Table 7. Results of extreme value analysis.
thicknesses varying around 0.1 in. and occur-
rence rates for extreme storms of between one
50-yr
and two per year. The results of the extreme
Period of
return period
value analysis for the superstations, based on
Superstation
record
ice load
the simple model results using the weather data
(years)
(in.)
for this storm and historical weather data, are
Watertown
94
0.9 (0.6)
summarized in Table 7. The table lists the num-
Fort Drum
ber of years in the period of record for each
Plattsburgh AFB
superstation and the 50-year return-period ice
Burlington
loads, followed in parentheses by the 50-year
Barre
21
0.6 (0.6)
loads that were calculated prior to this storm.
Concord
136
1.1 (1.0)
The updated 50-year return-period loads in
Pease AFB
Table 7 were used to revise the ice-load map
Portland
currently in draft form for the 1998 edition of
Bangor
ASCE 7 Minimum Design Loads for Buildings
and Other Structures. That revised map is
shown in Fig. 25.
Note that Barre is the only weather station with hourly data at a high elevation in either
Vermont or New Hampshire. However, it cannot represent the wide variety of conditions that
occur over higher terrain in that region. Furthermore, because of the short period of record of the
weather data at Barre, determining extreme ice loads for even relatively short return periods is
very uncertain. In using the data at Barre, we are not only extrapolating 50-year loads from only
20 years of data, we are also doing that extrapolation from the recent past (since 1972) in which
the weather in the United States was mild, with relatively few storms of any sort. If we could use
the data recorded at coop weather stations, many of which have records dating back to 1948, we
could substantially increase our understanding of the severity and geographical distribution of ice
accretion from freezing rain in this difficult region. An empirical approach developed by Dr.
Charles Konrad at the University of North Carolina (in review) identifies days with freezing rain
from the daily precipitation and snowfall records and the daily maximum and minimum tempera-
tures. We could apply our ice-load models to data from those freezing-rain days to get estimates
of ice loads. However, these load estimates would not include the additional ice that may accrete
at elevations high enough to be in low clouds that may form in the cold air layer during freezing-
rain storms. As was indicated for this storm, ice accreted from supercooled clouds may be signif-
icant on structures built on mountaintops. To model the accretion of ice from supercooled clouds,
we would use the cloud layer data from the hourly weather stations, estimates of cloud liquid
water contents and drop sizes, and upper air wind and temperature data at the stations where
upper air data is obtained. In-cloud icing at all times, not just during freezing-rain storms, is
probably the dominant ice-loading mechanism on very tall towers (1000 to 2000 ft tall) and any
structures on mountaintops.
7.3.2 Return period for severe ice storms
The January ice storm was the worst ever in the experience of many people in upstate New
York and northern New England, both in the amount of ice that accreted on trees and structures,
and the extent of the storm. Using the parameters of the GPDs, we can estimate return periods
for the ice loads in this storm. In Table 8 are shown the return periods for ice loads ranging from
42
Back to contents page
Previous Page
