Data and Information
On this page...
Determining Imagery Data Needs
Imaging Systems and Sources
Imagery Archives
Determining Imagery Data Needs
Karl Brown - April 2006
Introduction and Background
The project defined imagery use, task objective, and scale of decision making will help determine the needs for remote sensing products. One logical pathway that can be employed addresses the three parts of remote sensing imagery resolution. These three parts are:
- Spectral Resolution – the bands of energy used in collecting the image
- Spatial Resolution – the ground sample distance and effective viewable product
- Temporal Resolution – the timing and / or revisit cycle desired
The data products section contains greater detail on these concepts, including definitions, examples, and technical details. The Hardware section of this webpage gives a description of the sensors, their return cycles if satellite based, and the resolutions of the products available.
Several NPS Inventory Projects deploy a variety of sensors, depending on the type of information needed. Monitoring efforts require multiple temporal data, with (consistent) repeat measurements a valuable criterion for selection.
Baseline Inventories:
For 1-time inventories, or single temporal studies, inexpensive aerial photography may provide the chosen product. This has the advantage of simplicity, relatively quick to contract for, has a wide variety on off-the shelf contracts in place, and generates a relatively common product of 9x9 hard copy prints that allow for stereo viewing with an inexpensive pocket stereoscope. Scales and format are variable, and the use of 1:10K to 1:12K scale has worked very well for vegetation sampling and ecological inventories, either with true color or color infrared. A photo scale at or above 1:24K is more difficult for field navigation and to conduct ecological inventories. Whether collected on film (analog) or by digital means, this product must be preprocessed for machine analysis if that is the desired output. Uncorrected contact prints are not georeferenced and are able to generate a stereo view because the parallax has not been removed during the georectification process. The steps, choices, pros and cons of this imagery format are extensive. The pre-processing steps from scanning to georectification are extensive and non-trivial. Recognize that photo-mosaics are not matched to the quadrangles as orthophotos, and do not allow GIS analysis because they are not georectified. If your project intends to scan and rectify to the 1:24K quadrangle base, then other image products may be more efficient in both collection and design to meet your needs.
Moderate Complexity:
Multi-scale or multi-temporal imagery collections increase the options and investment for the user. Digital aerial photography with GPS linked inertial navigation packages on board can produce rectified products more quickly. Recognize that the rectified image does not afford a stereo capability; however some sensor packages are flown with nadir and forward and or rear positioned cameras to not only rectify the imagery, but to provide stereo as well. A monitoring effort that looks for change detection will desire a stable sensor, with multiple views at the same scale and resolution for machine processing of the landscape change. The steps of preprocessing by either the vendor or the user before use and analysis increase project costs. Overall project efficiency may be increased, and could make the complexity well worth the investment. The Landsat products are standardized, provide moderate scale resolution, are easy and inexpensive to order, and allow relatively simple comparisons to be done quickly and relatively cheaply.
High Complexity:
Most multi-sensor or multi-process hybrids are more complex and require a higher understanding of remote sensing science to design and apply to a project. Specific signature searches like using a hyperspectral sensor for invasive plants involve expensive sensors and platforms ($80K per mission – AVIRIS) spanning to the space-based Hyperion sensors with NASA data agreements needed to task the imagery. The preprocessing is excessive, complicated, and expensive. The ability to discern specific signatures also requires the collection of that plant phenology with a ground sampling instrument, or the use of a signature library. Signature libraries for plants are extremely variable and problematic to have the appropriate time and plant condition represented in the library. Laser telemetry or Lidar collections also carry significant preprocessing tasks to utilize the data. Once obtained, the dataset can help with vegetation structure, bathymetry, bare earth models, or other height-related studies that need fine elevation resolution.
Potential Decision Steps for Identifying Needs
What is the appropriate scale for your project? Do you need to see individual plants, pockets of features, or specific target items to help you answer your management question? These cases could utilize a relatively fine scale imagery with maximum ground detail. High resolution digital imagery can address this, as can traditional aerial photography. One meter digital will have less ground resolution than fine scale aerial photos due to pixilation; however, digital imagery has the advantage of being machine ready for analysis. Ground rectification is still needed for either.
Do you need a particular spectral tool or analysis? Here are some examples of the use of color infrared imagery:
- Plant stress surveys: oak wilt, mountain pine beetle, defoliators, wooly adelgid, acid rain, etc.
- Deciduous versus coniferous forest cover
- Relative biomass, all living plants having pink to red hues
- Inert surfaces having a cyan hue
Here are some advantages of true color:
- Natural colors are easier to interpret by non color-infrared trained staff. This format is more likely to be recognized by other park staff members.
- Landform and landcover spectral information in shadows (CIR is just black – no data or signature information)
- Less expensive to process; slight difference in price possibly
Do the spectral bands need to include thermal, IR, etc?
A multi-spectral sensor that exceeds red, green, and blue would be needed. Depending on your thermal requirement, this band may be useful for you.
What is the revisit requirement to answer the management question?
A trend study needs multiple coincident measurements at a scale to discern the monitoring topic. A sixteen day return interval by Landsat may provide an appropriate cycle; however, cloud free collections on that frequency are not the norm in mountainous terrain. Change detection is a classic example of multi-temporal remote sensing.
Examples of classes of NPS imagery usage
Maintenance pavement restriping requirements: Single contract period for analysis of stripe condition. Spatial needs to detect stripe condition and brightness.
1:10K – 1:12K Color aerial photography – lowest cost, fine scale detail
“High Resolution” 1 meter product will be pixilated at the stripe evaluation level.
Forest health insect and disease evaluation: General landscape forest health may be visible on Landsat 30 meter scale products using live foliage ratios. Specific pockets seen on the 30-meter product could be evaluated with finer scale imagery to locate or evaluate pockets to verify and plan management efforts. High resolution (1 meter) digital would be more expensive, and would yield finer spatial and spectral detail. For multi-year evaluations, you would have to pay more for the product than the 30-meter Landsat imagery.
Leafy spurge mapping and control: (example Theodore Roosevelt NP, North Dakota)
Difficult by CIR or true color aerial photography due to yellow sweet clover as a partner on the landscape. Lots of ground checking needed for signature. Spectral requirement leads to considering an option of the CASI hyperspectral solution on an ARS plane flown at moderate flight height and 4-meter resolution. Significant pre-processing needed to map this plant with imagery, and still requires lots of ground survey effort.
Fire Perimeters and burn severity
This effort with the USGS has developed the NDBR methodology with Landsat products that is well documented and successful.
Summary:
Select the imagery product that produces the information at the scale you need to answer the management question, in the spectral range needed to define the signature of interest, and at a timing and revisit spacing that allows for the determination of the trend of interest.
Draft prepared by Karl E. Brown (karl_brown@nps.gov)
Contact Mike Story for imagery and acquisition questions or comments.
A few imagery archives
Earth Explorer
EOS Data Gateway
National Geospatial-Intelligence Agency
National Land Cover Data (NLCD)
National Satellite Land Remotes Sensing Data Archive
|
