Remote sensing and probabilistic sampling based forest inventory method

• US 20070291994A1
• Filed: 03/23/2007
• Published: 12/20/2007
• Est. Priority Date: 05/03/2002
• Status: Active Grant

• Current Assignee: GeoDigital International Inc.
• Original Assignee: ImageTree Corporation (GeoDigital International Inc.)
• Inventors: Kelle, Olavi, Mathawan, Neerai, Pliszka, Rubert, Macom, Eric
• Granted Patent: US 7,639,842 B2
• Time in Patent Office: Days
• US Class Current: 382/110
• CPC Class Codes: G01C 11/00 Photogrammetry or videogrammetry, e.g. stereogrammetry; Photographic surveying
• CPC Class Codes: G01S 17/88 Lidar systems specially adapted for specific applications
• CPC Class Codes: G01S 17/89 for mapping or imaging
• CPC Class Codes: G01S 7/4802 using analysis of echo signal for target characterisation; Target signature; Target cross-section
• CPC Class Codes: G06V 20/13 Satellite images

Abstract

A remote sensing and probabilistic sampling based forest inventory method can correlate aerial data, such as LiDAR, CIR, and/or Hyperspectral data with actual sampled and measured ground data to facilitate obtainment, e.g., prediction, of a more accurate forest inventory. The resulting inventory can represent an empirical description of the height, DBH and species of every tree within the sample area. The use of probabilistic sampling methods can greatly improve the accuracy and reliability of the forest inventory.

Claims

1. A remote sensing and probabilistic sampling based forest inventory method comprising:

a. processing remote sensing data indicative of tree attribute information for said forest, said remote sensing data comprising at least one of LiDAR data and digital images;

b. defining a sampling frame within said remote sensing data;

c. determining a field plot corresponding to said sampling frame and collecting field plot data therefrom, said field plot data comprising actual tree attribute information;

d. generating a correlated model by combining said field plot data with said remote sensing data corresponding to said sample frame; and

e. applying said correlated model to all said remote sensing data to produce a probabilistic forest inventory.

2. The method of claim 1 wherein said digital images comprise at least one of CIR and multispectral photography.

3. The method of claim 1 wherein said remote sensing data further comprises property boundary information.

4. The method of claim 1 wherein tree attribute information comprises at least one of tree height, diameter, and species.

5. The method of claim 1 wherein defining said sample frame comprises randomly selecting at least one said sample frame.

6. The method of claim 1 wherein said field plot data further comprises field plot measurement, and determining said field plot further comprises plot center location correction.

7. The method of claim 6 further comprising measuring said plot center using a survey grade GPS device.

8. The method of claim 6 further comprising manually correcting said plot center location to a location that results in the best match between tree locations indicated by said remote sensing data and tree locations verified by said field plot measurement.

9. The method of claim 1 wherein said correlated model comprises at least one formula for determining at least one of tree species, tree height and tree diameter.

10. The method of claim 9 further comprising:

a. at least one coefficient associated with said at least one formula; and

b. wherein said at least one formula is a regression formula derived from correlation of said field plot data with said remote sensing data corresponding to said sample frame.

11. The method of claim 1 wherein generating said correlated model further comprising using automatic field tree matching to create a table in which measured field tree records are merged with tree polygon objects based upon geographic proximity, wherein said tree polygon objects are derived from said remote sensing data.

12. The method of claim 11 further comprising manually adjusting said tree matching based upon interpreter estimate that a field tree is either contributing some pixels of a tree polygon that was created, or is not visible from the air because of a larger tree that contributed some or all pixels of said tree polygon.

