Irina SMIRNOVA, Alexandra RUSANOVA

Transcription

1 Irina SMIRNOVA, Alexandra RUSANOVA Monitoring of Landscape Changes Due to Petroleum Fields Exploitation, Construction of Oil Pipelines and Oil Terminal in the Northern Part of the Timan-Pechorian Petroleum Province Using Multitemporal ALOS and LANDSAT Data VNIIKAM Birzhevoy proezd, 6, St.-Petersburg, Russia

2 Test areas location The study area is situated in the Varandey region (northern part of TimanPechorian petroleum province) in tundra zone. The industrial development of the area was essentially made active in connection with construction of the oil terminal "Varandey", being basic object for export of petroleum extracted in the Timan-Pechorian petroleum province, and new oil pipeline Southern Khilchuou Varandey coming into service in These constructions render mechanical, chemical and thermal influence on frozen grounds, but also lead to significant pollution of lakes and rivers. 2 Landsat 4 RGB ( ) test areas and their numbers

3 Objectives to identify the effective tool for monitoring of different landscape changes using multisensor and multitemporal satellite data (ALOS AVNIR 2, Landsat 4 TM, Landsat 7 ETM +) by employing different change detection techniques; to detect the changes caused by human activity (mechanical disturbance of a surface, pollution of lakes and rivers and others) and changes caused by natural factors (changes in coastal line and thermokarst lakes) using ALOS AVNIR 2 and Landsat data; to estimate level of suspended sediments concentration in polluted lakes.

4 Methodology Computer processing and analysis of the data were realized on the basis of GIS software (ERDAS Imagine, Map Info) and includes: creation of databases of satellite images obtained in different years; image to image rectification; relative radiometric calibration; interactive interpretation of different satellite images with compilation of vector layers and calculation of the changed objects areas; processing of satellite images for change detection (using generation of color composite images from pair of images acquired in different years, image differencing); creation of spectral curves of test objects using different bands of multitemporal satellite images.

5 Methodology Relative radiometric calibration of one image in relation to another is made using equation: L' = a *L + b, where L' value of brightness of calibrated image, L - value of brightness of initial image, a (gains), b (offset) - coefficients of the equation. The coefficients a and b are calculated using the system from two linear equations: Y1 = b+ax1 Y2 = b+ax2, where X1, X2 values of brightness within the limits of reference sites (sand clear lakes) of the image which is corrected; Y1, Y2 values of brightness within the limits of reference sites (sand clear lakes) of the image to which calibration is made.

6 Data used Test area 1 ALOS AVNIR 2 (2007), RGB 432 LANDSAT 4 TM (1988), RGB 432

7 Changes in a coastal zone of sea: sites of shore abrasion (blue color) and sites of coast accumulation, increase of the area of sand deflation and the area of new constructions (red color) Changes caused by human activity and natural factors Test area 1 RGB: R ALOS AVNIR2 (3 band), 2007; G and B LANDSAT 4 TM (3 band),1988. The figures with arrows specify the areas of changed objects of coastal zone numbers of changed thermo-karst lakes

8 Images differencing Test area 1 Result of change classification based of difference ALOS AVNIR2 (3 band), 2007 and Landsat 4 TM (3 band), 1988 (red sites with increased brightness value; blue sites with decreased brightness value)

9 Data used Test area 2 ALOS AVNIR2 RGB 421 ( ) LANDSAT 7 ETM+ RGB 421 ( ) LANDSAT 4 TM RGB 421 ( ) test sites location and their number

10 Comparisons of spectral curves using Landsat data of 1988, 2000 and ALOS AVNIR2 of 2007 for different object Brightness value test site 1 Test area 2 Brightness value test site 2 Spectral bands Lake polluted up to 2007 Brightness value test site 3 Spectral bands Lake polluted up to 2007 Brightness value test site 4 Spectral bands Lake polluted up to 2007 Spectral bands Lake polluted up to 2007 Landsat 4 TM (red); Landsat 7 ETM+(blue); ALOS AVNIR2 (green)

11 Comparisons of spectral curves using Landsat data of 1988, 2000 and ALOS AVNIR2 of 2007 for different object Test area 2 test site 5 test site 9 Shallow lake (concentration of sediments is increased insignificantly) Lake with salt water test site 7 test site 6 test site 8 Dried up shore of the thermokarst lake Dried up from 1988 to 2000 thermo-karst lake Old dried up thermo-karst lake Landsat 4 TM (red); Landsat 7 ETM+(blue); ALOS AVNIR2 (green)

12 Pollution of thermokarst lakes Color compositing Test area 2 RGB: R ALOS AVNIR2 (3 band), 2007; G and B LANDSAT 7 ETM + (3 band), 2000

13 Changes classification based on images difference Test area 2 Result of change classification based on difference ALOS AVNIR2 (3 band), 2007 and LANDSAT 7 ETM + (3 band), 2000

14 Changes of thermo-karst lakes on based of images difference Test area 2 LANDSAT 7 ETM + (4 band), 2000 and LANDSAT 4 TM (4 band), 1988 ALOS AVNIR2 (4 band), 2007 and LANDSAT 7 ETM + (4 band), 2000

15 Conclusion The technique of color composite RGB is most simple and fastest way for change detection. It allows detecting and differentiating changes connected with human activity and natural processes. The technique of image differencing with classification of changes allows to obtain more evident picture of changes, but it results depend on size of the chosen threshold, and different types of objects can be carried to one class, for example, the dried up and polluted lakes. Comparison of ALOS AVNIR2 and Landsat data has allow to detect in Varandey region the changes caused by human activity (mechanical disturbance of a surface, pollution of lakes and rivers and others) and changes caused by natural factors (changes in coastal line and thermo-karst lakes). Analysis of spectral curves of multitemporal ALOS AVNIR 2 and Landsat satellite images has allowed estimating level of lake pollution due to human activity and comparing it with level of suspended sediment concentration caused by natural processes. Irina SMIRNOVA, Alexandra RUSANOVA VNIIKAM St.-Petersburg, Russia vniikam@mail.wplus.net