart. Cloud detection
is performed on a pixel-by-pixel basis by comparing satellite-measured
changes in visible counts and infrared brightness temperatures
to predicted changes in the corresponding clear-scene values over
the same time period. The image to the left contains the SERCAA
geostationary cloud mask for the 2100 UTC image above. Yellow
pixels denote regions where brightness changes over the last hour
exceeded the predicted clear-scene changes. Note that in the midlatitudes,
the yellow pixels are on the eastern edge of the clouds, since
the clouds are moving from west to east; in the tropics, the yellow
pixels are on the western edge of the clouds, since in those latitudes
the clouds are moving from east to west. Red denotes dynamic threshold
pixels whose brightnesses exceed those of the yellow temporal
differencing pixels. Finally, a static spectral test is employed
to capture stationary clouds whose brightnesses have not changed
significantly in the past hour; these pixels are cyan. All other
gray pixels are considered to be cloud-free by the geostationary
algorithm. As with the polar-orbiter cloud detection tests, note
that the overall SERCAA geostationary cloud mask is a composite
of several individual cloud detection tests.
 |
|