LAND COVER CHANGE DUE TO OIL AND GAS EXPLORATION IN THE HAYNESVILLE SHALE REGION FROM 1984 TO 2011

Size: px
Start display at page:

Download "LAND COVER CHANGE DUE TO OIL AND GAS EXPLORATION IN THE HAYNESVILLE SHALE REGION FROM 1984 TO 2011"

Transcription

1 LAND COVER CHANGE DUE TO OIL AND GAS EXPLORATION IN THE HAYNESVILLE SHALE REGION FROM 1984 TO 2011 D A N I E L U N G E R A P R I L 2 3, Division of Environmental Science Arthur Temple College of Forestry and Agriculture Stephen F. Austin State University

2 Introduction Objectives Methods Results Conclusion

3 Introduction Energy demands are on the rise Oil & gas are the primary sources for energy New technologies (hydraulic fracturing & directional drilling) have increased natural gas exploration Concerns have arisen about possible environmental impacts

4 Haynesville Shale Lies under approximately 2,890,770 ha of land 9 parishes in Louisiana 16 counties in Texas 3-4 km deep, 91 m thick Expected to contain 7 trillion m 3 of natural gas Drilling started in 2007 By October 2010, 891 wells were completed 715 in Louisiana, 176 in Texas

5 Introduction Objectives Methods Results Conclusion

6 Objectives Classify 6 satellite images ( ) Recode the classified maps to extract only forest, agricultural land, and well pads Determine the amount of change that occurred from 1984 to 2011 within each category Perform land cover metrics to determine surface disturbance

7 Introduction Objectives Methods Results Conclusion

8 Project Description Image Acquisition Radiometric Correction Geometric Correction Image Classification Accuracy Assessment Change Detection Land Cover Metrics Results

9 Methods Image Acquisition Landsat-5 Thematic Mapper (TM) images downloaded from USGS Global Visualization Viewer (Glovis) glovis.usgs.gov Year Date Row/Path November November November November /37 24/38 25/37 25/ November November November November March March April April /37 24/38 25/37 25/38 24/37 24/38 25/37 25/38 Year Date Row/Path November November December December /37 24/38 25/37 25/ December December December December October October October October /37 24/38 25/37 25/38 24/37 24/38 25/37 25/38

10 Methods Radiometric Correction Histogram Subtraction Sensor Sun Geometric Correction Geometrically correct when downloaded WGS 1984, UTM Zone 15 N

11 Methods Mosaicked Image

12 Methods Image Classification Unsupervised Classification Forest, agricultural land, well pads, and other Well pads did not have an unique digital signature Process created to isolate the well pad class from the agricultural land and other classes

13 Methods Accuracy Assessment 300 points randomly selected using a stratified random generation Minimum of 50 points per class (agricultural land, forest, and well pad)

14 Methods Change Detection Post Classification Comparison Summary matrix generated in ERDAS IMAGINE 2010 Between each 5-6 year interval Between 1984 and

15 Methods Land Cover Metrics Patch Per Unit (PPU) Patches = 3 Pixels = 25 PPU = per km 2 m = the total number of patches n = the total number of pixels in the study area λ is a scaling constant equal to the area of a pixel Patches = 17 Pixels = 25 PPU = per km 2 Square Pixel (SqP) A = the total area of all pixels P = the total perimeter of all pixels Perimeter = 480 m Area = 14,400 m 2 SqP = 1 Perimeter = 720 m Area = 18,000 m 2 SqP = 1.34

16 Introduction Objectives Methods Results Conclusion

17 Results Classified Image

18 Results

19 Classified Classified Results Accuracy Assessment 1984 Reference Other Forest Agricultural User s Well Pad Total Land Land Accuracy Other Forest Land % Agricultural Land % Well Pad % Total Producer s Accuracy % 81.91% 90.91% Overall Classification Accuracy = 87.33% Overall Kappa Statistic = Reference Other Forest Agricultural User s Well Pad Total Land Land Accuracy Other Forest Land % Agricultural Land % Well Pad % Total Producer s Accuracy % 80.68% 93.55% Overall Classification Accuracy = 88.33% Overall Kappa Statistic =

20 Classified Classified Results Accuracy Assessment 1994 Reference Other Forest Agricultural User s Well Pad Total Land Land Accuracy Other Forest Land % Agricultural Land % Well Pad % Total Producer s Accuracy % 85.44% 96.77% Overall Classification Accuracy = 88.00% Overall Kappa Statistic = Reference Other Forest Agricultural User s Well Pad Total Land Land Accuracy Other Forest Land % Agricultural Land % Well Pad % Total Producer s Accuracy % 79.12% % Overall Classification Accuracy = 84.67% Overall Kappa Statistic =

