PowerPointPrsentation

1
Institut für Softwaresystemein Wirtschaft,
Umwelt und Verwaltung
ISS
Multisource Data Fusion Supporting High Quality
Mapping of Sealed Areas
Fischer, P. Wannemacher, S.
University of Applied Sciences Trier Umwelt-Campus
Birkenfeld Informatics Department P.O.B. 1380, D
55761 Birkenfeld p.fischer_at_umwelt-campus.de
2
Motivation

• Dimension of new built-up areas in Germany
• 129 ha/day (1997 2000)
• 117 ha/day (2001)
• National plan reduction down to 30 ha/day (2020)
• actually 12,4 of the landscape in Germany is
  dedicated to settlements and traffic
• New built-up areas in Rhenania-Palatinate
• 54.000 m2/day (Statistical Survey Bad Ems, 2002)
• But only 40-60 of the areas mentioned above are
  definitely sealed soils (Bodenschutzprogramm
  Bayern 1991 Statistische Berichte 2002)
• ? There is a lack in valid information regarding
  quality and quantity of the sealed soils

3
Relevance of sealed area information

• Sealed areas
• are regions where the natural soil is covered
  by buildings, roads, concrete, etc. so there is
  no water infiltration possible anymore
• ... deliver base data for the organisation of an
  intelligent occupation of the landscape / soils,
  e.g.
• Questions regarding tolerable rise in settlement
  density ?
• parameter for the management of urbanized regions
  regarding further develoment
• ...
• ... deliver base information to environmental
  modeling, regarding
• Flood risk management
• Acquifer regeneration
• Urbanisation processes
• Protection of rare soil types
• ....

4
Objectives of the work

• Development of a method to quantify the sealed
  areas in the state of Rhenania Palatinate (test
  case River Rhine Valley)
• Satisfaction of user requirements
• Repeatable analysis (High data availability over
  space and time)
• Scale of the resulting thematic map about
  125.000
• automated analysis should be possible
• Low (better no) costs for data acquisition
  focus on use of GeoData available in different
  archives of the federal government

5
Region of Interest
Source CORINE Landuse
Source TERRA Aster
6
Methods to quantify sealed areas

• Terrestrial methods
• Land survey (z.B. landscape planning)
• Interviews (z.B. waste water management)
• EO-based methods
• Orthophotos (on-screen-digitizing)
• CIR-aerial photos
• airborne Scanners (e.g. HRSC)
• spaceborne Sensors

7
Potential of satellite Earth Observation Data

• Advantages
• Low cost data factor up to 4 with conventional
  methods.
• digital data, automated processing capabilities
• 3D information,
• frequent update of large regions, independent
  from administrative borders,
• objectivity and homogeneity of the data
• Disadvantages
• Often lack in geographical precision
• Very high resolution data with limited capacity
  in automated operational analysis
• Mixed signals
• Shadowing effects
• ....

8
Civil Satellite-Missions (01/2004)
recently 109 civil EO-missions available
(historical and recent)
9
Key parameter in EO-data processing Vegetation
Index

• Assumption vegetated areas are in general
  complementary to sealed areas
• Calculation of the vegetation index via ratioing
  of red/infrared band

10
Potential of GeoData 1 (landsurvey information)

• ATKIS (Authoritative Topographic-Cartographic
  Information System) is developed by the surveying
  authorities of the states of the federal republic
  of Germany (AdV).
• Components are object-based digital landscape
  models (DLM) and digital topographic maps with a
  high positional- (horizontal /- 3 m, vertical
  /- 0,5m) and semantic accuracy.
• The coverage is nationwide with an update cycle
  of 5 years.

IKONOS-PAN (1m resolution)
Overlayed with ATKIS landuse
11
ATKIS Object-Model

• Usage of ATKIS Objekt-model to define areas
  where we expect sealed soils (sealing masks)

12
Potential of GeoData 2 (cadastral information)

• Availability of a digital cadastral information
• system (ALK)
• High geometric accuracy ( lt 5 cm)
• But...
• Database not up-to-date
• Partially incomplete (e.g. missing of newly
  created buildings)
• Lack in relevant information (e.g. sealed parking
  lots) because of a reduction of the object-types
  to be monitored (reduction of the OBAK-LiKa-RP)
• ? Sealing information of the ALK ltlt Reality

13
Fusion of EO- GeoData

• ... will reduce misclassifications of EO-Data
• ... is able to update the GeoData available

14
Multisource Sealed Area Detection
ATKIS Mask (SUV)
Overlay
ALK Mask (sealed)
Resulting areas with need in EO-Data analysis
Traffic net (sealed)

• EO data analysis limited to ATKIS masks
  (encircling sealing activities) with exeption of
  the available building vectors (cadastral
  information) and the traffic net (accurate
  sealing info available)
• Maximum use of sealed information available in
  GeoDatabase
• Reduction of possible misclassifications in EO
  data processing (sensitive feature extraction)

15
Processing Chain
16
Ex. NDVI combined with Settlement Mask

• SPOT-5 NDVI

• NDVI and ATKIS settlement mask

17
Classified NDVI with ALK Buildings
18
Results City of Mainz, Kaiserstrasse
19
Comparison of different information sources

• ATKIS generalized geometry but semantic
  information about landuse available no
  information about the degree of sealing
• Multisource approach more detailed geometry,
  sealing and ATKIS feature information available
• Field Mapping generalized geometry with averaged
  sealing related info

Multisource
Field Mapping
ATKIS - GeoData
20
Results Visualisation and Environmental Atlas

• Visualisation and delivery of the sealed data on
  a pixel base or aggregated to administrative
  units
• http//www.umweltatlas-rlp.de

21
Conclusions and Outlook

• Conclusions
• Good reproductability of the results
• High spatial resolution
• High data availability in federal administration
• Low percentage of misclassifications ( 5
  compared with ground trouth data)
• ... but difficulties to get high quality high
  resolution EO-data on demand
• Processing chain not fully automised
• Future work
• Development of a method to replace Satellite data
  by Orthophotos
• Improvement of automatised processing