A great deal of effort went into the homogeneity adjustments. Yet the effects of the homogeneity adjustments on global average temperature trends are minor (Easterling and Peterson 1995b). However, on scales of half a continent or smaller, the homogeneity adjustments can have an impact. On an individual time series, the effects of the adjustments can be enormous. These adjustments are the best we could do given the paucity of historical station history metadata on a global scale. But using an approach based on a reference series created from surrounding stations means that the adjusted station's data is more indicative of regional climate change and less representative of local microclimatic change than an individual station not needing adjustments.
A change in the type of thermometer shelter used at many Australian observation sites in the early 20th century resulted in a sudden drop in recorded temperatures which is entirely spurious. It is for this reason that these early data are currently not used for monitoring climate change. Other common changes at Australian sites over time include location moves, construction of buildings or growth of vegetation around the observation site and, more recently, the introduction of Automatic Weather Stations.
The impacts of these changes on the data are often comparable in size to real climate variations, so they need to be removed before long-term trends are investigated. Procedures to identify and adjust for non-climatic changes in historical climate data generally involve a combination of:
Ø investigating historical information (metadata) about the observation site,
Ø using statistical tests to compare records from nearby locations, and
Ø using comparison data recorded simultaneously at old and new locations, or with old and new instrument types.
Full details of the procedure are described here