21 Classified Classified Results Accuracy Assessment 2006 Reference Other Forest Agricultural User s Well Pad Total Land Land Accuracy Other Forest Land % Agricultural Land % Well Pad % Total Producer s Accuracy % 86.05% % Overall Classification Accuracy = 89.00% Overall Kappa Statistic = Reference Other Forest Agricultural User s Well Pad Total Land Land Accuracy Other Forest Land % Agricultural Land % Well Pad % Total Producer s Accuracy % 90.91% 96.55% Overall Classification Accuracy = 88.00% Overall Kappa Statistic =

22 Results Comparison: well pad to well pad High resolution photo from Classified image from 2011.

23 Results Comparison: building/parking lot to well pad High resolution photo from Classified image from 2011.

24 Results Comparison: poultry houses to well pad High resolution photo from Classified image from 2011.

25 Results Change Detection 1984 Agriculture Forest Well Pad Other Agriculture 428,862 (59.5%) 90,708 (4.8%) 18,301 (61.6%) 25,557 (9.7%) Forest 254,260 (35.2%) 1,742,530 (92.9%) 6,691 (22.5%) 22,880 (8.6%) Well Pad 7,454 (1.0%) 1,444 (0.1%) 1,015 (3.4%) 1,865 (0.7%) Other 30,762 (4.3%) 41,359 (2.2%) 3,716 (12.5%) 214,301 (81.0%) 1989 Agriculture Forest Well Pad Other Agriculture 483,750 (85.9%) 372,399 (18.4%) 9,269 (78.7%) 40,039 (13.8%) Forest 58,700 (10.4%) 1,606,260 (79.3%) 291 (2.5%) 42,891 (14.8%) Well Pad 5,320 (0.9%) 4,016 (0.2%) 1,083 (9.2%) 2,253 (0.8%) Other 15,658 (2.8%) 43,692 (2.1%) 1,133 (9.6%) 204,961 (70.6%) 1994 Agriculture Forest Well Pad Other Agriculture 509,315 (56.3%) 145,526 (8.5%) 6,833 (53.9%) 17,558 (6.6%) Forest 353,914 (39.1%) 1,511,840 (88.5%) 3,337 (26.3%) 47,501 (17.9%) Well Pad 12,965 (1.4%) 1,912 (0.1%) 1,208 (9.6%) 1,391 (0.5%) Other 29,265 (3.2%) 48,862 (2.9%) 1,294 (10.2%) 198,997 (75.0%)

26 Results Change Detection 2000 Agriculture Forest Well Pad Other Agriculture 445,630 (65.6%) 117,491 (6.1%) 12,457 (71.3%) 24,904 (8.9%) Forest 202,440 (29.8%) 1,745,720 (91.1%) 1,711 (9.8%) 56,214 (20.2%) Well Pad 4,927 (0.7%) 4,145 (0.2%) 1,297 (7.4%) 1,597 (0.6%) Other (3.9%) 49,243 (2.6%) 2,011 (11.5%) 195,703 (70.3%) 2006 Agriculture Forest Well Pad Other Agriculture 406,034 (67.6%) 100,666 (5.0%) 4,722 (39.4%) 24,705 (9.0%) Forest 156,732 (26.1%) 1,812,050 (90.3%) 1,940 (16.2%) 42,515 (15.6%) Well Pad 6,220 (1.1%) 5,947 (0.3%) 2,506 (21.0%) 2,126 (0.8%) Other 31,495 (5.2%) 87,411 (4.4%) 2,797 (23.4%) 203,838 (74.6%) 1984 Agriculture Forest Well Pad Other Agriculture 359,648 (49.9%) 135,806 (7.3%) 15,219 (51.1%) 25,453 (9.6%) Forest 317,405 (44.0%) 1,645,890 (87.7%) 10,629 (35.8%) 39,308 (14.9%) Well Pad 6,703 (0.9%) 8,071 (0.4%) 789 (2.7%) 1,236 (0.4%) Other 37,583 (5.2%) 86,267 (4.6%) 3,086 (10.4%) 198,606 (75.1%)

27 Results Change Detection Total amount of land cover change = 24% Total change to well pads from agricultural land and forest = 0.1% (14,774 ha)

28 PPU (per sq km) Results Land Cover Metrics Patch per unit changes in the classified areas Agricultural Land Forest Land Well Pad Year

29 PPU (per sq km) Results Land Cover Metrics Patch per unit changes in Haynesville Shale region Year

30 SqP Results Land Cover Metrics Square pixel changes in the classified areas. 600 Agricultural Land Forest Land Well Pad Year

31 SqP Results Land Cover Metrics Square pixel changes in the Haynesville Shale region Year

32 Introduction Objectives Methods Results Conclusion

33 Conclusion Low user s accuracy resulted in lower accuracy in change detection and land cover metrics More agriculture land was converted to well pads than forest land A higher amount of agricultural land was fragmented than forest land. The entire Haynesville Shale region has experienced increased fragmentation over the last 25 years Each category and the entire study area had high complexity in shape Overall, oil and gas exploration had impacted forest and agricultural land within the Haynesville Shale region

34 Questions