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.
the late 1970s, brought these changes to light even more. Today, these systems and
ASOS provide us with very detailed data.
Figure 7 shows the passage of a cold front as recorded by three noncollocated
hygrothermometers at Sterling, Virginia. Note the drastic changes in temperature
over a very short period of time. All three systems logged temperature data quality
errors and the sensors were flagged inoperational as a result of this temperature
drop.
Based on temperature rate data from the HWG’s studies over 3 years, the HWG
recommended that the temperature and dew-point data quality algorithm be changed.
The change was implemented in a subsequent firmware load in ASOS. Now, the
quality failure does not occur until the last reading from the temperature or dew-
point sensor varies from the previous reading by more than 5.6
C (10
F).
7 RELOCATION BIAS
In addition to the instrumental biases discussed so far, another factor was to affect
temperature readings at most ASOS sites. ASOS sensors are typically located near
the touchdown zone of the primary designated instrument runway and on occasion at
a center field location. These exposures for temperatures at airports were generally
better where the H083s were not already remotely located from the airport office.
Some earlier station temperatures were taken on roofs or near parking lots. See
Figure 8. Even where the H083 was already remotely located, the ASOS sensor
was sometimes located a mile or two away from the previous location. There were
Figure 7 One-minute temperature data showing cold front passage from three noncollocated
ASOS hygrothermometers at Sterling, Virginia.
756
CONSEQUENCE OF INSTRUMENT AND SITING CHANGES
sometimes significant exposure differences over these distances. Runways are flat
but the areas around them are not always so. Suitable exposures for meteorological
instruments are not always available because of the proximity to taxiways and
instrument intrusion into safety regulated airspace. See Figure 9. As a result, sensor s
are sometimes located in swales where the temperatures are subject to cold air
drainage. Regardless of the reason for the change in location, the differences
needed to be accounted for.
Figure 8 Sensors poorly exposed in a parking lot. See ftp site for color image.
Figure 9 Sensors located too close to a jet takeoff area. Windscreen around the rain gauge is
damaged from jet blast. Blast also causes temperature to rise and wind gusts to increase. See
ftp site for color image.
7 RELOCATION BIAS 757
8 DATA CONTINUITY STUDIES
The basic dilemma was that even if the 1088 were perfect, it was now known that the
H083 was not. Thus, at all sites, an account must be made for that bias given by the
uncertain mixture of H083 bias and removal factors. This could only be done by
leaving the H083 in place and comparing it to the modified and much improved
1088. It was obvious that the simple sling psychromet er would not suffice. If the
sling is perfect and the 1088 is perfect and they are collocated, nothing is proved. For
true data continuity, there was a need to develop for each station its unique bias
signature.
Unfortunately, because of resource limitation and schedules, only a subset of the
ASOS sites could be studied. The H083s had to be returned to the manufacturer and
modified so they would be the sensor equivalent of the improved 1088. Only a
limited numbe r of the original H083s could be left in place with the ASOS 1088
version. With this restriction, the sites selected for a climate data continuity project
took into account as many different climatological regimes as practical. This
research was conducted by the Department of Atmospheric Science at Colorado
State University in Fort Collins, Colorado, and was reported by McKee et al.
(1996, 199 7). Studies of data continuity began with the 15 sites given in Table 1
for data collected from June, 1994, through August, 1995. All sites had received all
of the modifications described above by June, 1994. These 15 locations retained the
H083 during this period so the maximum and minimum daily temperatures were
available.
TABLE 1 Climate Data Continuity Study
(CDCP) Comparison Sites for Daily
Maximum and Minimum Temperatures
Number Site ID Station Name
1 AMA Amarillo, TX
2 AST Astoria, OR
3 BRO Brownsville, TX
4 BTR Baton Rouge, LA
5 COS Colorado Springs, CO
6 DDC Dodge City, KS
7 GLD Goodland, KS
8 GRI Grand Island, NE
9 ICT Wichita, KS
10 LNK Lincoln, NE
11 OKC Oklahoma City, OK
12 PWM
a
Portland, ME
13 SYR Syracuse, NY
14 TOP Topeka, KS
15 TUL Tulsa, OK
a
Station commissioned in August, 1994.
