Using daily minimum air-temperature (T-min) data from the state of Illinois
, the dates of spring and fall freezes - and the resulting growing-season l
ength - are examined for trends during the period 1906-1997. Of the station
s in the Daily Historical Climate Network, most show trends toward earlier
spring freezes; however, trends in fall freezes are not consistent over the
station network. Although the time series are highly variable (noisy), res
ults suggest that the growing-season length in Illinois became roughly one
week longer during the 20th century. To examine how changing freeze-date st
atistics relate to changing air-temperature probability distributions, perc
entiles of T-min for moving 10-year periods were analyzed for trends during
the typical times for spring and fall freezes in Illinois (i.e., the month
s of April and October). The lower portion of the April probability distrib
ution shows substantially larger warming (0.5-0.7 degrees C/100 yrs) than t
he upper portion of the distribution (0.2-0.3 degrees C/100 yrs), suggestin
g that although cold events are warming during April, warm events are not w
arming as fast. Conversely, the lower portion of the October probability di
stribution shows modest cooling in T-min (-0.2 degrees C/100 yrs for the 10
th percentile), while middle and upper portions of the distribution show ve
ry large rates of cooling (up to -1.5 degrees C/100 yrs for the 40th-70th p
ercentiles). Analysis of the entire probability distribution provides a mor
e-comprehensive perspective on climatic change than does the traditional fo
cus on central tendency.