Potter T.D., Colman B.R. (co-chief editors). The handbook of weather, climate, and water: dynamics, climate physical meteorology, weather systems, and measurements
Подождите немного. Документ загружается.
Even if a perfect gage were available, there is still a major problem limiting the
accuracy and introducing uncertainty into gage measurements of the water content of
snow. The problem is wind. Gages protr uding into the air present an effective
obstacle to the wind resulting in a deflection. Lightweight snow crystals are easily
deflected. The result is gage undercatch of precipitation. The degree of undercatch is
a complex function of wind speed, snow crystal type, gage shape, and exposure (see
Fig. 3). In the measurement of rain, gage undercatch is not significant until winds are
very strong. However for snow, even a 5 mph breeze can have a very large effect on
gage undercatch.
One approach to improving gage catch efficiency is the installation of a wind
shield surrounding the gage to reduce the effects of wind-caused undercatches (see
Fig. 4). The most common shield in use in the United States is called an Alter shield,
named in honor of its designer. While the Alter shield clear ly imp roves gage catch
efficiency, it by no means solves the problem. The Nipher shield is a favorite in
Canada, and the results above show why. Unfortunately, this shield does not perform
well in heavy, wet snows common in many regions of the world and does not adapt
to other types and sizes of precipitation gages, thus making it impractical for use
with most precipitation gages in use in the United States. Currently, only a fraction
of the U.S. precipitation gages are equipped with wind shields.
The World Meteorological Organization (WMO) has been diligently investigating
the challenge of measuring solid precipitation. An extensive international study was
completed during the 1990s thoroughly investigated the performance characteristics
of a variety of gages and wind shields used in snowy regions of the world (Goodison
and Metcalf, 1992). Many consider the Double Fence Intercomparison Reference
(DFIR) to produce the most representative gage catch under a full range and snow-
Figure 3 Gage catch efficiency in percent as a function of wind speed at the top of the gage
for unshielded, Alter-shielded and Nipher-shielded gages (after Goodison, 1978).
936
CHALLENGE OF SNOW MEASUREMENTS
Figure 4 Alter wind shield on a tower-mounted precipitation gage (from Doesken and
Judson, 1997).
4 PROCEDURES FOR MEASURING SNOWFALL 937
fall and win d conditions (Yang et al., 1993). Unfortunately, the size (12-m diameter)
alone makes this wind shield too bulky and expensive for most common weather
stations. No simple solution exists, and few countries show a willingness to change
their long-standing practices. But there is no excuse to ignore the problem of gage
undercatch since it is now well documented.
A practical approach to the gage catch problem is to be as wise as possible when
first deploying weather stations in order to achieve the best exposure for gages. The
ideal exposure for a precipitation gage is a delicate compromise between an open
and unobstructed location and a protected site where the winds in the vicinity of the
gage are as low as possible during precipitation events. The center of a small clearing
in a forest or an open backyard in a suburban neighborhood are examples of good
sites. The closer to the ground the gage is, the lower the winds will be—a plus for
improving gage catch. But the gage must also be high enough to always be above the
deepest snow. Rooftop exposures are not recommended because of the enhanced
wind problems and the potential for building- induced updrafts that will further
reduce gage catch.
When gage undercatch is apparent, observers are encouraged to take a secondary
observation by finding a location where the accumulation of fresh snow approxi-
mates the area average. A core sample is then taken, and this core is melted or
weighed in order to determine its water content. Unless an unrepresentative sample
is taken or melting has reduced the water in the remaining layer of fresh snow, the
water content of a core often exceeds that of the gage. In practice taking supple-
mental core samples is seldom done to check for gage undercatch, but it could
improve data quality consid erably.
Snowfall
The traditional measurement of snowfall requires only a measurement stick (ruler)
and a bit of experience. While this may be the simplest meteorological measurement,
in practice, it may also be the most inconsistent. The public interest in snowfall is
always great, but the measurements are often more qualitative than people realize.
The inconsistency is a direct result of the dynamic nature of fresh snow. It lands
unevenly, melts unevenly, moves with the wind, and settles with time. The location
where measurements are taken, the time of day, the time interval between measure-
ments, the length of time since the snowfall ended, the temperature, the cloudiness,
and even the humidity affect snow measurem ents. The imaginary but realistic exam-
ple from Snowyville dramatized these points.
