Kite Aerial Photography Techniques for Vegetation Monitoring in Rural Senegal

1
Kite Aerial Photography Techniques for Vegetation
Monitoring in Rural Senegal UC Berkeley GIF / MVZ
Geolunch Series February 6th, 2009
Jeremy T. Freund Geospatial Innovation
Facility College of Natural Resources UC
Berkeley jfreund_at_nature.berkeley.edu
2
Todays Outline

• Project Description
• Carbon Offset Credit Market
• Why KAP? Why Senegal?
• The Process / GIF Role, Involvement
• 1. Image acquisition ?
• 2. Image Referencing Stitching 3. Processing
  (Allometry, Biomass estimation techniques) 4.
  Validation techniques
• KAP role in Remote Sensing
• Monitoring change vs. absolute measure
• New Validation Paradigm? Future Prospects
• Pictures!

3
Project Description
Provide access to carbon and development credit
finance for community-based sustainable
development projects

• Battle Climate Change, Poverty, Global Health
• Combined carbon, health, poverty credits an
  ultimate goal
• New Model Focused on Efficiency, Cost
  Effectiveness
• Provide funds directly to rural villages based on
  their individual environmental rehabilitation
  programs
• Project focused on verification

4
Project Description
Why Senegal?

• Social factors
• UNDP HDI 156th / 177 Countries with data
  avail1., placing Senegal in Low Human
  Development category. Prime candidate for
  development assistance.
• Local communities manage own resources, recognize
  value and potential of land resources.
• Good understanding of impact on Global
  environment
• Established network of NGOs in place to carry out
  village-level tasks
• Physical factors
• Millions of Mangroves in tidal pools ideal for
  KAP
• Flat, windy (coastal), unobstructed, vast areas
• Huge potential for carbon sequestration, global
  environmental impact.

1http//hdrstats.undp.org/countries/data_sheets/ct
y_ds_SEN.html
5
The Process (as a whole)
KAP for village-level verification

• Kap for Baseline Biomass Estimation
• Village Biomass database
• Techniques include KAP image acquisition,
  automated image stitching, image processing and
  documentation
• KAP for biomass change detection
• Process repeated at specified intervals (2, 5, 10
  yrs)
• Credits based on verifiable biomass increase
• KAP for Verification
• Multiple applications in remote sensing

6
The Process
GIFs Role

• Image Acquisition (GIF)
• Use KAP techniques to acquire multiple
  overlapping very high resolution imagery.
• Common kite, digital camera can be used
• Sutton Flowform 60 Kite (60 ft2, good for low
  wind or heavy payloads1) 320
• Nikon CoolPix 6000 13 MegaPixel Digital Point and
  Shoot with integrated GPS 450

• Camera must shoot automatically, or use a servo
  device to manually push the button
  periodically.
• Basic Kite / Image Parameters
• 500ft of string (ideal for conditions)
• 30- 45 string angle
• 28mm focal length
• FOV _at_ 45 124m 87m, 108m alt.
• FOV _at_ 30 87m 62m, 76m alt.

diagram courtesy of Nathalie Rivest, VPI Intern
1http//arch.ced.berkeley.edu/kap/equip/supplies2.
html
7
Photoset Example
Peth
8
Photoset Example
Mbelane
9
Photoset Example
Sakhor
10
The Process (verification)
Stitching

• Image Stitching and Referencing
• GIF working on automatic stitching method /
  software
• Current ideas
• Photo stitching software (autopano pro)
• Manual mosaicking (Picassa)
• Geo-referencing performed with visible flags

11
Stitching Examples
Nguelack Reforestation
12
Image Processing

• Segmentation / Classification (OBIA) Allometry
• Image segmentation
• Berkeley ImgSeg (or other OBIA solution) to
  obtain individual crown diameters, relate to
  Biomass
• Open source / customizable
• Developed at UC Berkeley
• Working with developers to create customized
  species-specific segmentation / extraction scheme

• Rhizophora Mangle (Above ground Biomass)
• ln(biomass) 4.89219C2.61724 ln (DBH)1
• DBH Diameter at Breast Height (can be
  determined from crown diameter)
• Approximated by max crown circumference

1Soares Schaeffer-Novelli, 2005 Estuarine
Coastal and Shelf Science
13
Validation

• Obtain samples of at least 20 trees location that
  was imaged
• Trunk diameter at breast height (D.B.H.)
• GPS locations
• Compare to exact trees in image!

1Soares Schaeffer-Novelli, 2005 Estuarine
Coastal and Shelf Science
14
Kite Launch!
15
Public Image Gallery
All 13 Megapixel Images Available (4224 3168)
http//picasaweb.google.com/jeremy.freund
The images in this public gallery may be
downloaded and used for research (academic)
purposes only. The Author must be cited in any
refereed or published work which utilize these
images in any form.
16
Full Resolution Image Example

• Individual leaves / saplings are discernable
• Early growth monitoring
• LAI, other green biomass related measurement
  possible

17
The Reality!
18
The Reality!
19
Acknowledgements

• This project is made possible by
• Village Projects International, a non-profit
  headed by Robert Van Buskirk
• Special thanks to the following partners
• West African Association for Marine Environment
  (WAAME)
• Oceanium Dakar
• United Nations Development Programme
• Support and hospitality
• The Lebot family, Michel Perlo, student and
  ecologist
• Ahmed Diame (WAAME) for amazing, enduring
  assistance through some fairly trying and
  difficult conditions