Geoscience Reference
In-Depth Information
Table 8.3
Comparative Analysis for State's and Analyst's Sample Unit
Change Maps
Standard
Deviation
95% Confidence
Interval
State
Source
Mean
New South Wales
State
0.003864
0.007294
0-0.018160
Analyst
0.001831
0.008726
0-0.018935
Northern Territory
State
0.000644
0.000478
0-0.001579
Analyst
0
0
0
Queensland
State
0.011593
0.012670
0-0.036427
Analyst
0.011075
0.025121
0-0.060313
South Australia
State
0.012995
0.014141
0-0.040712
Analyst
0.005625
0.027887
0-0.060284
Victoria
State
0.013600
0.019658
0-0.052130
Analyst
0.006055
0.029289
0-0.063462
Tasmania
State
0.039160
0.026792
0-0.091672
Analyst
0.014050
0.026977
0-0.066925
Western Australia
State
0.032650
0.026897
0-0.085369
Analyst
0.020183
0.061761
0-0.141235
Continental estimate
State
0.012898
0.018069
0-0.048313
Analyst
0.008903
0.032335
0-0.072279
Note:
Differences between state's and analyst's estimates were not significant
(
p
= 0.05) for any state shown.
Table 8.4
Comparative Analysis for State's and Analyst's Sample Unit Change Maps
Standard Deviation
Range
Source
Level
Mean Range
State change maps
Continental
0.012898
0.018069
State
0.000644-0.039160
0.000478-0.026897
Scene
0-0.100215
Analysts
Continental
0.008903
0.032335
State
0.001831-0.020183
0.008726-0.061761
Scene
0.000041-0.068090
0.000163-0.138222
Subsample
0-0.57344
Table 8.5
Potential Error in Land-Cover Change
Estimates for Australia if 90% or 97% of Images
Are Reliable and the Remaining 10% to 3% of
Images Have Various Amounts of Error
Error per Image
90% Correct
97% Correct
10%
1%
0.3%
20%
2%
0.6%
30%
3%
0.9%
8.5. If the interpretation that 10% of the change scenes failed is correct (even though an additional
7% passed with reprocessing) and the error on the failing change scenes is as high as 30%, the
difference in LC change is only 3%. It would be surprising if the error were as high as this, since
no scenes failed for the state with the largest amount (> 80%) of clearing (Queensland). Thus, the
error was likely to be distributed among the states having the least amount of change.
8.4 DISCUSSION AND CONCLUSIONS
The goal of this study was to provide an independent evaluation of the reliability of the estimates
made by state agencies of LC change in Australia from 1990/1991-1995. Traditional approaches
