Detecting environmental changes and trends

Abstract Detecting changes in climate variables is a difficult process that depends on several parameters. These parameters include not only the size of the trend and the accuracy of the measurements, but also the magnitude of variability and the autocorrelation of the observations. The ability to detect a trend can vary from location to location and can vary with height at a single location. These differences are illustrated using observations from the Forecast Systems Laboratory Radiosonde Database to determine the natural variability and autocorrelation of temperature data as a function of altitude. As an example, the results for Topeka, Kansas, a mid-continent, stable monitoring site, indicate that trends of a fixed magnitude are easier to detect in the free troposphere than in either the boundary layer or the lower stratosphere. The free troposphere may be an important region for climate monitoring, especially given general circulation model projections of a nearly vertically-constant warming throughout the tropospheric altitudes. These types of investigations, which use past data to determine the natural variability and memory inherent in environmental parameters, can help identify geographic and altitude regions that may be critical for climate monitoring.