13. The method of claim 1 further comprises verifying the accuracy of said correlated model.

14. The method of claim 1 further comprises verifying the quality of said correlated model.

33. The method of claim 1 further comprises verifying the accuracy of said correlated model.

34. The method of claim 1 further comprises verifying the quality of said correlated model.

15. A remote sensing and probabilistic sampling based forest inventory method comprising:

a. processing imagery data, said imagery data indicative of tree attribute information for said forest;

b. classifying tree polygons within said imagery data to derive said tree attribute information;

c. correlating field data, said field data comprising at least one of actual tree attribute information and plot center location;

d. generating a correlated model utilizing said tree attribute information derived from said imagery data and said actual tree attribute information; and

e. generating a probabilistic forest inventory by applying said correlated model to all said imagery data.

16. The method of claim 15 wherein said imagery data further comprises at least one of digital images, LiDAR data, and property boundary information.

17. The method of claim 16 wherein said digital images further comprise color infrared photography, and said imagery data further comprises at least one of stand shapes and tree crown polygon shapes.

18. The method of claim 17 wherein said digital image processing further comprises color infrared processing and LiDAR processing;

19. The method of claim 18 wherein said color infrared processing comprises:

i. splitting said color infrared imagery into smaller blocks;

ii. creating a smoothed block data set;

iii. stand fixing/photo interpretation;

iv. shape clipping; and

V. merging smaller blocks into one property file.

20. The method of claim 18 wherein said LiDAR processing comprises:

i. calculating DEM;

ii. selecting highest pixel and subtracting DEM;

iii. mapping digital surface value; and

iv. converting data to gray-scale.

21. The method of claim 15 wherein said classifying tree polygons comprises:

a. superimposing at least one of CIR photography, stand shapes, tree crown polygon shapes and LiDAR data;

b. creating a classifier formula using discriminant analysis; and

c. classifying polygons for all stands and strata.

22. The method of claim 21 wherein said classifying tree polygons further comprises:

a. manually reviewing species strata;

b. calculating an average CIR band for individual tree crowns;

c. calculating a second order gray level texture feature;

d. selecting a subset of stands for classification; and

e. creating a training set for species at strata level.

23. The method of claim 15 wherein correlating said field data further comprises:

a. defining a sampling frame within said imagery data; and

b. wherein said field data is collected from a field plot corresponding to said sampling frame.

24. The method of claim 23 wherein correlating said field data further comprises:

a. determining a geo-referenced plot center such that said field plot corresponds to said sampling frame; and

b. correcting said plot center based upon field measurements.

25. The method of claim 24 further comprising measuring said plot center using a survey grade GPS device.

26. The method of claim 24 further comprising correcting said plot center location to a location that results in the best match between tree locations indicated by said remote sensing data and tree locations verified by field plot measurements.

27. The method of claim 15 wherein correlating said field data further comprises:

a. capturing actual tree attribute information indicative of at least one of tree height and location; and

b. creating match data correlating said actual tree attributes with said tree attributes derived from said imagery data.

28. The method of claim 27 further comprising using automatic field tree matching to create a table in which measured field tree records are merged with tree polygon objects based upon geographic proximity.

29. The method of claim 28 further comprising manually adjusting said tree matching based upon interpreter estimate that a field tree is either contributing some pixels of the tree polygon that was created, or is not visible from the air because of a larger tree that contributed some or all pixels of the tree polygon.

30. The method of claim 15 wherein generating said correlated model further comprises deriving at least one formula for determining at least one of tree species, tree height and tree diameter.

31. The method of claim 30 further comprising:

a. at least one coefficient associated with said at least one formula; and

b. wherein said at least one formula is a regression formula derived from correlation of said field plot data with said remote sensing data corresponding to said sample frame.

32. The method of claim 31 wherein generating said correlated model further comprises utilizing said regression formula and said at least one correlation coefficient to create at least one of a species probability prediction model, a diameter probability prediction model, and a height probability prediction model.

35. The method of claim 31 wherein generating said probabilistic inventory further comprises

a. applying said species probability prediction model, diameter probability prediction model, and height probability prediction model to all said imagery data;

and

b. calculating at least a volume per acre.