|
|
|
Poster Session
|
Using Remote Sensing technology for dynamic
monitoring forest resources
Designing of RGC method
The designing & implementation of RGC method can be summarized as follows:
- The Landsat Image Map Making.
The landsat image which is geometrically corrected and registered with geographic information is called image map: it is the base map for monitoring forest resources, the carrier for sample layout and the fundamentals of interpretation and classification:
When it is used to monitor an area larger than a county, the image map should be made at the scale of 1:50,000 using TM data. The indispensable elements on the image map are administrative boundaries, inventory boundaries, longitude lines and latitude liens, kilometer nets. The density and intervals of boundary lines are determined by inventory aim and range.
When an image map is made as the base map for monitoring forest resources, the features of forest vegetation and the identifications of other land covers must be taken into consideration in selection of TM bands, ranges of contrast stretch. Gauss-Kruger projection is the standard coordinate system for image geometric correction and geographic information input. The errors of geometric correction and registration with GIS should be less than 0.5 pixel.
On land types interpretation, it is necessary to pay more attention to mosaic precision of different scences of TM data and difference between different seasons, because usually the forest investigation covers very large area.
- Determining the Number of Sample Plots
The RGC method is required to determine the number of sample plots. The RGC has modified CFI system which is based on the principal of systematic two-phase sampling to stratified double sampling. Therefore it is necessary to know:
- The number of very ground factor sample plots which use average stock as major factor. These sample plots should be set up n the field and measured.
- The number of sample plots for calculating area proportions. These sample plots are used to estimate the area proportions of each land types and forest types. Normally, it is done by computer automatic recognition and visual interpretation.
- Ground corrected sample plots. They are sub sample plots of double sampling which are used to amend the area proportions and should be laid in the area proportions sample plots. It is finished mainly in the field and partly indoor such as water bodies.
A. Determining the number of ground sample plots
It is determined by the alternate coefficient and the precisions of forest increment, forest depletion and forest average stock. The formula is shown s follows:
N = ( t2 c2 / E2 % )
In the function :
t is reliable index
c is alternate coefficient
E is desired error
N the number of ground sample plots
At last, it is determined by the largest number of sample plots of forest increment, depletion and average stock plus a certain amount of safety factor.
B. Determining the number of area proportions sample plots
It is determined by the precision of major land type and minimum land type. It can be calculated by following function.
N = ( t2 ( 1-p ) / pE2 %)
p is the estimating proportions of appointed land type
E is the allowable error land type area
t is the reliable index
N is the number of area proportions sample plots
C. Determining the number of ground corrected sample plots
It is determined by the accuracy of automatic recognition and visual interpretation. It can be calculated by following function:
n = N (1-q)
N is the number of area proportions sample plots
q is the accuracy of synthetic interpretation of each land types
n is the number of ground corrected sample plots
|
|
|
|
|
|
|
