Remote Sensing Imagery Types

And Sources

GIS Management and Implementation

GISC 6383

October 27, 2005

Neil K. Basu, Janice M. Jett, Stephen F Meigs, and Jody A. Urbanovsky

Introduction

Remote sensing can be simply defined as the collection of data about an object from a distance. Remote sensing is usually done with the help of mechanical devices known as remote sensors. These technological advancements have greatly improved the ability to receive and record information about an object without any direct contact. Most of these sensors record information about an object by measuring an object’s transmission of electromagnetic energy from reflecting and/or radiating surfaces. In the past few decades this technology has advanced on three fronts: 1) from predominantly military uses to a variety of environmental analysis applications; 2) from basic photographic systems to remote sensors that utilize the electromagnetic spectrum; 3) from aircraft to satellite platforms. Even though remote sensing imagery was primarily used in military applications it has many commercial applications in mapping land-use and cover, agriculture, soil classification, forestry, urban planning, and archaeological investigations to name a few. It is the purpose of this document to provide a brief overview of four broad categories of remote sensing imagery. These imagery categories are multispectral, hyperspectral, radar and high resolution.

Criteria for analysis of remote sensing imagery:

Spatial Resolution

Spatial resolution is another way of stating the size of each pixel. Pixel size is a direct indicator of the spatial resolution because the pixel size defines the smallest elements that can be detected by the sensor. Spatial resolution is one of the major criteria used in evaluating and choosing an imagery product. The required spatial resolution for a project is what usually determines the cost of the data.

Spectral Resolution

Spectral resolution refers to the spectral position and bandwidth of the bands used in the acquisition of the imagery. The bandwidths and position of a particular sensor is an important factor when deciding which platform will be suited for each application. When considering the number of spectral bands the width of those bands will also needs to be considered. Narrow bands can offer more precise coverage but can often exclude need frequencies for many applications.

Coverage Extent

The main factor included in coverage extent is temporal resolution. Temporal resolution refers to the time it takes an imaging system to return to an area and collect another image. Temporal resolution must be considered when choosing an imagery product because the platform may not be over the area of interest when the acquisition is needed or the area may be cloud covered during imaging. Two other coverage topics to think about are whether or not the satellite overflies the area for which imagery is needed and the scene size / swath width. The scene size is the area on the ground that each individual image covers.

Data Format and Cost

It is crucial to know what software application you will be using to analyze the imagery. When purchasing imagery from a commercial provider they tend to have multiple data formats with GeoTIFF being a common format. Although most data is available over the Internet or by purchase on a CD-ROM older archival data may only be available on an outdated medium such as 9-tracks. Prices of imagery are usually a direct result of resolution and the provider. Prices can range from extremely cheap of lower resolution government data sources to extremely expensive for high resolution custom scene commercial providers.

Imagery Categories:

Multispectral

MSS (Multispectral scanner) satellites map the Earth’s surface by using the three visible channels along with the near-infrared band on the electromagnetic spectrum. In 1972 the USA launched Landsat 1 which was the first earth observing satellite specifically designed for analysis of land features using the MSS. Landsat satellites 1-5, ASTER, and IRS (Indian Remote Sensing Satellite) are all MSS equipped satellites. Landsat 1-5 are the primary imagery source for archived data of the United Sates. Landsat data have been used by government, commercial, industrial, civilian and educational communities for over 30 years. Landsat data have potential applications for monitoring changes on the Earth’s surface over periods of several months to two decades. With the introduction of higher resolution and hyperspectral sensors the need for newly acquired MSS data was ceased in October 1992 on the Landsat 5. MSS data although is still available from ASTER and IRS.

Example Platforms:

Landsat 1 – 3 MSS

  • Spatial Resolution: 60m x 80m
  • Spectral Resolution: MS 4 bands, 500 – 1100nm
  • Revisit Time: 18 days
  • Scene Size: 185 x 185km
  • Coverage Extent: Global data between 81 degrees North latitude and 81 degrees South latitude.
  • Data Format: NDF
  • Cost: $200 -$375

Landsat TM 4 – 5

  • Spatial Resolution: 30m x 30m
  • Spectral Resolution: MS 7 bands, VNIR - TIR
  • Revisit Time: 16 days
  • Swath Size: 170 x 183km
  • Coverage Extent: Global data between 81 degrees North latitude and 81 degrees South latitude.
  • Data Format: NDF, EFF, & GeoTIFF
  • Cost: $425 -$625

Hyperspectral

Hyperspectral imagery is similar to multispectral imagery, but instead of only having 3 to 10 bands, hyperspectral data has hundreds of bands which make it possible to distinguish between extremely detailed spectral response curves. Hyperspectral imagery is used for a variety of purposes. A tank is camouflaged and firing on your troops by use of a land based hyperspectral image the tank can be located and destroyed. Some other uses include law-enforcement finding cannabis in remote fields. Scientists can evaluate pollution, soil types, and even differentiate between oak and elm trees when using hyperspectral imagery.

With hyperspectral imagery there are two different types of systems available and they are categorized by either airborne or spaceborne systems. In general the spectral range of the airborne system is 380 nm to 12,700 nm, while the ranges of the spaceborne systems are 400 – 14,400 nm. The number of bands and bandwidths very from system to system the overall ranges for hyperspectral images are 1-288 bands at 2-2000 nm bandwidths.

Pixel size per spaceborne system is fixed like it is 30m for HYPERION while those for airborne system can fluctuate depending on the height or altitude of the aircraft an example is CASI2 whose pixel size is sub-meter to 10m. Resolution and whether or not the system is airborne or spaceborne will have an effect on the price of each swath. Most airborne systems price their data from $250 -$1000 while spaceborne systems can go as high as $2500 per swath.