For climatological and business applications, for spatial mapping, and for station-
to-station comparisons, the best funct ional definition of snowfall is ‘‘the greatest
observed accumulation of fresh snow since the previous day prior to melting,
settling, sublimation, or redistribution.’’ The typical observer may only go out to
measure snowfall once per day at a designated time. Depending on weather condi-
tions and timing, that may or may not coincide with the time of g reatest accumula-
tion. If there had clearly been 4 in. of fresh snow on the ground early in the day but
only 1 in. still remains at the scheduled observation, did it snow 4 in. or 1 in.? Four
938 CHALLENGE OF SNOW MEASUREMENTS
inches is obviously the better answer, but if the observer were not there to see it, he
or she wouldn’t know for sure. Ideally, the observer is available to continuously
watch the snow accumulate, note the greatest accumulation, and then note the
settling and melting that occurs later. But, in reality, this may not be the case,
since much of the historic snowfall data in the United States has come from the
National Weather Service Cooperative Observer Program, many of whom are volun-
teers who can take only one observation each day (National Research Council,
1998).
Because of the nature of fresh snow, perfectly consistent measurements may not
be possible. However, there are several simple steps that lead to measurements that
achieve a high degree of consistency.
The location for taking measurements is critical. An unobstructed yet relatively
protected location (such as a forest clearing or open backyard away from
buildings, trees, and fences) is best since the goal is to measure where snow
accumulation is as uniform and undrifted as possible and represents the
average snow accumulation of the area.
The use of a snowboard (square or rectangular flat, white surface positioned on
the ground and repositioned daily on the top of the existing snow surface) for
measuring the accumulation of new snowfall greatly helps to achieve consis-
tency by providing a smooth, solid surface on which to measure and from
which core samples can be taken (Doesken and Judson, 1997).
When snow is falling, observers should periodically check the accumulation of
new snow on the snowboard and note the greatest amount before settling or
melting begin to reduce the depth of fresh snow. The greatest accumulation will
typically occur just before the snowfall diminishes or changes to rain.
Observers should only clear the new snow at the scheduled observation time
and then reposition the snowboard on the top of the new snow surface.
To account for blowing and drifting snow and the resulting uneven accumula-
tion patterns, observers should assess the representativeness of the snowboard
measurement by taking an average of several measurem ents in the surrounding
environment, making sure to include only that snow that has fallen since the
previous observation. If the snowboard measurement is found to not
be representative, either because it has partially or totally blown clear, because
drifts have formed in its immediate vicinity, or because snow melted on
the snowboard but not on most ground surfaces, the reported snowfall should
be the average of as many readings as are needed to obtain an appropriate
average over an area including both moderate drifts and moderate clearings.
With the creation and expansion of airport weather stations from the 1930s
through the 1950s, came a new emphasis on weather forecasting for air transporta-
tion and safety. Hourly weather observations were initiated that included manual
measurements of precipitation type and intensity, visibility, and many other weather
elements. These became the foundation of the surface airways weather observations
4 PROCEDURES FOR MEASURING SNOWFALL 939
(reference) and provided more detai ls about snowfall characteristics than had been
previously available. Every hour a complete report of weather conditions was used in
a consistent manner from a large number of stations. Conditions were also monitored
between hourly reports, and any significant changes were reported in the form of
‘‘special’’ observations. During snowfalls, depths were measured every hour, and
special remarks were appended to observations whenever snow depth incre ased by
an inch or more. These ‘‘SNOINCR’’ remarks always caught the attention of meteor-
ologists since they signaled a significant storm in progress. But this also introduced a
new complexity into the observation of snow. Instead of observing snowfall once
daily, some weather stations reported more frequently. The instructions to airways
observers stated that snowfall was to be measured and reported every 6 h. The daily
snowfall was then the sum of four 6-h totals. Some weather stations then used the
seemingly appropriate procedure to measure and clear their snowboards every hour
and add these hourly increments into a 6-h and daily totals.
For some applications, short-interval measurements are extremely useful.
However, for climatological applications, snowfall totals derived from short intervals
are not the same as measurements taken once daily. For rainfall, a daily total can be
obtained by summing short-interval measurem ents. However, for snowfall, the sum
of accumulations for short increments often exceeds the observable accumulation for
that period. To demonstrate this, volunteer snow observers were recruited from
several parts of the country and measured snowfall for several winters at several
locations in the United States. Each observer deployed several snowboards and
measured snow accumulations on each board. One board was cleared each hour,
one every 3 h, one every 6 h, and every 12 h, and finally one was only measured and
cleared every 24 h (see Fig. 5). While results vary from storm to storm, it was very
clear that snowfall totals are consistently and significant ly higher based on short-
interval measurements (Doesken and McKee, 2000). Based on 28 events where
measurements taken every 6 h throughout the storms were summed and compa red
to measurements taken once every 24 h, the 6-h readings summed to 164.4 in., 19%
greater than the 138.4-in. total from the once-daily observations. When hourly read-
ings were summed and compared to once-daily measurements, the total was 30%
greater. What this means to the user of snowfall data is that two stations, side by side,
measuring the same snow amounts at different time intervals may report greatly
different values.
