1.6 m Wavelength Important for
Operational Remote Sensing at NOAA
Snow and Cloud Mapping Supports Hydrologic and Meteorologic Forecasting
The National Operational Hydrologic Remote Sensing Center (NOHRSC; National
Weather Service, NOAA) uses NOAA Polar Orbiting (AVHRR) and Geostationary
(GOES) satellite data to map snow and clouds throughout the United States.
Daily map products are generated to support operational hydrologic forecast
models and numerical weather prediction models. To map snow and clouds,
NOHRSC analysts use a supervised image classification algorithm that uses
multi-band (wavelength) satellite data. The reflectance of snow and clouds
is similar in the wavelengths measured by GOES channel 1 and AVHRR channels
1 and 2 (Figure 1), therefore discrimination between snow and clouds using
these channels is difficult.
Figure 1. Satellite channel wavelengths in microns (m), and typical
reflectance spectra for snow and clouds.
Snow Can be Distinguished From Cloud at 1.6 m
The 1.6 m wavelength allows significantly improved discrimination between
snow and clouds. At 1.6 m, snow has very low reflectance, while the reflectance
of clouds remains high (Figure 1). Therefore, both cirrus and optically
thick clouds can be directly classified and distinguished from snow at
the 1.6 m wavelength (Warren, 1982). This has been clearly demonstrated
using the operational Landsat Thematic Mapper satellite, which has a channel
centered near 1.6 m (channel 5; 1.57-1.78 m) (Dozier, 1987; Baglio, 1989).
AVHRR Channel 3a Demonstrates Effectiveness of 1.6 m
The NOAA-15 Polar Orbiting Advanced Very High Resolution Radiometer
(AVHRR) satellite sensor includes a 1.6 m channel (Ch. 3a) that was turned
on for testing between March 20 - April 22, 1999. Analysts at the NOHRSC
evaluated the effectiveness of the 1.6 m AVHRR data for mapping snow cover
(Figure 2).
Figure 2. NOAA-15 AVHRR imagery of the vicinity of the
Snake River Valley, Idaho, March 24, 1999.
AVHRR Chanels 2, 3a, and 5 illustrate the reflectance of clouds and
snow in these three wavelength regions (Figure 2). Channel 2 includes portions
of the visible and near-infrared spectrum (e.g. Figure 1). The image contains
significant cloud cover on the left side of the dotted line. The clouds
are transparent in some areas and opaque in others. Clouds and snow have
similar reflectance in this wavelength region. For example, the brightness
of transparent cloud cover at point A is similar to the brightness of snow
at points A' and B. Although the two features can be discriminated visually
based on their different textures, their similar brightness in this channel
does not readily permit discrimination between the two features using numerical
classification techniques.
Channel 3a is the 1.6 m test channel. The low reflectance of snow at
this wavelength, indicated in Figure 1, is clearly apparent here. Cloud
reflectance remains relatively high. For example, point A is significantly
brighter than points A' and B. The darkest areas in the 3a image are unforested,
snow-covered areas. The dark areas further north are also snow covered,
but are less dark because of increased reflectance of forest cover at 1.6
m. The large difference in snow and cloud reflectance at 1.6 m even permits
identification of snow beneath thin transparent clouds, as evident in the
small dark area directly below point A.
Channel 5 lies in the thermal infrared portion of the spectrum, and
brightness in this channel is related to the temperature of the cloud and
land surfaces. In this case, the more opaque clouds have much cooler temperatures
than the land surface, while the more transparent cloud and snow temperatures
are similar (e.g. at A and A'). The ability to discriminate between clouds
and snow using this channel is complicated by the variable temperatures
of both clouds and snow, and their often similar temperatures.
The relative abundance of snow, clouds, and forest cover were determined
using the1.6 m 3a channel and linear spectral unmixing techniques (Figure
3a, c, and e) using AVHRR Channels 1,2, 3a, 4, and 5. The relative abundance
of each feature in a pixel is depicted as
Figure 3. Relative abundance and threshold classification of
(a,b) snow cover, (c,d) forest cover, and (e,f) cloud cover determined
using linear spectral unmixing.
Shades of gray, with darker shades indicating less abundance and lighter
shades indicating more abundance. Images (a) and (c) indicate that both
forest and snow cover contribute to the reflectance of individual pixels
in the northern part of the image. Images (a) and (e) illustrate areas
with transparent clouds, where the land surface features (e.g. snow) contribute
to the pixel reflectance. These relative abundance images were classified
using a simple threshold (Figure 3 b, d, and f) to illustrate the benefit
of the 1.6 m channel for snow/cloud discrimination.
AVHRR 1.6 m Channel Improves Snow and Cloud Classification
Under normal AVHRR operations (without the 1.6 m channel), snow and
cloud classification is based on information illustrated by channels 2
and 5 in Figure 2. Classification is based on all five channels, but channels
1 and 2 are highly correlated with each other, as are channels 4 and 5.
AVHRR channel 3 (3.55 - 3.93 m) adds little or no useful information for
snow/cloud discrimination. The different reflectance characteristics between
snow and clouds at 1.6 m significantly improve snow and cloud discrimination
(Table 1).
Table 1. Assessment of AVHRR Channel 3a for operational snow
and cloud mapping tasks.
| |
AVHRR with Normal Ch. 3 |
AVHRR with 1.6 m Channel |
| Snow/Opaque Cloud Discrimination |
Fair |
Improved |
| Snow/Tranparent Cloud Discrimination |
Poor |
Improved |
| Identification of Snow beneath Transparent Cloud |
Poor |
Improved |
References
Baglio, J.V., and Holroyd, E.W., 1989. Methods for operational snow
cover area mapping using the advanced very high resolution radiometer:
San Juan Mountains Test Study, Research Technical Report, U.S. Geological
Survey, Sioux Falls and U.S. Bureau of Reclamation, Denver.
Dozier, J., 1989. "Remote sensing of snow in visible and near-infrared
wavelengths," Theory and Applications of Optical Remote Sensing,
G. Asrar, ed., John Wiley and Sons, New York.
Warren, S., 1982. Optical properties of snow, Reviews of Geophysics
and Space Physics, 20, 67.
|
weather.gov
