landscape scale patterns of canopy gaps in the old growth...
TRANSCRIPT
Landscape‐scale patterns of canopy gaps in the old growth
forest of Lom,Bosnia‐Herzegovina
Matteo GarbarinoEnrico Borgogno MondinoEmanuele LinguaTom NagelVojislav DukicZoran GovedarRenzo Motta
Torino
PadovaLjubljana
Banja Luka
Canopy gaps spatial pattern
Ecological importance of canopy gaps spatial pattern and distribution:‐they drive the gap‐phase regeneration of the canopy
‐they influence stand structure and biomass
‐they influence regeneration composition & dynamics
INTRODUCTION
Canopy gaps at landscape scaleINTRODUCTION
• Gaps at landscape scale are poorly studiedMany studies are based on field data collection (time consuming)
• Fine resolution remote sensing data are now available to perform this kind of studies
‐ Satellite images (VHR < 1m resolution)
‐ LIDAR data (0.5m resolution)
Objectives
• Evaluating the potential of fine spatial resolution data to detect gaps at landscape scale
• Analyzing the spatial patterns of canopy gapsin an old‐growth forest
• Assessing the role of gap geometry in conditioning the regeneration composition
INTRODUCTION
LOM forest reserve
LOMLOM
JanjJanj
PeruciPeruciççaa
CORE AREA
Established in 1956: Klecovaca Mountains North Western Bosnia‐Herzegovina (44.482 N & 16.427 E)Reserve: 300 ha ; Core Area: 50 ha; Buffer Zone: 250 haElevation:1250‐1522 m a.s.l.; Soils: brown; Vegetation: Piceo‐Abieti‐Fagetum illyricum (Maunaga, 2001)
BUFFER ZONE
METHODS
Image analysis: orthoprojection
KOMPSAT‐2
Korea Multi‐Purpose SATellite‐2Resolution: 1m Pan; 4m MSAcquisition date: 11/06/2009Type: Bundle 2A (UTM/WGS84)
ASTGTM
ASTER Digital Elevation ModelResolution: 30m
NASA and METI (Slater, 2009)Absolute accuracy: 10m
15 Ground Control PointsCollected with a GeoXM
GPS
GPS + +
ORTHOPROJECTION ORTHOPROJECTION
Toutin rigorous model for Kompsat‐2 data
using PCI Geomatica 10.2
METHODS
Spatial distribution of planimetric error of 15 Ground Control Points
collected with a GeoXM GPS
GCPs Accuracy: RMSE: 1.92 m
Planimetric precision: 1.00 – 3.20 m
ORTHOIMAGE Accuracy:RMSE x: 1.15 mRMSE y: 0.69 m
RMSE Tot.: 1.35 m
Image analysis: orthoprojectionMETHODS
GCP Residuals (m)
Image analysis: ClassificationMETHODS
• Image enhancement: atmospheric correction
(radiance to apparent reflectance)
• Unsupervised pixel based classification based on Neural network trained by the photointerpretation of18 large (> 700 m²) gaps
• Land Cover Classes: Forests; Bare soils; Fields; Gaps; Soil‐meadow mosaic
The classification was performed with ENVI EX software
Landscape analysisMETHODS
Canopy gaps map derived as vector data from LC map adopting 32 m² as Minimum Mapping Unit
‐ Landscape metrics (size and shape of the gap)
‐ Gap size distribution
‐ Spatial pattern analysis (Ripley’s K) multi‐distance spatial cluster analysis in ArcGis 9
RESERVERESERVE
BUFFER ZONEBUFFER ZONE CORE AREACORE AREA
Stand scale: field surveys
Data collected on the field: ‐ Gap fillers & Adjacent trees
‐ Seedlings & Saplings
‐ GPS position of the centroid
‐ CWD volume (gap makers)
Direct ordination analysis (RDA)Exploring correlations between regeneration composition and
gaps geometric characteristics A
DJA
CEN
T
GAP FILLERS
METHODS
Orthoimage
Land Cover map
RESULTS
Canopy Gaps MapRESULTS
Landscape scale
• Landscape metrics (gaps as patches)
METRICS Unit CORE BUFFER RESERVETotal Area ha 59.14 240.36 299.50Gap Number n 102 548 650Gap Density n/ha 1.72 2.28 2.17Gap size mean m² 62.59 81.16 78.24Gap size max m² 320 1776 1776Gap size SD m² 50.00 106.72 100.17Gap Fraction % 1.08 1.85 1.70
RESULTS
Landscape scaleRESULTS
LARGEST GAPS
Gaps spatial pattern (Ripley’s K)RESULTS
RESERVEClustered in the Reserve
Clustered in the buffer zone
BUFFER ZONECORE AREA
Random within the core area
Stand scale• Regeneration composition in relation to gap size and shape (Redundancy analysis)
RESULTS
RDA - I
Explained variability % 12
Correlation 1st axis 73.3
P‐ value (Monte Carlo test) 0.004
regeneration compositiongap characteristics
1 = seedlings (H < 1m)2 = saplings (DBH < 7.5cm)
Light demanding species positively correlated to large gaps
Beech saplings slightly associated to gap fillers
The geometrical correction of the Kompsat‐2 image allowed to reach a good (1.35m) RMS error.
This landscape approach proved to be sound in detecting canopy gap > 32 m². Smaller gaps thatproved to be important for this kind of forestsmust be studied through a field survey (local‐scale).
DISCUSSIONS
Image processing
DISCUSSIONS
Core area VS Buffer zone
The Core area of the reserve was dominated bysmall gaps that were randomly spatiallydistributed.
The Buffer zone, more disturbed by humanactivities (harvesting & grazing), was dominatedby larger and more clustered gaps.
The influence of geometric characteristics ofgaps on regeneration composition emerged as animportant factor
Light demanding species (maple, and rowan) are more common in large gaps
Future research: compare Lom with other Balcanold‐growth forests
DISCUSSIONS
Regeneration
Aknowledgments
• All the people who helped us in the field data collection: Fabio Meloni, Roberta Berretti, Miroslav Svoboda, Tihomir Rugani, Dejan Firm, Alessandra Bottero, Daniele Castagneri, Beppe Dolce
• Planet Action project for providing the satellite images http://www.planet‐action.org/