Example Platform:

EO-1 (Hyperion Sensor)

  • Spatial Resolution: 30m x 30m
  • Spectral Resolution: 220 bands, 400 – 2500nm @ 10nm
  • Revisit Time: 16 days
  • Scene Size: 7.5 x 100km
  • Data Format: HDF or GeoTIFF
  • Cost: $250 per image

Radar

Radar measures the strength and round-trip time of the microwave signals that are emitted by a radar antenna and reflected off a distant surface or object. RADAR is
an acronym for RAdio Detection And Ranging or Radio Angle Detection And Ranging. For an imaging radar system, about 1500 high- power pulses per second are transmitted toward the target or imaging area. Typical bandwidths for imaging radar are in the range 10 to 200 MHz. The radar moves along a flight path and the area illuminated by the radar, or footprint, is moved along the surface in a swath, building the image as it does so.

There are several benefits, and drawbacks for choosing to use radar imagery. One of the main benefits of using radar imagery is that images can be generated though clouds and in darkness. Others advantages include, Near-Real Time processing of data, direct downlink and onboard recorder storage capacity, and data calibration for change detection studies Drawbacks of using radar include, difficulty mapping areas with dense vegetation or areas where it cannot see the earth surface, such as the shadowed areas of mountains or buildings. This is caused by the angle of flight with respect to the target areas. The military also may restrict the use of radar around certain facilities and areas where information may be needed.

A number of factors affect the final cost of radar data. They include the size and location of the project area, the vertical and horizontal accuracy requirement for the data, and the amount of post-processing needed to produce certain product types. There are seven beam modes and 35 beam positions for a wide range of imaging options. Radar also has varying resolutions (8 - 100 meters), and varying swath widths of 50 - 500 km.

Example Platform:

RADARSAT

  • Spatial Resolution: 10 – 100m
  • Spectral Resolution: C-Band 35 – 500km
  • Revisit Time: 24 days
  • Scene Size: 50 x 50km up to 500 x 500km
  • Data Format: CEOS
  • Cost: $2750 - $3750 per image

High Spatial Resolution

The spatial resolution of data is what defines the smallest possible space that can be resolved. For example, a car may be approximately 3 meters in length, so the pixel size necessary to capture the image of the car would need to be 3 meters or less. Hence, the advancements in spatial resolution have provided greater accuracy and precision in the fields of remote sensing. High spatial resolution platforms exist as both airborne platforms as well as satellite platforms. Satellite platforms have a fixed spatial resolution while airborne platform resolutions are adjusted depending on altitude of the plane as well as the dimensions of the area being covered.

Commercial uses for high spatial resolution remotely sensed data ( < 10m), like all other RS data are ever growing beyond the traditional uses for weather prediction, surveying and mapping. The utilization of such data is expanding to new fields because of the lowering cost of high resolution platforms and the greater number of proposed platforms. The current and future market for spatial data is for spatially attributed temporal information such as the niche in the transportation sector for real-time navigation with dynamic map displays created from satellite imagery coupled with GPS. Just a few of the uses for high-resolution data include: disaster monitoring, disease detection, real estate appraisal, city and urban planning, navigation safety, financial and insurance services, reconnaissance and entertainment.

At present, the highest resolution commercial satellite data available comes from QuickBird, which was launched in October 2001 and collects 61 centimeter resolution panchromatic (black and white) and 2.44 m multispectral. The main use for QuickBird data is for focusing on the assessment and management of land, infrastructure and natural resources, although each of these topics consists of innumerable subtopics.

Example Platforms:

QuickBird

  • Spatial Resolution: 60 x 70 cm (Pan)

2.4 x 2.8m (MS)

  • Spectral Resolution: 0.45 to 0.9 microns (Pan)

0.45 to 0.9 microns at 0.08 microns over 4 bands

  • Revisit Time: 1-3.5 days
  • Scene Size: 16.5 x 16.5 km at nadir
  • Data Format: GeoTIFF 1.0, NITF 2.1 or 2.0
  • Cost: A basic scene will vary from $400 to $10,000. Many options for varying quality levels.

IKONOS

  • Spatial Resolution: 1 m (Pan)

4 m (MS) over 4 channels

  • Spectral Resolution: 0.45 to 0.9 microns (Pan) 0.45 to 0.88 microns over 4 channels
  • Revist Time: 3 days at 1 m, 1 to 2 days for 1.5 m
  • Scene Size: variable, usually 11.5 km at nadir
  • Data Format: GeoTIFF, NITF
  • Cost: $7 to $132 per sq.km

Other platforms for high spatial resolution include: ORBView, EROS, SPOT and IRS-1C and 1D

Conclusion

The great variety of remote sensing data collectors, types, and vendors clearly is a positive aspect for commercial use, but it also creates difficulty when choosing appropriate data. Projects requiring remotely sensed data must be evaluated on the basis of what type of data is most appropriate. The major factors to be considered are the relative importance of spatial resolution, spectral resolution, coverage extent and cost for each project. For example; while high spatial resolution may appear to be the best possible data for certain applications because it allows for the analysis of very small objects, it may not be useful for projects requiring refined and specific spectral bands such as the thermal bands. And in some cases, a specific wavelength range is more important rather than spatial resolution or revisit time. In order to successfully choose the appropriate data, the problem being addressed must be well understood in the terms of the issues discussed above.

Bibliography

Personal Correspondance with Dr. Mohamed Abdelsalam