Remote sensing information flow is a complex process involving five phases: (1) pre-launch calibration, (2) data ingest (collection), (3) digital image processing, (4) storage and archiving, and (5) retrieval and application. [1] Satellite data must be transformed from newly-collected petabytes of binary code, to calibrated data occupying terabytes of storage area, to gigabytes that are usable for modeling and observational systems, to megabytes that can be used in daily applications. [2] Potential for error exists in all of these transformations, but NASA and other satellite developers are continually creating and improving calibration tools to reduce amount of potential error. For most satellites, data handbooks exist that detail collection and calibration procedures.

Pre-launch
During the pre-launch correction process, scientists characterize and calibrate all satellite sensors to ensure accuracy. First they ‘characterize’ the instruments, a process that involves performing a set of operations to quantitatively express the instrument’s response to the conditions experienced in orbit. [3] Then they calibrate the sensor radiometrically (with respect to the electromagnetic spectrum) and geometrically, both pre-launch and repeatedly while in orbit, to reduce error resulting from sensor failure and space “noise.” [4] The launch of Landsat 7 introduced in a new generation of calibration strategies to bring its radiometric accuracy within a ±5% uncertainty over the five-year life of the mission. [5] All of the EOS satellites, including Terra, will also have onboard calibration instruments that will be monitored independently, and with respect to one another, throughout the fifteen-year mission. [6] [top]

Data Ingest (Collection)
To minimize error in receiving the data, satellites have counterpart ground systems (ingest systems) that receive, calibrate, and store the same data. The Landsat ground system, for example, includes, ground stations for uplinking commands and receiving data, a spacecraft control center, and a data handling facility. [7] Once the data is received by the ingest system, it is time-stamped and undergoes extensive quality and statistical sampling. Monitors located in control centers constantly observe the data for anomalies; Calibration software corrects incoming data and flags questionable data. [top]

Digital Image Processing
Once the digital pixels are obtained, they must undergo a three-step process to generate a meaningful product: (1) preprocessing, (2) display and enhancement, and (3) information extraction. [8] Preprocessing generally involves a first round of corrections that eliminate error caused by sensors and by environmental factors. Preprocessing also corrects the image geographically, so that the data corresponds to the representative point on Earth. Information enhancement adjusts pixels either individually or simultaneously to change the magnification, filtering, and textures of the image. Information extraction involves interpreting the pixels into recognizable patterns using primary colors. The enhancement processes are carefully controlled. Recently, scientists have used expert systems, in which the computer draws from a stored database of human knowledge to determine the best depiction of the data, and neural networks, in which the computer is ‘taught’ what decisions to make interpreting the data. [9] [top]

Storage and Archiving
The ground systems that receive and process data also store data. Both raw data and processed imagery are usually stored in duplicate to protect against loss. In the U.S., NASA has established nine Data Active Archive Centers (DAACs). [10] Each DAAC focuses on a specific scientific discipline and is responsible for processing, archiving, and distributing data from the Earth observing satellite missions, including Landsat, Terra and future EOS missions, and SeaWiFS. The DAACs also provide a full range of user support and data access.[top]

Retrieval and Application
Consistent with the ‘scientific method,’ a scientist states the problem encountered, determines a hypothesis, and then locates data to support or dispute the hypothesis. [11] Since NASA launches its satellites with particular research goals in mind, scientists hoping to use the satellite data for other purposes may find themselves working backwards, trying to identify a question that the data supports. While data may be used for purposes other than the original mission, decisions must be carefully made to ensure that other applications are legitimate. For example, the limitations of each sensor must be weighed against the potential application. [12] Satellites such as SPOT and IKONOS are taking advantage of the interest in satellite commercial applications by providing features such as global coverage, pointable sensors, spatial resolution between 1-10 meters, and high spectral resolution.

Once the data has been processed and the correct application has been determined, the data must be transformed to match the needs of the scientist or other end-user. This transformation may include further algorithmic analyses, finer definition of the spatial resolution, or overlaying the image with other information. It may also include data-distribution and interpretation. [top]