In an effort to standardize procedures among different types of weather stations,
the Nationa l Weather Service issued revised snow measurement guidelines in 1996
(U.S. Dept. of Commerce, 1996a). These guidelines stated that observers should
measure snowfall at least once per day but could measure and clear their snowboards
as often as but no more frequently than once every 6 h (consistent with long-standing
airways instructions). These guidelines were promptly put to the test when an
extreme lake-effect snowstorm in January, 1997, produced a reported 77 in. of snow-
fall in 24 h. Subsequent investigations by the U.S. Climate Extremes Committee
(U.S. Dept. of Commerce, 1997) found that this total was the sum of six observa-
tions, some of which were for intervals less than 6 h. While the individual measure-
ments were taken carefully, the summation did not conform to the national
940 CHALLENGE OF SNOW MEASUREMENTS
guidelines and hence could not be recognized as a new record 24-h snowfall for the
United States.
The main conclusion here is that where station-to-station comparability and
long-term data continuity are the goals, stations must measure in a consistent
manner. There may be justification for different observation times and increments,
but data from incompatible observation methods should not be interchanged or
compared. Many of the data sets in common use today are not fully consistent, so
end users may have the responsibility to determine the compatibility and compar-
ability of various station data.
Snow Depth
The measurement of snow depth is not subject to the qualitative and definitional
issues that plague the measurement of snowfall. The equipment is again very basic.
For areas with shallow or intermittent snowpacks, a sturdy measurement stick is
normally carried by an observer to the point(s) of observation and inserted vertically
through the entire layer of new and old snow down to ground level. In areas of deep
and more continuous snow cover, a fixed snow stake, round or square, is often
used—clearly marked in whole inches or in centimeters and permanently installed
in a representative location and read ‘‘remotely’’ by an observer standing at a
convenient vantage point so that the snow near the snow stake is not disturbed by
foot traffic.
Figure 5 Snowfall comparisons for different measurement intervals. Values were normalized
by dividing the total accumulated snowfall in each category by the number of events sampled
(Doesken and McKee, 2000).
4 PROCEDURES FOR MEASURING SNOWFALL 941
The key to useful comparable measurements of snow depth is identifying and
maintaining representative locations for taking measurements. Blowing and drifting
are inevitable challenges. Uneven melting adds further compl ications. For uneven
snow accum ulation, measurements should be taken from several locations propor-
tionally representing both the deeper and shallower areas. Since most traditional
weather stations are long distances apart, it is imperative that each measurement
represents the predominant conditions in the vicinity of each stat ion.
Snow Water Equivalent (SWE)
This is an extremely important measurement for hydrologic applications. Both flood
and overall water supply forecasting rely on accurate measurements of SWE from as
many locations as possible to represent the spatial patterns of snow water content
that will contribute to subsequent runoff.
The measurement of SWE is taken by only a fraction of National Weather Service
stations and is usually accomplished by taking full core samples of total snow on the
ground using the 8-in. diameter precipitation gage overflow can. Under deep snow
conditions, the melting of snow cores requires considerable amounts of warm water
and is tedious and time consuming for observers. Some stations are equipped with
special scales that make it relatively easy to take a core sample and immediately
estimate the SWE from the weight of the sample. When snow depths exceed 2 ft, the
NWS overflow can is inadequate for effectively coring the snowpack.
The Natural Resources Conservation Service and other water resources organiza-
tions have a long history of measuring SWE in high snow accumulation areas.
Records date back to the 1930s throughout the western mountains with even
longer records from a few sites. A wealth of literature, much of it informal and
non-peer-reviewed, exists detailing the results of years of experimentation and field
testing of devices and techniques to measure snow water in the deep snow regions of
North America. Proceedings of the Western Snow Conference and Eastern Snow
Conference are great sources for information on the evolution of snow measure-
ments and related research.
