|
Titel |
Monitoring changes in surface elevation of blanket peat and other land cover types using a novel InSAR processing technique |
VerfasserIn |
Barry Rawlins, Francesca Cigna, Colm Jordan, Andrew Sowter, Chris Evans, David Robinson, GMEP Team |
Konferenz |
EGU General Assembly 2014
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 16 (2014) |
Datensatznummer |
250088870
|
Publikation (Nr.) |
EGU/EGU2014-3047.pdf |
|
|
|
Zusammenfassung |
There is considerable interest in techniques which could detect changes in peat volume,
of which surface elevation is an important measure. We used new technique for
processing InSAR (satellite) to accurately measure changes in the elevation of soil and
vegetation across an area of north Wales (UK) which includes a large region of
blanket peat (10km2). We applied a novel technique for processing InSAR data;
Intermittent Small Baseline Subset (ISBAS; Sowter et al., 2013). This technique
considers pixels within the input radar stack which are only coherent for subsets of the
total time period for processing. We applied the ISBAS technique to satellite data
between 1993 and 2000 over a large region of interest (4460km2) that encompasses
several land use, soil and bedrock combinations† (Emmett B.E. and the GMEP
team, 2013). In two cases, the period between individual satellite acquisitions was
short (24 hours), whilst in general it was longer (35 days). The pixel size of each
elevation measurement after processing is a square area with side length 100 metres
(10000m2). We collated historical measurements of weekly rainfall and predictions of
potential evapotranspiration (each in mm) to identify major periods of wetting and
drying.
We used the short-interval scenes to estimate change in surface elevation over 24 hours
for different land cover types. In each case the mean change was close to zero (range
-0.03–0.02cm) whilst its standard deviation in areas dominated by peat (5.2–5.8cm) was
substantially larger by comparison to areas of improved grass, forest and heather
(1.7–3.2cm) and areas dominated by bedrock (0.5cm). In each case these distributions of
change were not strongly skewed (range of skewness coefficients -0.44–0.34), but they
generally had heavy upper and lower tails (range of kurtosis values 0.6–15; leptokurtic). We
focussed on data from the areas of blanket peat. We computed change in elevation between
the start and end of three of the driest and one of the wettest periods (between 105 and 175
days in duration). The mean change in elevation for these dry and wet periods was also
close to zero (-0.05–0.1cm) whilst their standard deviation (5.4–5.7cm) was
similar to those of the 24 hour periods (over peat). The similarity of changes in
surface elevation over these very different timescales (and moisture regimes) suggests
that the dominant processes controlling peat surface elevation occur over the short
timescale. There was no evidence to suggest that extended wet or dry periods led
to an overall change in surface elevation across the area of peat. We computed
variograms for the peat elevation data (relative to a fixed point) to explore the nature of
any autocorrelation, and cross-variograms to investigate joint spatial variation in
elevation over time (changes in elevation for each pixel). The variograms fitted to the
semi-variance estimates for change in peat elevation over the short periods (and the
wet/dry periods) did not exhibit spatial autocorrelation, indicating that the dominant
processes controlling surface elevation occur at scales shorter than the pixel resolution
( |
|
|
|
|
|