|
Titel |
Pseudo-spectral methodology for a quantitative assessment of the cover of in-stream vegetation in small streams |
VerfasserIn |
Yaron Hershkovitz, Yaakov Anker, Eyal Ben-Dor, Guy Schwartz, Avital Gasith |
Konferenz |
EGU General Assembly 2010
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 12 (2010) |
Datensatznummer |
250042407
|
|
|
|
Zusammenfassung |
In-stream vegetation is a key ecosystem component in many fluvial ecosystems, having
cascading effects on stream conditions and biotic structure. Traditionally, ground-level
surveys (e.g. grid and transect analyses) are commonly used for estimating cover of aquatic
macrophytes. Nonetheless, this methodological approach is highly time consuming and
usually yields information which is practically limited to habitat and sub-reach
scales.
In contrast, remote-sensing techniques (e.g. satellite imagery and airborne photography),
enable collection of large datasets over section, stream and basin scales, in relatively short
time and reasonable cost. However, the commonly used spatial high resolution
(1m) is often inadequate for examining aquatic vegetation on habitat or sub-reach
scales.
We examined the utility of a pseudo-spectral methodology, using RGB digital photography
for estimating the cover of in-stream vegetation in a small Mediterranean-climate stream. We
compared this methodology with that obtained by traditional ground-level grid methodology
and with an airborne hyper-spectral remote sensing survey (AISA-ES).
The study was conducted along a 2 km section of an intermittent stream (Taninim
stream, Israel). When studied, the stream was dominated by patches of watercress
(Nasturtium officinale) and mats of filamentous algae (Cladophora glomerata).
The extent of vegetation cover at the habitat and section scales (100 and 104 m,
respectively) were estimated by the pseudo-spectral methodology, using an airborne Roli
camera with a Phase-One P 45 (39 MP) CCD image acquisition unit. The swaths
were taken in elevation of about 460 m having a spatial resolution of about 4 cm
(NADIR). For measuring vegetation cover at the section scale (104 m) we also used
a ’push-broom’ AISA-ES hyper-spectral swath having a sensor configuration of
182 bands (350-2500 nm) at elevation of ca. 1,200 m (i.e. spatial resolution of ca.
1 m). Simultaneously, with every swath we used an Analytical Spectral Device
(ASD) to measure hyper-spectral signatures (2150 bands configuration; 350-2500
nm) of selected ground-level targets (located by GPS) of soil, water; vegetation
(common reed, watercress, filamentous algae) and standard EVA foam colored
sheets (red, green, blue, black and white). Processing and analysis of the data were
performed over an ITT ENVI platform. The hyper-spectral image underwent radiometric
calibration according to the flight and sensor calibration parameters on CALIGEO
platform and the raw DN scale was converted into radiance scale. Ground level
visual survey of vegetation cover and height was applied at the habitat scale (100 m)
by placing a 1m2 netted grids (10x10cm cells) along ‘bank-to-bank’ transect (in
triplicates).
Estimates of plant cover obtained by the pseudo-spectral methodology at the habitat scale
were 35-61% for the watercress, 0.4-25% for the filamentous algae and 27-51% for plant-free
patches. The respective estimates by ground level visual survey were 26-50, 14-43% and
36-50%. The pseudo-spectral methodology also yielded estimates for the section scale (104
m) of ca. 39% for the watercress, ca. 32% for the filamentous algae and 6% for plant-free
patches. The respective estimates obtained by hyper-spectral swath were 38, 26 and
8%.
Validation against ground-level measurements proved that pseudo-spectral methodology
gives reasonably good estimates of in-stream plant cover. Therefore, this methodology can
serve as a substitute for ground level estimates at small stream scales and for the low
resolution hyper-spectral methodology at larger scales. |
|
|
|
|
|