Over time, two devices have emerged as the standards for measuring snow water
equivalent in deep snow accumulation regions. The Federal Snow Sampler, a porta-
ble set of tubes, handle, and a cutter to cleanly penetrate deep snow and ice layers,
has been in use for many years (see Fig. 6). Core samples of the snowpack are taken
with this instrument, and the sample is then weighed in situ with specially calibrated
scales to determine the snow water equivalent of the core. To account for the
nonuniform accumulation of snow, several cores are taken at each site. A measure-
ment site is called a ‘‘snow course.’’ The final SWE value for a snow course is the
average of a set of measurements across the snow course. Each time a snow course is
read, core measurements are taken at the same approximate set of individual points.
Snow courses have been traditionally read once or twice a month beginning in
midwinter and continuing into the spring until all annual snow has melted.
The second instrument, in wide use since the late 1970s, is called a snow pillow
(see Fig. 7). This is essentially a scale bu ilt at ground level to measure the weight of
942 CHALLENGE OF SNOW MEASUREMENTS
the snowpack as it accumulates and melts. Snow pillows were developed to provide
remote measurements of SWE without requiring the huge expense of time and effort
involved in sending teams of scientists and hydrologic technicians into the back
country every mont h.
As with snowfall and snow depth, the utility and comparability of the observa-
tions are only as good as the repres entativeness of the measurement location and the
averaging process. Snow pillow measurements have their own set of challenges. For
example, ‘‘bridging’’ and other nonuniformities in load-bearing characteristics
within the snowpack can compromise the accuracy of snow pillow meas urements.
5 CONTRIBUTION OF TECHNOLOGY TO SNOW MEASUREMENTS
Many aspects of the measurement of snow continue to use only the simplest of
instrumentation in the hands of trained and experienced observers. Increasingly,
Figure 6 Federal snow sampler consists of a cutter, tubes, and scales for measuring snow
water equivalent in moderate and deep snowpacks (from Doesken and Judson, 1997).
5 CONTRIBUTION OF TECHNOLOGY TO SNOW MEASUREMENTS 943
Figure 7 U.S. Department of Agriculture Natural Resources Conservation Service SNOTEL
(Snow Telemetry) site in the western United States including snow pillow (in foreground), a
shielded standpipe precipitation storage gage (left background), and radio telemetry equip-
ment (from Doesken and Judson, 1997).
944
CHALLENGE OF SNOW MEASUREMENTS
scientists and practitioners turn to new techno logies and new measurement techni-
ques to acquire the data needed to better understand and apply our knowledge of
snow to improve forecasts. Remote sensing is providing data sets showing greater
areal coverage and finer resolution information on the spatial distribution of snow.
This leads to improved models of snowmelt processes and water supplies. At the
same time, improved physical models have pointed out the deficien cies in surface
data motivating greater efforts to employ technology to gather more and better data
about snow.
In these few pages we cannot do justice to all the technological innovations that
are helping improve the measurements of snow and its properties. Rather, we will
touch on a few technologies and instruments currently in use. Some of these ideas
have been around for many years, but the actual implemention of operational
measurement systems is made possible by the remarkable expansion in low-cost
computing power in recent years.
Visual Satellite Imagery
From the time the first satellites orbited Earth, it was apparent that snow cover was
easily detectable from space. Depth and water content are not easily estimated from
reflected visi ble light, but the extent of snow cover can be readily evaluated from
space. The primary limitation is cloud cover, which makes it impossible to see the
snow cover below using only visible wavelengths.
Meteorological Radar
Radar refers to the remote-sensing technique of transmitting microwaves of a
specified wavelength and receiving, processing, and displaying that portion of the
transmitted energy reflected back to the transceiver. Rain and mixed-phase precipita-
tion reflect microwave energy relatively effectively. Snow crystals can also be
detected but, depending on crystal structure and temperature, are not detected as
well as ‘‘wetter’’ forms of precipitation. The U.S. National Weather Service routinely
uses radar to monitor the development, movement, and intensity of snowfall.
Improvements in radar realized by the NWS WSR-88D allow estimates of snowfall
intensity (Holyrod, 1999) and potential accumulation rates. However, ground truth
data remain essential for radar cali bration. Detection efficiency varies greatly with
distance from the radar, cloud height, and other factors. Still, this technology affords
wonderful opportunities for studying snowfall processes in action.
Passive Microwave Remote Sensi ng
Energy in the form of microwaves is continuously emitted from Earth’s surface.
Snow crystals within the snowpack scatter and attenuate these microwaves. Aircraft
or satellites overhead can detect these emissions a nd through a series of approxima-
tions and algorithms can estimate regional snow cover, water content, and appro-
ximate depth. There are various limitations of this technology. For example, shallow
5 CONTRIBUTION OF TECHNOLOGY TO SNOW MEASUREMENTS 945