1993 satellite data is 48 ft above empty, the area and
RESULTS
capacity estimates for these lower elevations are
Surface area results
not as reliable as those for higher elevations. In
Appendix A presents composite images and
classified water surfaces for each of the Landsat
yielded negative area estimates for elevations below
scenes. In the classification images, blue repre-
546 ft. The underestimate of capacity was estimated
sents water, bluegreen represents shoreline, and
to be not more than about 2000 acre-ft. This error,
white represents pixels that are either unclassi-
because it is associated with the lowest reservoir lev-
fied or masked out. Figure 4 shows the multitem-
el, is carried through to the upper levels as well.
poral water classifications from 1993. Table 9 lists
500,000-acre-ft-loss scenario
the reservoir water surface area as of 1953 and
1985 (from survey data) and as of 1993 as derived
Reservoir managers initially estimated that
from Landsat data. These values are for the reser-
500,000 acre-ft of storage capacity might have been
Table 9. Painted Rock Reservoir water surface area in 1953, 1985, and
1993.
Reservoir
Surface area (acres)
Surface area loss (acres)
elevation
1953
1985
1993
1953 to
1985 to
Date
(ft)
(survey)
(survey)
(Landsat)
1993
1993
4 Dec 93
532.10
112
33
0
112
33
2 Nov 93
578.51
8,571
8,599
7,779
793
821
1 Oct 93
594.65
14,960
15,367
13,328
1,632
2,039
30 Aug 93
604.88
19,840
19,552
17,623
2,217
1,929
29 Jul 93
612.32
23,376
23,332
21,455
1,921
1,877
27 Jun 93
620.61
27,736
27,730
25,888
1,847
1,842
26 May 93
631.06
33,489
33,512
32,303
1,186
1,208
10 May 93
637.75
37,825
37,913
36,485
1,340
1,428
24 Apr 93
644.22
41,732
42,059
40,934
798
1,124
8 Apr 93
651.68
46,792
46,846
45,795
997
1,051
31 Mar 93
655.24
49,106
49,263
48,633
473
630
7 Mar 93
665.86
56,602
56,660
55,141
1,461
1,519
Spillway
661.00
53,200
53,213
52,101
1,099
1,112
voir elevations at the time of the Landsat over-
lost after the 1993 flood. The question is whether
passes. Figure 5 graphs the elevation vs. surface
the loss measured with the Landsat procedure is
area curves for the 1953 and 1985 surveys, the
significantly different from this, given the proce-
dural uncertainties.
500,000-acre-ft-loss case.
In order to make an estimate of area and capac-
Figure 4 shows the location of the 15 ground
ity curves for the 500,000-acre-ft-loss scenario, it is
survey profiles overlaid on the 1993 multitempo-
first observed that because the capacity curve is
ral water classification. The matrix in Table 10
computed as the integral under the area curve, if
summarizes how the elevation as determined
the area curve is multiplied by a factor, the capacity
from the multitemporal Landsat classification
curve computed from it will be multiplied by the
same factor. Given this, a ratio was made of the
compares with that from the 1993 ground survey
capacity of the reservoir at spillway elevation if
profiles. Diagonal elements in this matrix repre-
500,000 acres were lost since 1953 (1,991,700) to the
sent agreement between the Landsat classifica-
1993 estimated capacity at spillway elevation
tion and the ground survey profile.
(2,334,804). The 1993 area and capacity curves were
Appendix B shows cross sections of the 15
then multiplied by this ratio at all points to create
ground survey profiles. The Landsat elevation
the new curves. This yielded a capacity curve with
range is delimited by the dashed lines while the
the capacity at spillway elevation equal to
ground survey elevation is represented by an un-
1,991,700 acre-ft. This is only one method of model-
broken line. The Landsat data can be viewed as
ing these curves, and does not account for different
correct if the unbroken line lies between the
rates of sediment deposition in different parts of
dashed lines.
the reservoir.
11
Back to contents