D. C. White 3 ’ *, M. Williams 3 , S. L. Barr 3
School of Civil Engineering & Geosciences, Newcastle University, UK - davina_cherie@yahoo.com
KEY WORDS: Hyperspectral, Vegetation, Soil, Radiometry, Spectral, Environment, Analysis, Detection
ABSTRACT:
Previous studies conducted under controlled laboratory and controlled field conditions have demonstrated the ability of
hyperspectral techniques to detect vegetation stress associated with elevated levels of soil gas and associated soil oxygen depletion.
This paper investigates the capability and transferability of these hyperspectral techniques, in particular the Smith et al. (2004)
725:702 nm ratio, to identify vegetation stress features associated with sub-soil disturbance resulting from gas pipeline earthworks
under heterogeneous field conditions. Field spectroradiometry data of barley were acquired in 2005 and 2006 at selected transects
perpendicular to a stretch of buried gas pipeline in Aberdeenshire. Spectral reflectance and absorption features in the VIS-NIR are
evaluated through first derivate analysis to establish their position, shape and magnitude and used to determine narrow waveband
ratios which are tightly coupled to changes in photosynthetic function resulting from vegetation stress. First derivative ratios,
723:700 and 725:702 nm, detected vegetation stress above the gas pipeline where soil had been disturbed and were the same order of
magnitude for the 2005 and 2006 data. Ratio values were similar to those conducted under controlled conditions by Smith et al.
(2004), with differences of up to 56 % for spring barley between locations of soil disturbance and locations away from the pipeline,
R 2 accounting for up to 62 % of the variance in the ratios of the regression. Student’s T-tests revealed that the ratios were
statistically significantly different between stress and no known stress within zones of soil disturbance at the 0.05 confidence level.
1. INTRODUCTION
Remediation involves the reinstatement of the land, flora, and
other land use, where the sub-surface soil has been disturbed, as
close as possible to its original condition. Sub-surface soil
disturbance associated with gas pipeline earthworks can have an
impact on adjacent land potentially resulting in drainage
misalignment and collapse, sub-soil compaction (Pers. Comm.
Donal Cullen, Bruce Mackie and Gerald Banks), and mixing of
surface and subsurface horizons. These impacts can lead to
changes in organic matter, clay, pH, aeration, water retention
and nutrition (De Jong and Button, 1973; Rimmer, 1991).
Vegetation stress associated with sub-surface soil disturbance
has been observed as visual symptoms of chlorosis, stunted
growth and sparse vegetation cover. To date research evaluating
the health of vegetation overlying gas pipelines has been
confined primarily to controlled laboratory and controlled field
test sites, such as Smith et al. (2004), Noomen et al. (2006), and
van der Meijde et al. (2006). These studies have demonstrated
the capability of hyperspectral techniques (first derivative
analysis, band ratios and continuum removal) to detect
vegetation stress intimated to be a generic response to soil
oxygen depletion resulting from elevated levels of soil gas.
Smith et al. (2004) also suggested that their first derivative
725:702 nm ratio could be employed to detect other forms of
soil oxygen depletion, such as sub-soil compaction and
waterlogging, which can result from pipeline earthworks.
Hyperspectral remote sensing has the potential to provide a
solution for detection and monitoring of pipeline earthworks
associated sub-surface soil disturbance which can be inferred
from resultant surface vegetation stress features. A key
capability of hyperspectral data are their near contiguous
narrow spectral wavebands which lend to first derivative
analysis providing a more robust approach (much less affected
by extraneous variables, such as partial plant cover under stress
conditions) than broad band vegetation indices such as NDVI.
Hyperspectral remote sensing is also potentially capable of
providing more rigorous, reliable and repeatable detection and
monitoring than current visual inspections from aircraft
overflights. Visual inspections can be unreliable at identifying
stress due to human subjectivity, and are personnel and time
intensive (Hausamann et al., 2003; Zimig et al., 2002).
The aim of this paper is to determine the capability of
hyperspectral field spectroradiometry data to identify surface
vegetation stress features associated with sub-surface soil
disturbance resulting from gas pipeline earthworks under
heterogeneous field conditions. Spectral reflectance and
absorption features in the VIS-NIR are evaluated through first
derivate analysis to establish their position, shape and
magnitude. These first derivative VIS-NIR features are used to
identify which narrow waveband ratios are tightly coupled to
changes in photosynthetic function resulting from vegetation
stress, in particular the Smith et al. (2004) ratio.
2. FIELD METHODS AND DATA ANALYSIS
2.1 Study site
The study area consists of a 9 km stretch of 508 mm buried
Natural Gas Liquids (NGL) pipeline in Aberdeenshire, Scotland,
installed and remediated in 1983. A second 457 mm gas
pipeline is also present within the same corridor and was
installed in 1980 (Figure 1). Land use is predominantly arable
crops (barley, wheat and oats) with some grassland. The
Corresponding author.