758 CONSEQUENCE OF INSTRUMENT AND SITING CHANGES
The goal of the analysis was to understand the physical causes of the observed
temperature difference between the two hygrothermometers. An initial step was to
confirm the absolute accuracy of the ASOS hygrothermometer. The NWS examined
several of the ASOS instruments at Sterling, Virginia, and found a range of 0.20
F.
Three additional comparisons were made in the field at Colorado Springs, Colorado
(COS), Oklahoma City, Oklahoma (OKC), and Tulsa, Oklahoma (TUL), which
showed a range of 0.30
F relative to a field standard (RM Young), which had
been calibrated against a secondary standard at Sterling, Virginia. Thus, ASOS does
not have a temperature bias. The model used to assess the temperature differences
had the analytic form:
DT ¼ DT
i
þ DT
l
þ DT
s
ð1Þ
where DT is the observed temperature difference of ASOS–H083, and the subscripts
are i (instr uments), l (local effect of location), and s (solar heating effect). All of the
separations between the hygrothermometers were less than 1 mile and the hygro-
thermometers were collocated at four sites, which would have no local effect, by
definition. Local effects due to site location could be different for day and night,
various weather systems, and seasons. Solar heating was included since it was
known that the H083 had a solar heating problem.
Results of the analyses are given in Table 2 for the 15-month period. Observed
differences in the maxi mum and minimum temperatures (labeled M
x
and M
n
) show
ASOS to be cooler by 1.17
F(M
x
) and 0.86
F(M
n
) averaged over all sites. Two
methods were used to determine the instrument bias of the H083 (DT
i
) assuming
ASOS had no bias. The first was to examine comparison s when wind speeds were
high enough to reduce local effects resulting in a narrow frequency distribution of
DT. A speed of nearly 15 miles per hour was required, but there were not enough
observations remaining to be meaningful at all sites. The second approach was to
have an ASOS reported overcast sky at night, which meant cloud base was 12 ,000 ft
or lower. Downward infrared radiation from an overcast low cloud would reduce
horizontal temperature differences. In fact the frequency distribution of observed
temperature differences was very narrow, which indicated the local effects were
reduced. The instrument bias determined from this analysis is given in Table 2 in
the column marked DT
i
. The average DT
i
was 0.57
F, and the range was quite large
from 0.16 to 1.06
F. The ASOS at Lincoln, Nebraska, was moved midway in our
data collection period. Two LNK sites are thus included. LNK-1was the first location
and the ASOS instrument was moved to LNK-2, which was essentially collocated
with the H083. The cloud analysis yielded the same estimate of DT
i
from the two
locations. Next the H083 bias was subtracted from the M
x
and M
n
observations, and
the remainder for the M
n
is DT
l
at night and for the M
x
is a combination of DT
l
in the
daytime plus the solar effect. Notice the resulting DT
l
from the M
x
has a wide range
of 0.56 to 1.10
F. This shows that local effects can be quite important for distances
less than 1 mile. For the daytime local effect plus solar effects, several of the sit es
showed a magnitude of 1
F or larger indicating a likely strong solar effect, which
was previously established as a problem for the H083. As a consequence of the
8 DATA CONTINU ITY STUDIES 759
differences for M
x
and M
n
, the ASOS observed diurnal range also decreased in
comparison to the H083 observations. A summary of this data continuity analysis
is that ASOS is a better instrument than the H083 with much less bias, less solar
influence, and a new recognition that local effects are important.
