[Back to Detailed Agenda]
|
| Chairs: Hank Shugart, Frank Muller-Karger
|
Discussion Questions:
- Given our science questions, what are the remote sensing observation and modeling requirements for biodiversity? What should a biodiversity sensor suite include? What would a biodiversity mission look like? How do current and planned missions address biodiversity requirements? Where are the gaps?
- How can remote sensing programs link to in situ efforts to understand and monitor the condition of biodiversity? What are the opportunities to work across scales and/or connect to work at finer (e.g., plot) scales?
- More specifically, how can NASA contribute to the global effort to "achieve by 2010 a significant reduction of the current rate of biodiversity loss" — the so-called 2010 Biodiversity Target? What satellite observations and associated models are appropriate here in terms of a monitoring strategy and how can we link them to other efforts?
- What are the opportunities to transition NASA biodiversity research projects into applications activities? How can we improve upon this transition?
|
|
|
|
--- Please forward any additional comments to the relevant Program Manager at NASA Headquarters. ---
|
I heard today that hyperspectral remote sensing was excluded as a requirement for marine and terrestrial biodiversity research in the Joint Workshop.
I would like to challenge this decision and provide some reasons by considering my research on remote sensing of coral reef ecosystems experience in your workshop. If you are interested in just "mapping" the extent of coastal zone ecosystems in very broad categories (coral, seagrass, sand) then essentially multispectral imaging at moderate spatial resolution (Landsat) is adequate. Reefs can run in size from large barrier structures to small patch reefs. Therefore, the higher the spatial resolution the better so as to consider patch reefs. Further, after collecting benthic type spectra (from a handheld underwater spectroradiometer) since 2002 in coral reefs and adjacent seagrass beds in the Bahamas, Puerto Rico, USVI, and the Florida Keys I would argue that hyperspectral remote sensing is needed to distinguish differences in benthic types and variability in these benthic types. Coral reef ecosystems are vastly heterogeneous structures with not only varying morphology, but color as well, even within species, due to conditions of health, herbivory, mortality, and composition of zooxanthallae for starters. Since only radiance from the visible range of the EM spectrum can penetrate the water column, shallow water remote sensing is done best with channels in the visible range. A high spatial resolution sensor that can provide detail for texture classifications is useful for some delineation in morphology, but doesn't take into account the variation due to color. Color (exhibited by the photosynthetic algae w/in coral or lack of it, for example) is very important in my research as I am trying to quantify coral bleaching, coral recovery, disease, mortality, living coral, and algal overgrowth. More research in this area is needed since there is so much variability in benthic type spectra. These coral reef conditions are important factors in understanding the biodiversity and productivity of a coral reef ecosystem. Therefore, a high spatial and high spectral resolution (in the visible range) sensor is superior for shallow water ecosystem research.
Please let me know if you would like me to provide some references and/or examples from my spectral library.
I could speak to the terrestrial side of this, but I feel Greg Asner would have the best first-hand knowledge and expert opinion on this.
– submitted by Liane Guild at 2006-08-24 22:28:41 |
I would like to comment on the value of hyperspectral technology for aquatic measurements. I have had a research project for four years,in which we have collected hyperspectral data to map aquatic weeds in the Sacramento-San Joaquin Delta, an area of about 3500 km2, requiring more than 100 flightlines, with 3m resolution hyperspectral airborne imagery (see poster abstract). We have been able to map water hyacinth and egeria (more correctly submerged aquatic weeds) each year with about 90+% accuracy (based on some 2000 GCP per year).
Prior to this study, the CA rResources Agency tried for several years to obtain good measurements with other camera and four band systems but without success. This is because the conditions change across the delta (algae, sediment, tide height, etc.) and the differences between these weeds (and several species of native vegetation including cattails, tules, primrose, and a number of others) is to variable to accurately map individual species. Based on these four years of data (we are now doing our second year of fall data collections to map before and after weed eradication), I can say with certainty, that it could not be done with multiband data.
– submitted by Susan Ustin at 2006-08-24 14:25:39 |
Comment from Greg Asner regarding spectroscopy and biodiversity:
Our ability to remotely measure species richness and abundance in any ecosystem rests on our ability to remotely measure biochemical and structural variability. At this time, the terrestrial ecology community lacks the data and synthetic knowledge to link observed spectral variability to biochemical and structural variability in ecosystems.
