The staff of the Goddard Earth Sciences DISC is continually striving to make Giovanni a preeminent data analysis and exploration tool for the use of scientists, educators, and students. It must be noted, however, that Giovanni primarily utilizes remote-sensing data from NASA instruments and satellites, as well as other remotely-sensed datasets. There are a few other types of data, such as output from models or ground station data, but remote-sensing data constitute most of the data in the various Giovanni instances.
The use of remote-sensing data for research requires a researcher to understand how “good” the data actually is. The production of remotely-sensed geophysical parameters from satellite-borne instruments is usually the result of a multi-step computational process. In general, every remotely-sensed data parameter in Giovanni is generated by one or more algorithms that are applied to the actual measurements an instrument acquires. Some of the algorithms are simple, and others are quite complex. Some algorithms require other data (“ancillary” data) from other sources to complete the generation of a particular data product.
Remotely-sensed data products therefore have their own individual data quality considerations. Each of the chapters in the Giovanni-3 Online Users Manual provides information on the instrument and the data obtained from it. Additional work may be required to fully understand the accuracy of a remotely-sensed data product and the degree of error associated with it. This information may be obtained by searching the Web for related resources; the GES DISC staff will assist with data quality inquiries submitted to email@example.com.
Remotely-sensed datasets are rarely complete. Satellites and their instruments experience “down” times when they cannot make observations. Some types of data can’t be acquired when obscured by clouds or dust or haze in the atmosphere. Other types of data cannot be acquired if the Sun is not in the right position. There are many other conditions that cause remotely-sensed data sets to have missing data.
There are frequently conditions that degrade the accuracy of a remotely-sensed data product. Some of these conditions cause the algorithm generating the product to fail, and hence there will be no data; other conditions will make a data product less accurate, even though the existence of such conditions may not always be apparent. Thus, all remotely-sensed data products must be evaluated with caution and with respect to the conditions that may cause them to be incomplete or inaccurate.
Many of the datasets used in Giovanni are called “Level 3” datasets. Level 3 data indicates that the data are mapped or “gridded”, a process which requires average over an area, a time period, or both. This computational process also affects the quality and accuracy of the data. For example, the value of a data product averaged over an entire month of observations may be considerably different than the value of that data product acquired by a single observation. Giovanni also allows users to average data over time and area, which can mean that data which has been processed once to create a Level 3 data product is processed again to generate a Giovanni data output. We provide documentation of the computational steps performed by Giovanni, but a researcher may also need to know the methods used to create the datasets used by Giovanni.
In summary, we encourage the use of Giovanni for research, and we are striving to make it an excellent tool for that purpose. Successful research investigations, however, require that researchers fully understand the characteristics and limitations of the data they are using, as well as the characteristics and limitations of the tools they utilize to analyze such data.
Return to Chapter 1 of the Giovanni-3 Users Manual >