synergies of sentinel-1a sar and sentinel-2a msi data for...
Post on 10-Oct-2020
1 Views
Preview:
TRANSCRIPT
Jan Haas (jhaas@abe.kth.se) and Yifang Ban, Division of Geoinformatics, Department of Urban Planning and Environment
KTH Royal Institute of Technology, Stockholm, Sweden
The objective of this study is to evaluate the potential use and synergetic effects of
novel ESA Sentinel-1A C-band SAR and Sentinel-2A MSI data for mapping of
ecologically important urban and peri-urban space at medium to high spatial
resolutions. The combined use of Sentinel-1A SAR in Interferometric Wide Swath
mode and simulated Sentinel-2A MSI data is being evaluated in classification of an
urban area over Zürich, Switzerland. Based on the outcome from fused, segmented
and SVM-classified data, landscape metrics and ecosystem services are used to
characterize ecosystem service provision and demand budgets, drawing spatial and
topological considerations of service providing landscape patches into account.
INTRODUCTION
Zürich is the largest city in Switzerland with an increasing urban population of over
400,000 inhabitants and about 1.8 million people residing in the metropolitan area.
The city centre lies about 400 m above sea-level in the temperate climate zone. The
living quality in Zürich is considered high with noticeable progress in terms of
environmental quality over the past two decades. Major urban classes are continuous
and discontinuous urban fabric, industrial/commercial areas, the infrastructural
road/railroad network including Zürich airport, construction sites, green urban
spaces, sports/leisure facilities and allotments. The urban hinterland is characterized
by Lake Zürich, agricultural land and forest.
Synergies of Sentinel-1A SAR and Sentinel-2A MSI Data for
Urban Ecosystem Mapping
METHODOLOGY
characteristics of natural green and blue land cover patches that are expected to alter ecosystem service provision capacities were derived under these following assumptions:
Figure 3. Classification result
• The combined use of Sentinel-1 SAR and simulated Sentinel-2 MSI data has
proven effective in urban land cover and ecosystem services mapping and
continued use of the data is endorsed for future ecological and urban applications.
• Spatial characteristics and topological aspects of landscape patches that are believed
to influence the provision of ecosystem services have been integrated to the
concept of ecosystem service supply and demand budgets.
RESULTS AND DISCUSSION
CONCLUSIONS
Range-Doppler
Terrain Correction
Sentinel-1 IW Level-1
GRD
VH and VV Intensity
2015-03-16
Simulated Sentinel-2
level 1c product (APEX)
RGB, NIR and SWIR
2011-06-26
5x5 adaptive Lee
Speckle filtering
Coregistration, RMSE X/Y
<0.5 pixels, CH1903
Resampling (8-bit,
2% tail-trim for S1, 10m2)
Canny Edge Detection
Accuracy Assessment
Continuity
Area
Perimeter-to-area
Distance
Ecosystem Service Budgets
(accounting for space/topology)
Final
classification
14 classes
Ecosystem Service Budgets
(Burkhard et al., 2012)
Segmentation with KTH-SEG
Object-based SVM Classification
Class aggregation and post-
classification
STUDY AREA AND DATA
Figure 2. Methodology Flowchart
Figure 1. RGB true-colour-composite of a Sentinel-2A APEX scene from June 2011 (left) and
Sentinel-1A 5x5 adaptive Lee speckle filtered intensity data from March 2015 (VV-VH-VV) (right)
For this study, a 16x22 km2 Sentinel-2A scene, simulated with high resolution
airborne imaging spectrometer (APEX) data and including all spatial and spectral
characteristics corresponding to a Sentinel-2 level 1c product dating from 2011-06-
26 was used in combination with Sentinel-1A C-band SAR IW mode data as a
Level-1 GRD product from 2015-03-16.
This research is supported by a grant from FORMAS. The research is also part of
the project “Satellite Monitoring of Urbanization for Sustainable Urban
Development” within the European Space Agency (ESA) and the Chinese
Ministry of Science and Technology (MOST) Dragon III program.
ACKNOWLEDGEMENT
The classification into twelve urban and peri-urban classes resulted in an overall
accuracy of 79.81% with a Kappa coefficient of 0.78 indicating the suitability of
the chosen classification method and underlying data. Largest confusions occurred
between construction sites and industrial/commercial areas and between the built-
up classes discontinuous urban fabric, sport and leisure facilities and between
airport runways and roads. The classified image was sieve-filtered and agriculture
and mixed forest classes were transformed into urban green spaces and urban
forests, respectively under an urban mask. From the final classification, four spatial
• Distance: Increased ecosystem service provision of patches close to urban dwellers.
• Perimeter-to-area: The lower the ratio, the higher the service provision because less
edge is shared with other classes and patch centres are more pristine.
• Area: Larger patches are capable of providing more services.
• Contiguity: High patch connectivity and less fragmented landscapes are considered
beneficent in several ways, e.g. for species dispersal or recreational purposes.
Figure 4. Spatial patch characteristics influencing ecosystem service provision
Figure 5. Ecosystem service budgets
Ecosystem service budgets were then
generated by modifying the supply
values initially presented in Burkhard et
al. (2012).
REFERENCE
Burkhard, B., Kroll, F., Nedkov, S. & Müller, F. (2012). Mapping ecosystem service
supply, demand and budgets. Ecological Indicators 21:17–29.
top related