An observation can be taken for many different purposes. An exploratory observation is one taken in the spirit of exploration—no firm scientific rationale can be given for it because there are not enough data to make a scientific argument for a measurement. Such observations are very hard to come by (it is almost impossible for an investigator to propose such a measurement), yet all knowledge begins with exploratory measurements and every research program should include them. There are regions of the ocean that have never been measured, as well as parts of the upper atmosphere, and the land surface.
A critical measurement is one that tests a specific hypothesis. While common in particle physics, in which most accelerator experiments are specifically designed to test aspects of the current theory, these measurements are relatively rare in geophysics and even more rarely successful. Contradictory as it may seem, a program of observations can be hypothesis driven, even though a critical measurement may not exist.
A measurement can be made to document the secular change of some relevant climatic quantity, such as global surface temperature or upper-tropospheric humidity, for the purpose of documenting some aspect of global change and for providing the data to compare to models. Such a measurement could be critical if a prediction of such changes has been precise and unambiguous, but this rarely happens in geophysics because it is hardly possible to control the surroundings of a measurement.
Measurements can be taken as part of a forecast-analysis cycle, and these would generally be classified as an operational measurement. These measurements tend to be taken in a regular and systematic manner as input to an ongoing prediction system. While not performed (or funded) as research, such measurements can be extremely valuable since they make available to the research community observations that could (or would) not be supported through research—the upper-air observing network is a good example of this.
Some measurements are taken mainly to validate other measurements. Examples are measurements of SST from drifting buoys taken to calibrate the operational AVHRR satellite measurements of surface radiance. Because the satellite measurements are subject to cloud obscurations and are affected by aerosols in the atmosphere, which are not carefully measured, the in situ measurement of SST provides an absolute measure from which the satellites can derive global SST on a regular basis.
Some measurements are taken in a regular and systematic manner and, while serving the purpose for which they are taken, do not have the accuracy or reliability for some different purpose. An example is the upper-air network, which by itself has been incapable of documenting small temperature changes in the upper atmosphere. Increasing the accuracy or reliability of such measurements leaves