The potential future of Remote Sensing in Hawaii

							Very High-resolution
Imagery for Remote
 Sensing in Hawai`i
         Stephen Ambagis1
            Jim Jacobi2
  1Hawai`i   Cooperative Studies Unit, UH Hilo
              2U.S. Geological Survey
  What happens in a vacuum
• Presently there are no local commercial aerial digital
  imagery providers
• Two systems are breaking onto the scene
• Both are focused on research and conservation
• No overlap in base products, complimentary systems
• Both systems are producing new products

This presents the user community with a very unique
  situation, one that could progress Hawaii’s
  conservation goals more than ever imagined
Carnegie Airborne
    Observatory
– Hyperspectral imagery,
  350 bands, 30 cm
– Full waveform LIDAR

Contact: Greg Asner
email gpa@stanford.edu
website- http://cao.stanford.edu/
 Resource Mapping
   – Very high resolution
     multispecral, 4 bands,
     20 cm resolution
   – Ultra high resolution
     natural color, 3 bands,
     1 cm resolution
Contact: Dana Slaymaker
Email    info@resourcemappinggis.com
Web site http://resourcemappinggis.com/
                       Resource Mapping plane
CAO plane and system        and system
• Fist contracted through USGS to do pilot research
  for TNC, Army, and Fish & Wildlife lands.
• USGS funded development of next generation
  system that included image normalization for direct
  comparison and dual scale system
• Initial results were promising
• Recently TNCH has taken a leading role in
  implementation creating Resource Mapping Hawaii,
  local capacity
Matching multispectral and
natural color imagery over
the Hakalau National
Wildlife Reserve
This dual-scale approach is
especially helpful with
species like Australian tree
fern which is not
distinguishable from the
native fern species using
any spectral identifiers.
Resource Mappings goal is
usually to find the lowest
resolution adequate to a
specific mapping need to
reduce costs, but the dual
camera set up gives them
several options to meet
these needs.
A subsection of the natural color with it’s corresponding multispectral. The multispectral shows better spectral discrimination
between trees while the natural color has more detail. A three times increase in spatial resolution in the natural color was
sufficient to distinguish the tree species needed to map in the the Hakalau National Wildlife Reserve.
                   Example of Ultra High Resolution Image Interpretability



                                                     Psidum
                                                     cattelianum




                                                                       Pandanus
                                                         Psidum        tectorius
                                                         cattelianum




       Melastoma
       candidum                     Schefflera
                                   actinophylla

                                                                         Metrosideros
                                                                         polymorpha




                                                                                        Macaranga
                                                                                         mappa

                                                  Cecropia
                                                  obtusifolia



                                                                                                    Nephrolepis
                                                         Melochia                                    multiflora
                                                         umbellata
                                                                            Trema
                                                                            orentalis




2 cm resolution
Range of sub-sampling options, from 7 cm (above) to 1cm per pixel   Detail of a single fern at both resolutions
          Current Status of Systems
• Current product lines
   – Resource mapping products are focused on very fine detailed image
     interpretation over small to moderate AOI’s.
   – CAO data is geared toward large scale, automated, image analysis.
• Availability
   – Resource Mapping will be local in June of this year with a dedicated
     plane, pilot, and processing facility
   – CAO is local but still in research phase of development from a large
     scale implementation perspective.
• Cost
   – Resource Mapping produces full coverage, 2 scale, ortho-imagery at
     less than $2.00 per acre.
   – CAO costs requests should be directed to Greg Asner of Carnegie.
                          Summary
• Carnegie Airborne Observatory
   – Huge potential for automated large scale mapping of certain species,
     vegetation structure, and physiology; high-resolution DEMs.
   – True costs still being determined; currently available with conditions;
     long term availability being determined.

• Resource Mapping
   – Straight forward approach to mapping and monitoring plant
     communities and species using image interpretation with ultra high
     resolution data
   – Known costs, fast and simple tasking ability.
Carnegie Airborne                                Resource Mapping
    Observatory                               – Very high resolution
– Hyperspectral                                 multispecral, 4 bands,
  imagery, 350 bands,                           20 cm resolution
  30 cm resolution                            – Ultra high resolution
– Full waveform                                 natural color, 3 bands,
  LIDAR                                         1 cm resolution


Contact: Greg Asner                         Contact: Dana Slaymaker
email gpa@stanford.edu                      Email    info@resourcemappinggis.com
website- http://cao.stanford.edu/           Web site http://resourcemappinggis.com/


Note: Any use of trade, product, or firm names in this presentation is for descriptive
    purposes only and does not imply endorsement be the U.S. Government.