watershed boundaries. The database may be
same event is found, all of it is presented for the
accessed via the CRREL web site at http://
user to evaluate.
www.crrel.usace.army.mil. The user interface
Some ice events are well documented, per-
allows for database queries that are displayed in
haps by more than one source, while only sketchy
a manner that allows additional data screening
information exists for others. The most complete
and processing.
documentation consists of gage records (usually
containing quantitative stage and flow informa-
tion) supplemented by other sources (which can
supply either qualitative or quantitative informa-
OVERVIEW OF THE
tion, or both). We can safely assume that ice events
CRREL ICE JAM DATABASE
did occur at the locations and dates noted in the
database, but the converse is not true. That is to
Definition of ice events
Ice events included in the CRREL Ice Jam Data-
say, we cannot assume that no ice events took place
base can be described by the broad definition used
at times or locations not listed, since often an addi-
by the International Association for Hydraulic
tional search of records or interviews with local
Research (IAHR) Working Group on River Ice
residents will reveal ice events that are not in the
Hydraulics (1986): "a stationary accumulation of
database.
fragmented ice or frazil that restricts flow." The
Therefore, the database is best viewed as a col-
database includes ice jams that form in the early
lection of identified ice events, subject to selection
winter during ice cover formation (freezeup
bias introduced by the literature search process
jams), those that form during ice cover breakup
and the types of records examined thus far. For
(breakup jams), and those that contain elements
statistical analysis, we must consider the database
of both. Ice cover formation that results in
to be a biased, limited sample of the entire ice
increased upstream water levels or decreased
event population. The presence of bias does not
downstream water levels is also considered an ice
prevent us from undertaking statistical analyses,
event. Thermally induced expansion of an ice
but the biases must be recognized and taken into
cover that damages the shoreline or a structure,
account.
and aufeis--or thickened surface ice accumulated
through the successive freezing of sheets of
Temporal bias
water, also called naled (Carey 1973, Schohl and
Because about 70% of the database entries
Ettema 1986)--are also included.
originated from the mid-1960s USGS series on the
States (USGS Water-Supply Papers 1671 through
Sources of ice event information
The database draws largely from two sources:
1689), a significant temporal bias was introduced
the United States Geological Survey (USGS)
into the database, as is shown in Figure 2. With-
report series that provides gaging station data col-
out considering the bias introduced by heavy
reliance on a single source, it might be erroneously
of floods (e.g., Green 1964) and the annual USGS
concluded from a search of the database that most
Water-Data Reports (e.g., Toppin et al. 1993). Both
ice events in the United States occurred between
of these contain information on peak stage and dis-
about 1935 and 1965. This bias is accentuated
charge events at USGS gages. About 85% of the ice
when a smaller subset of the database is exam-
events in the database are documented at USGS
ined. For example, in Montana, where data col-
gages from these sources.
lection efforts are adequate for the historical (i.e.,
Other data sources are newspapers, historical
before 1965) and recent (since 1995) periods, but
records, such as town histories and government
poor for the intervening period (Fig. 3).
agency reports, anecdotal reports by local resi-
Further investigation may reveal that what
dents obtained from personal interviews, and
appears to be an obvious temporal bias in the
CRREL files. These sources may provide good
database entries for a particular location may
narrative information about an ice event and the
actually reflect changes in river hydraulic condi-
damage it caused, but quantitative information of
tions that significantly altered the ice regime. An
the type found in USGS sources is often lacking.
example of this is the Israel River in Lancaster,
Although sources vary in reliability, everything
New Hampshire, an ungaged location for which
identified as an ice event is included in the data-
the database includes one ice jam in 1895 and 13
base. Where conflicting information about the
ice jams between 1950 and 1992. One might
3
Previous Page
