the rain rate as observed during a period of 10 days has been reported earlier
(Schanda and Mätzler, 1981).
GLOBAL SNOW MAPS RETRIEVED FROM NIMBUS-7
SCANNING MULTICHANNEL MICROWAVE RADIOMETER DATA
The Nimbus-7 satellite, launched into a sun-synchroneous orbit on
October 24, 1978, carries a five frequency (6.6, 10.7, 18, 21, 37 GHz) dual
polarized, conical scan microwave radiometer (SMMR) with correspondipg surface
resolution (cell sizes) of approximately 150 x 150 km^ to 30 x 30 km^. The I
SMMR operates on alternate days due to space-craft power limitations. One full I
coverage of the entire earth is completed every 5 days with many overlapping Hl
cells in the near to polar regions. SMMR was designed for the global determi- |
nation of environmental parameters of weather and oceans. |
|
For snow detection combinations of the outputs of the various
frequency channels can be selected to achieve a significant discrimination
and quantitative determination of a few important parameters of the snow cover.
In a recent investigation (Kuenzi et al. 1981 and 1982) with the aid of an
interactive image display system the retrieval algorithms for the extent of
dry snow, for the onset of snow melt and for the determination of the snow ih
depth have been developped. SMMR data sets of two periods October/November Ii
1978 and February/March 1979 have been used.
The algorithm for calculating the extent of dry snow makes use of the
decrease of brightness temperatures with increasing frequency : T, (37 GHz) - |
Tg (18 GHz) « D, where the value D can be adjusted to obtain correspondence to |
the ground truth data.
36 GHz
ue to
Figure 6 presents a map of snow cover of the northern hemisphere of
the dates November 11, 12, 13, 1978. The dark grey level corresponds to
Tg (37) - Tg (18) » - A Kc and- isiddentified snow-free land surface, the values
- 5 K to - 8 K are presented as light grey areas corresponding to thin (« 10 cm)
or patchy snow, while all values below - 9 K, presented as white areas correspond i
to a cover of dry snow of more than ~ 10 cm thickness, it should be noted here Il
that this algorithm is valid only for snow cover over land and not over ice Hil
shields (Greenland) or sea ice. We remind here to the fact that moist snow And
exhibits a brightness temperature spectrum like Fig. 2 while the same cover of |
dry snow behaves like Fig. 1. If now within the days of observation due to
sunshine over day or due to warm air masses a large area of the snow cover
becomes moist, the resulting change of the spectrum can be detected and the
regions of melting snow can be delineated. Figure 7 presents à snow cover map
of Northern America of March 1, 3, 5, 1979. Additionally to the algorithm for
mapping of dry snow (as applied in Fig. 6) a supplementary algorithm utilizes |
here the temporal change of the T, (37) - Tg (18) values. Areas for which this |
change during the three days surpässed a given limit (> 12 K) were identified |
as melting snow and are presented in light grey. A more detailed explanation
of algorithms used and a more thorough discussion of the results achieved wi th
the Nimbus-7 SMMR data is given in the above-mentioned publications. It is worth
mentioning that all these results have been verified by the ground truth made
available by a large number of snow observing stations in Canada, Finland and
the USSR. This was particularly important for the estimation of the snow depth.
a day
fallen
» full
ation.
807
a M P
EP
"m. .
(aL MEE ke ipee eati