---

The Problem

• Time series used for studying long-term climate often contain artificial inhomogeneities, especially jumps at times when measuring instruments are replaced. A striking example occurred at Calcutta, India in 1968. Many other such cases can be found, although typically they are not as extreme.
• The difficulties are greater for radiosonde than surface or satellite data. The obstacles include:
  1. Historical documentation of changes in instruments or measurement practices is often incomplete, ambiguous or erroneous.
  2. The magnitudes of artificial inhomogeneities are often comparable to those of real climate variability, complicating the distinction between the two.
  3. Instrument changes are controlled at the country or station level; sometimes changes affect a large, contiguous area, hampering efforts to use “buddy-checks” of neighboring stations; in other instances changes occur at different times at each station, greatly magnifying the number of inhomogeneities which must be identified.
  4. The sparseness of the upper-air station network makes use of “buddy-checks” of neighboring stations more problematic.

Back to the main page