Utilizing daily maximum/minimum temperature data for a first-order station with a lengthy available history (Los Angeles, CA Civic Center -1921-2002), Principal Components Analysis(PCA)is explored as a tool for identifying and characterizing extreme-most daily maximum/minimum time-series patterns for a hierarchy of calendar intervals (annual, seasonal, and calendar month), throwing light on the station's history in this regard.

For a given period in question, daily data (29 February data not included) are arranged by year into rectangular arrays and converted to correlation matrices. Principal components, along with loadings and standardized scores, are then calculated for each.

Results showed for the seventeen calendar intervals considered that the first principal component explained between 69.2% (December-January) and 87.9% (July) of the variance, and that first component scores were very highly correlated with climatological mean daily maximum and minimum statistics.

Given this fact, first component loadings (one statistic for each year), which measured the correlations of first component scores with original daily max/min temperatures, could serve as an objective (linear) statistic for identifying years with the most anomalous or "non-climatological" patterns (lowest loading statistic).

In addition to the statistical tabulations, time-series charts are presented for some of the more outstandingly anomalous calendar-period patterns, as ranked by standardized loading statistics. While their contributions were always meager relative to overall variances, for informational purposes a few time-series plots of second-component scores are also included.