Changes in observing instrumentation are a natural part of the maturing process for environmental monitoring that cannot easily be avoided. The instrumentation may be simple or complicated. It doesn't really matter. In either case, it is a matter of time before it is replaced. The difference in how an instrument takes measurements and the materials used in the instrument can significantly affect the quantity measured (i.e., the data). These factors introduce biases into the climate record. Instruments can also slowly drift or deteriorate, and show discontinuities when swapped out.

To understand data, we must understand how the instrumentation works over the range of environmental conditions it is expected to be exposed to. Otherwise, we cannot be sure of the full range of its accuracy.

As we have seen, accuracy is important to the science of climate, especially with respect to extreme events, whose frequency and intensity are being monitored closely as a sign of climate change. Requirements for accuracy can be basically summarized as follows:

a) the instrument should have and maintain calibration under given conditions to within the desired precision, and,

b) The errors under other conditions should be known and constant in time within required limits.

Thus, one must understand instrumentation responses before we can determine whether a sensor meets climate requirements. Even though there are plenty of manufacturers producing quality instrumentation, if we have not seen validated test results, we do not understand which ones meet our requirements for climate data accuracy. And another consideration….don't assume that advertised capabilities are 100 percent factual. We are all well aware of the limitations of advertising. The proof is in the observation itself.

The discussion of instrument accuracy brings us to the topic of installing non-approved instrumentation into the nation's published climate network. Simply stated, this is not a wise choice. In fact, it's a no-no (Figure 6). Refrain from the temptation. If you need to open a site with non-approved instrumentation for the less demanding accuracies of the public forecast and warning missions, and you can do so within policy and resource limitations, then that's your call. But even here, it is prudent to understand the capabilities (and limitations) of the instrumentation you are using.

Approved Thermometer
Figure 6. Install only Approved Instrumentation at Published Climate Stations

It is especially harmful to install non-approved equipment at a published climate station and not document that change in the station's metadata. This only compounds the problem. Data users may discover significant discontinuities (inhomogenities) in the record at this point, but unknowingly, without any documentation of what happened, reach false conclusions regarding an "apparent" (false or artificial) climate change.

The good news is that a complete, detailed station metadata record allows data users to better understand discontinuities (note that the reasons for the discontinuities are shown on the illustration). Metadata information, combined with instrument performance specifications, allows users to adjust (or filter out) artificially-induced climate shifts so that the naturally occurring trends and variability are more accurately detected.

Another problem with the use of unapproved instrumentation is maintenance. Keep in mind we are in the climate monitoring game for the long run. This means network operation for decades and longer. We are responsible for long term maintenance and operation which is much more difficult with unapproved instrumentation that is not supported through the logistics pipeline.

A good example that illustrates the importance of instrument change on climate data is that associated with the implementation of wind shields on precipitation gauges (Figure 17). For example, researchers using precipitation data published in Monthly Climatic Data of the World, one of the primary sources of global climate data, must address the fact that by 1980, 44 percent of all the contiguous U.S. precipitation gages in the published set had shields installed around the orifice of the gage. Prior to 1948, shields were absent.

Recommended Instrumentation-Related Actions:

The following actions are recommended to minimize unwanted discontinuities in the climate record:

1) Use only approved instrumentation whose functional characteristics are understood for published climate stations. To do otherwise, increases the uncertainty in the quality, continuity, and comparability of the climate record.

2) Ensure that all instrumentation changes and the dates thereof, whether a new instrument, replacement or modification of existing stock, are documented in station metadata.

3) Coordinate with your RCSPM and other climate services partners (NCDC, RCCs, SCs) on any related issues.