Geoscience Reference
In-Depth Information
Fig. 6 Brightness temperature difference retrievals for special sensor microwave imager (SSM/I)
(triangles), advanced microwave scanning radiometer-earth observing system (AMSR-E) (squares) and
the average simulated snowpit data processed through the Helsinki University of Technology (HUT)
microwave emission model. HUT simulations are provided for the N-layer case (circles) and for a single-
layer case where all properties were averaged to one layer (diamonds). Legend in top right identifies marker
shapes and line styles
Table 4 Average brightness
temperature difference for each
intensive observation period
(IOP) as simulated by inverting
the Chang algorithm, using dif-
ferent numbers of layers in the
Helsinki University of Technol-
ogy (HUT) microwave emission
model, and the average retrievals
for advanced microwave scanning
radiometer-earth observing sys-
tem (AMSR-E) and special sensor
microwave imager (SSM/I)
T
B
(19H)-T
B
(37H) (K)
IOP3
IOP4
Chang
18.58
24.72
1-layer
20.23
13.86
2-layer
20.19
13.18
3-layer
20.24
13.31
4-layer
20.21
13.35
5-layer
20.23
13.32
N-layer
20.24
13.46
AMSR-E
16.03
15.71
SSM/I
18.40
15.64
5.3 Differences Due to Layering Detail
It appears that for the CLPX pits, using the HUT radiative transfer model to generate the
scene brightness, temperature difference improves the simulation relative to using the
Chang algorithm approach. Furthermore, Lemmetyinen et al. (
2010
) reported that RMSE
and bias were reduced at these microwave channels when HUT accounted for the multiple
layering of snow, rather than using bulk averages in a single layer.
It was therefore assumed that the best simulation of DT
B
;
V
was provided by the HUT
model run with the N-layer realisation of the CLPX snowpit properties, and the perfor-
mance of simplified layer models should be compared to this. Here the same brightness
