Surface Climate and SnowWeather Relationships of the Kuparuk Basin
In discussions of the annual climate cycle, it is common to decompose the
yearly cycle into three-month seasons defined on monthly boundaries rather than
the traditional seasonal definitions based on the solar cycle. We follow the for-
mer course here, using the monthly groupings of Sep.Oct.Nov., Dec.Jan.
Feb, Mar.Apr.May, and Jun.Jul.Aug and further naming these seasons Early
Cold (EC), Deep Cold (DC), Late Cold (LC), and Warm Season (WS), respec-
tively. (Refer to Table 1 for a summary of acronyms.) Our rationale for this
nomenclature is based on the annual temperature cycle and will be discussed
further as that cycle is considered.
Table 1. Acronyms used in this report.
Temperature Data Averaging
Climate data typically show a great deal of temporal variability on scales
ranging from diurnal to interannual to millennial. The five-year period of record
(19941998) discussed here is clearly insufficient to address those time scales
(decadal and upward) relevant to what is heuristically referred to as climate
change. Rather, our intent here is to quantify the current climate along the Arctic
Slope as well as possible with the limits of the data record available to us.
The five-year period of record used here shows a considerable degree of
interannual variability for any given day of year, and the limited number of sam-
ples (at best five, assuming no missing data points) permit extreme short-term
excursions from a mean state to produce highly variable (i.e., noisy) time series
of such things as mean daily temperature when averaged over the period of
record. Seasonal trends, a central focus of this work, are more apparent after
some form of smoothing is applied to the data. Much of the data presented are in
the form of running averages. After experimenting with different averaging peri-
ods, we found that an averaging period of 11 days was sufficient to inhibit the
short-term variability associated with propagating weather disturbances while
permitting the best possible resolution of evolving seasonal trends. Clearly, if the
period of record were several times longer, such filtering would become less