BIO_SOS - GA N. 263435
Biodiversity Multi-Source Monitoring System:
From Space To Species.
www.biosos.eu
On behalf of BIO_SOS consortium
Palma Blonda° and Richard Lucas*
°CNR-ISSIA, Bari-Italy
blonda@ba.issia.cnr.it
* Aberystwyth University, UK
3rd Space call FP7-SPACE-2010-1
(Dic. 2010-Nov. 2013)
BIO_SOS working objectives
 The development of pre-operational automatic HR and mainly
VHR EO data processing and understanding techniques for:
 LCLU and LCLU change maps production as an improvement
of GMES core services.
 The development of an ecological modelling framework at both
habitat and landscape level to combine EO and in-situ data for:
 Habitat and Habitat change mapping (GHC and Annex I)
 Biodiversity indicator extraction.
 Scenario analysis
as an extension of GMES downstream–services
BIO_SOS peculiarities
 The system adopts deductive learning schemes, i.e. it based
on expert knowledge elicitation to fill the gap between domains
 Ontologies and semantic networks are used for
 LCLU and Habitat class description;
 LCLU to Habitats translation (domain ontologies)
 The Food and Agriculture Organization Land Cover Classification
System (FAO_LCCS) is adopted for LCLU classification since it
provides the best description of natural vegetated classes compared
to other taxonomies (e.g. CORINE) (Tomaselli et al. 2013).
V. Tomaselli, P. Dimopoulos • C. Marangi et. al., (2013). Translating land cover/land use
classifications to habitat taxonomies for landscape monitoring: a Mediterranean
assessment. Landscape Ecology DOI 10.1007/s10980-013-9863-3. Published on
March 2013.
BIO_SOS peculiarities
 The proposed architecture is multi-modular. The modules have been
developed independently to reduce the risk of a major project
breakdown
 Based on expert knowledge elicitation, validated LC/LU maps can
be integrated with in-situ data to provide:
 General Habitat Category (GHCs) and Annex I maps
 LCLU and habitat change maps
 Pre-existing LCLU maps produced by photo-interpretation and infield campaigns (e.g. LCLU and in-situ data) can be exploited mainly
in Mediterranean regions were habitat maps are often not available
The EODHaM System: Overview
www.biosos.wur.nl
A12/A2.A5.E7+O3.M233.N3.N12-AR
Habitat: 1210
EDGES:
Non - spatial relationships:
Class (IS A) Natural and semi-natural terrestrial vegetation (A12)
Class Core:
• T1 April- September: herbaceous forbs (PART-OF) (height in [0,01–0,3]m)
• T2 October-March: sand, washed by waves during storms (PART OF)
Class Context:
• Sandy
T1 AprilSeptember: sand (PART-OF)
shore
• T2 October-March: Context coincides with Core
Spatial topological relations:
Adjacent:
• Towards the sea: to sandy shore
• On the other side (towards the inland part): A12/A2.A6.E6 (habitat 2110)
Temporal relations:
 Class phenology: annual
Geometric attributes:
 Shape: linear
 Cover: open (40-(20-10)%)
June
Combining LCCS Codes
www.biosos.wur.nl
Habitat discrimination within EODHam
LCCS, Annex I and GHC habitat map:
Le Cesine site (IT)
GHC MAP:
Entropy texture for
height information
GHC with LIDAR
www.biosos.wur.nl
Le Cesine
Classification of Annex 1 Habitats
North Pembrokeshire Commons, Wales
LCCS Classification: Veluwe
Translation to GHC: Veluwe
WP 6: EODHaM modelling modules
development
Aim:
Task 6.4
Ecological Niche
Modelling (ENM)
To investigate the relationship between the
GHCs, landscape features, pressures, and
the abundance and composition of some
taxa, both animals and plants
D6.6 “Selected bioindicators”
D6.7 “ENM”
Aim:
Verify if GHCs can provide an useful link
between EO data and the species distributions
(species of conservation interest or representing
biodiversity threats)
WP 6: EODHaM modelling modules
development
Ecosystem state assessment (Task6.3)
 Software generation (in R) to simulate spatially distributed
water connectivity:
 to relate slope gradient to upslope basin area
 to deal with roads as connectivity modificators
Impact of ecosystem fragmentation – metapopulation
Methodology to identify locally recognizable (Task
6.8) pressures and quantify their impact on habitats
extract pressure growth
www.biosos.wur.nl
BIO_SOS achievements
 EODHaM System Designed and Implemented
 1st stage: spectral rules
 2nd stage: context-sensitive features and spectral rules
 3rd stage: translation of LCLU to GHCs and Annex 1
 Potential to include expert knowledge, ancillary data
 Open source software with translation out of commercial software
(Python)
 Scalable: UAS, WV2, Landsat, MODIS
 LCCS/GHC maps generated for primary BIO_SOS sites
 Capacity for change detection:
 Change in class
 Change in indices (pixel/object-level).
 New accuracy assessment tools, approaches and data
BIO_SOS contribution to GEO Task BI-01
 Long-term baseline data (e.g., thematic maps at 1:5000 scale or
finer) of land covers and habitats as well as new automatic,
standardized, rapid and cost-effective monitoring techniques.
 A modelling framework for evaluating the combined impact that
different drivers affecting soils and/or vegetation may have on
biodiversity.
 BIO_SOS products will be more familiar to the End Users since
they are built on their expertise and can be improved as they
further engage with the process
BIO_SOS contribution to GEO Task BI-01
 Continuity with EBONE and links with Ms.Monina for biodiversity
monitoring at regional level: VHR GHC and Annex I.
 FAO-LCCS taxonomy is used as brokering tools to combine
different domains (e.g. LCLU and habitats)
 The output products will be made available for policy decision
making; to follow up impact of existing policies and meet the
commitments related to European Directive obligations.
Conclusions and lessons learnt
 Multi-spectral/temporal EO data are more effective and
less costly than available single-date hyper-spectral data.
 WorldView2 data and LIDAR seems very useful for habitat
monitoring.
 More research is still needed for biodiversity indicator
extraction from VHR imagery
 Protocols for in-situ data collection should be:
 harmonized based on ecological modeling (e.g. niche modeling)
 agreed with Management authorities who institutionally collect data