WMOD00275

• Title: A Comparison of Snow Telemetry and Snow Course Measurements in the Colorado River Basin
• Prepared For: Journal of Hydrometeorology
• Prepared By: Dressler, Fassnacht, Bales
• Date: 8/1/2006
• Weather Modification - Doc Type: Report

i'-J[> ~.

400

DRESSLER ET AL.

AUGUST 2006

a. Gunnison

300

200

100

o
100

80
E
E 60
w
$: 40
C/)
20

o
100

80

60

40

20

o
1500 2000

b. San Juan

c. Salt-Verde - SNOTEL
- - SNOTEL + snowcourse (mm!)

-0- SNOTEL + snowcourse Qinear fill)

........... snowcourse linear fill

2500 3000 3500 4000 4500
elevation, m

FIG. 5. The 30 Mar 1993 hypsometrically interpolated SWE
(normalized by basin area) by subbasin using various combina-
tions of data. Refer to Fig. 3 for dataset combinations.

basins, SWE estimated from interpolated SNOTEL
data (Figs. 5a-c) were consistently higher than that for
snow course data, with the greatest differences, as much
as 150 nun or 40% over the Gunnison at 3000-3500-m
elevation, at peak SWE. The rmse for interpolated
SWE (1 km2 grid) versus observed point values, calcu-
lated as the root mean square of the absolute difference
between the estimated SWE in the 1-km grid block
from interpolation and that from the observed station
value, increased through May and then declined as
snow cover became nearly depleted (Fig. 6). Average
interpolated SWE for the Colorado basin as a whole
was greater when using snow course versus SNOTEL
data, with the largest difference at lower elevations
(1500-2000 m) (Fig. 7).
Variograms were constructed to examine spatial con-
tinuity as it related to distance, with a view to measure-
ment error and the geospatial relationships of the two
datasets. Variograms were calculated by plotting vari-
ance in SWE against lag distance, determined from the

500
450

400

E 350
E 300
u.i
en 250
::E
0::: 200

150

100

50
o
12/29

709

-SNOTEL
",/1<" SNOTEL + snowcourse (mmt)
- SNOTEL + snowcourse (linear fill)
- snowcourse (linear fill)

2/2

5/18

6/22

3/9 4/13
date

FIG. 6. The 30 Mar 1993 rmse interpolation error, calculated
from the difference of the 1 km2 gridded values and the point data
used to compute the gridded values within the Colorado basin.
Refer to Fig. 3 for dataset combinations.

fixed station locations. Important features of the vari-
ogram were the (a) range, or the separation distance of
stations for which no significant increase in the squared
difference (variance) between station pairs occurs; (b)
sill, or the plateau variance value at the range; and (c)
nugget effect, or the variance discontinuity at the vari-

*- 40
J
w
I-
a 20
Z
(f)
E
e 0
-
Q)
(,)
c
Q)
Q) -20
:t::
'5

-40
1500

-+- SNOTEL + snowcourse (mmt)
___ SNOTEL + snowcourse (linear fill)
- snowcourse (linear fill)

2500 3500
elevation zone, m

450C

FIG. 7. The 30 Mar 1993 elevation dependence of interpolated
SWE volume differences for collocated sites over the Colorado
River basin. Interpolations are masked by maximum snow extent
from Advanced Very High Resolution Radiometer snow-covered
area. Refer to Fig. 3 for dataset combinations.