In: Wagner W., Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B
In: Wagr
68
LAND USE CLASSIFICATION WITH HIGH-RESOLUTION
SATELLITE RADAR FOR ESTIMATING THE IMPACTS OF LAND
USE CHANGE ON THE QUALITY OF ECOSYSTEM SERVICES
D. Bargiel ab ’ *, S. Herrmann 3 , P.Lohmann b , U. Sörgel b
a Gottfried Wilhelm Leibniz Universität Hannover, Institute of Environmental Planning, Herrenhäuserstr. 2, D-30419,
Hannover, Germany- (bargiel, herrmann)@umwelt.uni-hannover.de
b Gottfried Wilhelm Leibniz Universität Hannover, Institute of Photogrammetry
and Geoinformation, Nienburger Str.l, D-30167, Hannover, Germany - (lohmann, soergel)@ipi.uni-hannover.de
KEY WORDS: Ecosystem, Land Use, Classification, Radar, Multi-temporal
ABSTRACT:
Ecosystems provide several services for human well-being. The quality of these ecosystem services is among others affected by
agriculture, the main land user in Europe. Thus, it is essential to provide information about land use in agricultural areas.
In this research, land use classification of agricultural areas is carried out based on high-resolution Spotlight TerraSAR-X images
(TSX-images) of two different polarisations (HH and VV). A stack of several radar images taken during the vegetation season is
used for multi-temporal classification of land cover. The typical phenology of agricultural vegetation types and their individual
phenological development during the year cause differences in the backscatter of the radar signal over time.
Two different study areas are investigated, one in the North East of the city of Hannover, Germany called “Fuhrberger Feld” and one
in the “Gorajec area” in the very South East of Poland. These two areas represent extremely diverse European regions with regard to
agro-technological level, population density, cultivation form as well as geological and geomorphological conditions. Thereby, the
radar signal backscatter for different regions is tested.
Preliminary results show significant differences in the backscatter of crop types in SAR data of about 3 m, especially for grasslands,
grain and broad-leaved crops. Furthermore the VV polarised radar signal has clearly lower backscattering for grains during
summertime and for grasslands in general than for broad-leaved crops.
1. INTRODUCTION
Ecosystems have an important function for the quality of human
life. They provide material goods and intangible values as
“ecosystem services” for human well-being. They comprise all
basic requirements for human well-being, e.g. food, water, air,
climate, or recreation. Ecosystem services base on a complex
system of ecosystems and their interactions. They enable
security, health, basic material goods, and good social
relationships (Millennium Ecosystem Assessment, 2005;
Myers & Reichert, 1997).
Ecosystem services are affected by different direct and indirect
drivers of change. Beside natural drivers, also human impacts
influence the ecosystem services; one important impact is land
use (Millennium Ecosystem Assessment, 2005). Although land
use and ecosystem services interacted since the beginning of
land cultivation, there has been an unprecedented increase of
intensity in land use in the twentieth century (Poh Sze Choo et
al., 2005; Ramankutty et al., 2006). This leads to numerous
negative impacts on ecosystems services (Poh Sze Choo et al.,
2005; Ramankutty et al., 2006; DeFries, 2004). To assess the
impacts of land-use changes on ecosystem services, it is
indispensable to provide precise and up-to-date information
about land use and land-use change. Remote sensing affords the
opportunity to derive this information. New high-resolution
sensor types are particularly suitable to improve land use
classification results (Poh Sze Choo et al., 2005;
Ramankutty et al., 2006). One of the new sensors is the
TerraSAR-X satellite based radar sensor.
TerraSAR-X allows acquisition of multiple polarized radar
images (products) with a high ground resolution of up to one
meter (DLR, 2007; Fritz & Eineder, 2009). As a satellite-based
radar system it is able to provide reliable and regular
information about earth surface. Hence, it is especially suitable
for multi-temporal land-use classification. The basic idea of the
multi-temporal land-use classification is to use a stack of
several products during the vegetation period. Different
phenological conditions of the vegetation cause an individual
backscatter of radar signal in time. In this way, a higher content
of information for the classification method is available. Studies
with elder systems like ERS 2 or ENVISAT-ASAR showed
already the general feasibility of this approach (Schieche et al.,
1999; Foody et al., 1988; Borgeaud et al., 1995; Tavakkoli
Sabour et al., 2008).
The objective of this study is to conduct a multi-temporal land-
use classification of TerraSAR-X images and to examine the
suitability of the classification results for assessing impacts of
land-use change on ecosystem services. Individual temporal
backscatter patterns are identified for different crop types.
Observations concentrate on European agricultural areas and on
two selected ecosystem services, namely biodiversity and soil.
2.1 Stu<
Two stu<
transferal
their soc:
conditior
Germany
9.84 E), ’
South Ei
area is
productic
share of
structure:
ground v
Hannove
In contri
methods
standard
regions
methods
structure
size of 1
relief an(
erosion p
2.2 Dat:
For both
Resolutio
the year
polarisati
products
range dii
available
region be
region in
area all a
those for
Ground i
meters fc
meters fo
in azimut
HS-Mode
* Corresponding author.