Geoscience Reference
In-Depth Information
Fig. 8 Residual between brightness temperature difference simulated for models with layer thickness
[10 cm versus the 10-cm layer simulation, and plotted as a function of this bulk layer thickness. All data
represented as open circles and snowpits of depth [100 cm are excluded
gradient of the standard deviation graph in Fig.
9
, the increase in error due to thickening of
snow layers beyond 10 cm was found to be 0.053 ± 0.006 K cm
-1
. The total error from
loss of stratigraphic information can therefore be estimated from the snow depth and
number of layers and is presented as a function of SWE (calculated by applying the CLPX
average density for snowpits\100 cm depth) in Fig.
10
for snowpits whose stratigraphy is
averaged to 1, 2 or 3 layers.
Additionally, the estimated error in DT
B
;
V
can be interpreted as an estimate of the SWE
error for this regime where the roughly linear relationship between DT
B
and SWE holds,
similarly to Eq. (
1
) but for vertically rather than horizontally polarised microwaves.
DSWE
=
DDT
B
;
V
was calculated from linear regression of the known SWE and N-layer
calculated DT
B
;
V
values
for
the
pits
of
depth \100 cm,
and
found
to
be
2.45 ±
was converted into an approximate SWE
0.09 mm
SWE
K
-1
, and so error in DT
B
;
V
, DDT
B
;
V
error using:
D SWE
¼
D SWE
DDT
B
;
V
DDT
B
;
V
ð
17
Þ
For a CLPX snow profile of 100 cm depth (170 mm SWE), simplification of the stra-
tigraphy from the measurement resolution of 10 cm down to a single layer of average
properties leads to DT
B
;
V
simulations that contain a 4.8-K error related to the loss of
stratigraphic information, equivalent to 13 mm SWE (7 % of total). In a two-layer model,
this error would be reduced to 2.1 K (5.6 mm SWE, 3 % of total) and for a three-layer
model 1.2 K (3.3 mm SWE, 2 % of total).
The individual user must decide model detail based on the trade-off between precision
and computational expense, and it is hoped that this approach will inform such decisions. A
user might determine a given threshold for fractional or absolute error in SWE, and from
this information could determine the number of layers to use in their model based on the
snow depth.
