It is difficult to develop a consistent, unbiased long record of the thermal
structure in regions where conventional observations are sparse in space and
time. Here, we make use of two hemispheric datasets. The first is the NCEP-NCAR
reanalysis (Kalnay
et al. 1996; Kistler
et al. 2001), which provides a full suite of atmospheric parameters on
a 2.5° × 2.5° grid from 1948 to the present. There is good coverage
of rawinsonde data along the Arctic coast for assimilation into the reanalysis.
The second is the TOVS Path-P dataset (Francis
and Schweiger 2000), which provides temperature and humidity analyses north
of 60°N on a 100 × 100 km grid
for the period 197998. The reanalysis uses a "frozen" data
assimilation model system over the period of record and hence has temporal
consistency, but includes relatively few observations from the central Arctic.
The TOVS Path-P dataset is based on satellite-measured radiances that provide
relatively good horizontal resolution in temperature and humidity in the central
Arctic, but with uncertainties in calibration. In an overall sense, the reanalysis
is best suited for describing temporal changes and for diagnostic calculations,
since it includes both thermodynamic and kinematic parameters in dynamical
balance, while the TOVS Path-P is best suited for describing details in horizontal
gradients in temperature. The TOVS Path-P dataset depends on measurements from
a series of satellites, and the calibrations that need to be applied to the
data prior to 1986 are unknown. We have adjusted the temperature from the TOVS
Path-P for the purposes of the present analysis; the details of this adjustment
are covered in the appendix.
Our analysis relies principally on the reanalysis dataset. This source is used to illustrate the changes in the climate of the western Arctic over the last half of the twentieth century, and for estimates of terms in the thermodynamic energy budget. The TOVS Path-P dataset is used to compare the details of the structure of the temperature anomalies of the 1990s with those in the 1980s. An important result of our analyses using the two datasets is that the temperature fields in the reanalysis were found to be similar to those in the TOVS Path-P, even though the former has relatively meager data available for the central Arctic. This result indicates that diagnostic calculations from the reanalysis for the Arctic are reasonably robust.
Return to previous section or go to next section