[EI
66
68
)
andscape structure
and Assessment,
ing and predicting
t Paul, Winnipeg,
f Environmental
/04/2010, from
CID/711/ItemID/8
/12/2011, from
ge processing: a
, Vol, pp. Pages,
‘cover change and
ie Bindura district,
Development, 18,
008. Urbanization
1al directorate of
using landsat data,
ment, 140, 1-14.
1d E Ene, 2002.
ysis Program for
oftware program
e University of
the following web
h/fragstats/fragsta
ication of remote
ver mapping and
western Nigeria,
al Science and
decision approach
ng Markov chain
jach, Canadian of
id use change and
l'aihu Lake Basin.
informatics 18th,
,
er Changes Using
orth-East Turkey,
Landsat data to
ng basin, China,
7.
1e role of land use
ital change, Land
78.
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B8, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
HIGH TEMPORAL FREQUENCY BIOPHYSICAL AND STRUCTURAL VEGETATION
INFORMATION FROM MULTIPLE REMOTE SENSING SENSORS CAN SUPPORT
MODELLING OF EVENT BASED HILLSLOPE EROSION IN QUEENSLAND
B. Schoettker * *, R. Searle °, M. Schmidt “, S. Phinn *
* The School of Geography, Planning and Environmental Management, The University of Queensland, 4072 St Lucia,
Queensland, Australia — b.schoettker@ug.edu.au
? Commonwealth Scientific and Industrial Research Organisation, Ecosciences Precinct, 4102 Dutton Park, Queensland,
Australia
* Queensland Department of Environment and Resource Management, Remote Sensing Centre
Environment and Resource Sciences Ecosciences Precinct, 4102 Dutton Park, Queensland, Australia
Commission VIII/8: Land
KEY WORDS: vegetation, dynamic, multisensor, erosion modelling, MODIS, terrestrial, management.
ABSTRACT:
This study demonstrates the potential applicability of high temporal frequency information on the biophysical condition of the
vegetation from a time series of the global Moderate Resolution Imaging Spectroradiometer (MODIS) Fraction of Photosynthetically
Active Radiation absorbed by vegetation (FPAR) from 2000 to 2006 (collection 4; 8-day composites in 1 km spatial resolution) to
improve modelling of soil loss in a tropical, semi-arid catchment in Queensland.
Combining the biophysical information from the MODIS FPAR with structural vegetation information from the Geoscience Laser
Altimeter System on the Ice, Cloud, and land Elevation Satellite (ICESat) for six vegetation structural categories identified from a
Landsat Thematic Mapper 5 (TM) and Enhanced Thematic Mapper 7 (ETM+) woody foliage projective cover product representing
floristically and structurally homogeneous areas, dynamic vegetative cover factor (vCf) estimates were calculated. The dynamic vCf
were determined in accordance with standard calculation methods used in erosion models worldwide. Time series of dynamic vCf
were integrated into a regionally improved version of the Universal Soil Loss Equation (USLE) to predict daily soil losses for the
study area. Resulting time series of daily soil loss predictions averaged over the study area coincided well with measures of total
suspended solids (TSS) (mg/l) at a gauge at the outlet of the catchment for three wet seasons (R? of 0.96 for a TSS-event). By
integrating the dynamic vCf into modified USLE, the strength of the dependence of daily soil loss predictions to the only other
dynamic factor in the equation - daily rainfall erosivity - was reduced.
1. INTRODUCTION
1.1 Motivation and aim
The relevance of the vegetative cover components to mitigate
soil loss effects by water and their potential to improve water
quality downstream is widely accepted and has been proven
valid over a range of ecosystems worldwide (Renard, Smith et
al. 1997; Vrieling 2006; de Asis and Omasa 2007).
High quality information on the biophysical and structural
properties of the total vegetation cover (TVC), optimally taken
at high temporal frequency, is thus indispensable to support
sustainable Natural Resource Management (NRM) of land and
water. This is particularly valid in complex and highly dynamic
savanna ecosystems, such as the tropical, semi-arid coastal
catchments of Queensland adjacent to the Great Barrier Reef
(GBR), where key challenges include declining water quality,
land degradation and soil erosion, and terrestrial discharges into
the lagoon (Hutchings and Hoegh-Guldberg 2008).
Remote sensing applications and broad-scale catchment
modelling offer invaluable potential to complement classical
field-based NRM in the assessment of temporal and spatial
aspects of soil erosion in the savanna ecosystems of these
tropical, semi-arid coastal catchments of Queensland (Searle
* Corresponding author.
and Ellis 2009). However, tropical savannas pose a particular
challenge to remote sensing applications due to abundant
senescent plant material being present at most times of the year
in a structurally complex and heterogeneous landscape (Asner
1998), which all influence the biophysical and spectral
properties of TVC at canopy and landscape (Asner and
Wessman 1997).
For the detection of non-photosynthetic vegetation (NPV) in
remote sensing applications the wavelength of
photosynthetically active radiation (PAR) (400-700 nm) has
also proven useful, since PAR is not always used for
photosynthesis (‘functional PAR’) (Asner 1998; Thomas, Finch
et al. 2006). A significant component of incident PAR can be
absorbed by NPV material in savanna ecosystems, particularly
in areas with a leaf area index (LAI) of less than 3.0; standing
grass litter canopies absorbed almost as much PAR as green
grass canopies (Asner 1998). How much PAR is absorbed at the
landscape scale is greatly affected by overstorey (trees) but the
relative differences in absorbed radiation are also affected by
the understorey (mostly grasses) LAI.
Global remotely sensed products provide free of charge, high
temporal estimates of biophysical properties that relate to
relevant ecosystem structure and function and provide estimates
of vegetation structure at different scales. Examples of these are