A spectroscopic signature of a species is not a series of bands put together. The spectral signature is a single measurement containing scattering and absorption features that probe the biochemical and structural properties of plants. It is possible that a set of narrow bands, which inherently lose the shape of the spectroscopic signature of a species, will under-determine the biochemistry and structural attributes of that species, and thus biodiversity measurements will be inaccurate. More research is needed to address these issues.
Greg Asner
gasner@globalecology.stanford.edu
– submitted by Robert Green at 2006-08-24 14:10:49 |
Regarding the section below, I do not believe this is a NASA biodiversity community consensus.
=============
Biodiversity Discussion Group,Summary Bullets,August 22, 2005,Frank Muller-Karger and Hank Shugart,Rapporteurs: Chris Moses and
Technological Issues:
o Hyperspectral not a requirement in either marine or terrestrial efforts, but both communities need improved multispectral (narrower bands, more bands, improved signal to noise, improved revisit)
==============
Imaging spectroscopy is narrower bands, more bands, and improved SNR.
The literature documented ability of imaging spectroscopy to measure and monitor through time the following parameters relate directly to NASA biodiversity science questions of the present and future.
Fractional Vegetation Cover:Green vegetation (GV), Non-photosynthetic vegetation (NPV) , Soil, etc
Plant Functional Types: Grass/forb/shrub/tree, Thick/thin leaves, Broad/needle leaves, Deciduous/evergreen
Disturbance and Response of Vegetation Cover: Recovery/change in fractional cover, Recovery/change in ecosystem function, Spectral signatures of vegetation stress and damage
Canopy Physiology and Function: Light use efficiency, Canopy liquid water, Canopy chemistry
l will work with my colleagues to provide you journal references documenting the measurement role of imaging spectroscopy in addressing a range biodiversity sciences questions.
Robert O. Green
rog@jpl.nasa.gov
– submitted by Robert Green at 2006-08-24 13:46:12 |
Comment from Dr. Curtis Mobley regarding:
=============
Biodiversity Discussion Group Summary Bullets August 22, 2005 Frank Muller-Karger and Hank Shugart Rapporteurs: Chris Moses and
Technological Issues:
o Hyperspectral not a requirement in either marine or terrestrial efforts, but both communities need improved multispectral (narrower bands, more bands, improved signal to noise, improved revisit)
==============
Definitely not concensus. I'm now a firm believer in the value of well calibrated hyperspectral Rrs, especially for applications like retrieval of bathymetry, bottom classification (mapping and monitoring sea grass beds, coral reefs, etc), and coastal water quality parameters. See Mobley et al, as one example. I believe the thrust should be towards high quality hyperspectral (well calibrated sensors, better atmospheric correction, etc), not just better multispectral.
Curt Mobley
-----------------------------------
Curtis D. Mobley, Ph.D.
Vice President and Senior Scientist
Sequoia Scientific, Inc.
2700 Richards Road, Suite 107
Bellevue, WA 98005
voice: 425-641-0944 ext 109
fax: 425-643-0595
email: curtis.mobley@sequoiasci.com
WWW: www.sequoiasci.com
personal: www.curtismobley.com
-----------------------------------
– submitted by Robert Green at 2006-08-24 13:20:45 |
One thing I didn't hear much about was the use of vegetation status derived from remote sensing data as *primary* data for biodiversity analyses, e.g. as the response variable in an analysis, not just a predictor. Remote sensing and modelling do a pretty good job with direct estimations of abiotic conditions which are underlying much of biodiversity, but many of the biodiversity mapping projects, even the ones focused on plants, still require field data to derive the models. Why were species mapping projects not even mentioned in this session? Research with CASI and other hyperspatial, hyperspectral sensors have shown the ability to map per-tree species at a reasonable level.
It seems the issues of sensors fall more into the realm of whether you are trying to map ONLY covariates of species diversity, or whether you want to actually map species or non-taxonomic groupings such as lifeforms.
– submitted by Jonathan Greenberg at 2006-08-21 23:47:41 |
While I agree with much of the break-out group discussion that it is important to focus on diversity "hot spots" (in terms of acquiring very high resolution imagery and 3D structural sampling), I think that priorities for conservation planning (and 2010 targeting) indicate we must maintain and should focus on a landscape perspective for biodiversity research applications. As such our priorities should be on spatially extensive moderately high resolution (~20-30m) multi-sensor (optical, InSAR, Lidar-sampling) suite of regular observations supplemented, if and were possible, with higher resolution (spatial, spectral, temporal) and field observational data sets.
– submitted by Nadine Laporte at 2006-08-21 23:22:17 |
|
|
|
+ QUESTIONS? CONTACT US
|