9 INTERFACE WITH OUTSIDE GROUPS
The ongoing work on temperature measurement was done in coordination with
groups such as the National Academy of Science, the National Center for Atmos-
pheric Research, a few universities, and a number of state climato logists. What
became apparent was that the measurement of temperature is more complex than
originally thought. For inst ance, one reason electronic thermometers report higher
temperatures is because of their quicker response time. Liquid-in-glass ther mometers
miss quick temperature fluctuations because of their slower response. Cotton Region
temperature shelters have an even larger bias than the H083 under the low-wind,
high-solar-load conditions. The climatological community is looking at these
problems, and the NWS will be working with them on documenting the biases
TABLE 2 ASOS–CONV (
F) June, 1994 Through August, 1995
Bias Removed
Station M
x
M
n
DT
i
M
x
(DT
s
þ DT
l
)
M
n
(DT
l
)
ABOS–CONV
Diurnal Range
AMA 0.76 0.59 0.35 0.41 0.24 0.17
AST 0.59 0.03 0.28 0.31 0.31 0.62
BRO 0.93 0.32 0.45 0.48 0.13 0.61
BTR 1.96 1.19 0.89 1.07 0.30 0.77
COS
a
1.38 0.41 0.16 1.22 0.25 0.97
DDC 0.48 0.91 0.61 0.13 0.30 0.43
GLD 1.16 1.19 0.21 0.95 0.98 0.03
GRI 1.25 0.69 0.67 0.58 0.02 0.56
ICT
a
0.79 0.27 0.29 0.50 0.02 0.52
LNK-1 2.14 2.00 0.96 1.18 1.04 0.14
LNK-2
a
2.38 1.03 0.96 1.42 0.07 1.35
OKC 0.64 2.03 0.93 0.29 1.10 1.39
PWM 0.78 1.18 0.47 0.31 0.71 0.40
SYR
a
0.80 0.42 0.29 0.51 0.13 0.38
TOP 0.48 0.06 0.50 0.02 0.56 0.54
TUL 2.17 1.55 1.06 1.11 0.49 0.62
Average 1.17 0.86 0.57
b
0.60 0.29 0.31
a
Co-located sites.
b
Value has LNK twice.
760 CONSEQUENCE OF INSTRUMENT AND SITING CHANGES
among the various sensors, climatological regimes, sensor locations, and by certain
weather parameters.
Without this cooperation, unexplainable temperature shifts would be a murky
blend of a number of biase s caused by different factors, and certain data would
go down in the books with the comment: ‘‘Data quality=continuity uncertain.’’
REFERENCES
Federal Aviation Administration and Department of Commerce (1965). Aviation Weather for
Pilots and Flight Operations Personnel, Washington, DC, U.S. Government Printing Office.
Flanders, A. F., and J. W. Schiesl (1972). Hydrologic data collection via geostationary satellite,
IEEE Trans. Geosci. Electron. GE-10(1).
McKee, T. B., N. J. Doesken, and J. Kleist (1996). Climate Data Continuity with ASOS (Report
for the period September 1994–March 1996), Climatology Report No. 96-1, Colorado
Climate Center, Atmospheric Science Department, Fort Collins, CO, Colorado State
University, March.
McKee, T. B., N. J. Doesken, J. Kleist, and N. L. Canfield (1997). Climate Data Continuity
with ASOS—Temperature, Preprints, 13th International Conference on IIPS, AMS,
2–7 February, Long Beach, CA, pp. 70–73.
National Weather Service (1992). ASOS (Automated Surface Observing System) User’s Guide.
National Oceanic and Atmospheric Administration.
Schiesl, J. W. (1976). Automatic Hydrologic Observing System. Paper presented at the
International Seminar on Organization and Operation of Hydrologic Services, July, 1976.
Ottawa.
Schrumpf, A. D., and T. B. McKee (1996). Temperature Data Continuity with the Automated
Surface Observing System, Atmospheric Science Paper No. 616, Climatology Report No.
96-2, Colorado State University, Fort Collins, CO.
U.S. Army Quartermaster Research and Development Command (1957). Environmental
Protection Research Division, Weather Extremes around the World, Natick, MA, May.
U.S. Depar tment of Commerce, National Oceanic and Atmospheric Administration (1994).
Federal Standard for Siting Meteorological Sensors at Airports, FCM-S4-1994.
REFERENCES 761
CHAPTER 38
COMMERCIAL RESPONSE TO
MEASUREMENT NEEDS:
DEVELOPMENT OF THE WIND
MONITOR SERIES OF WIND
SENSORS
ROBERT YOUNG
The Wind Monitor wind sensor is an example of a commercial product development
in response to customer measurement needs. Requirements of a single government
agency led to the eventual development of an entire series of wind speed and
direction sensors. The current models are the result of input from a multitude of
agencies and end users whose application requirements vary dramatically—from
sensitivity and responsiveness to survival in high winds, snow, and ice as well as
desert heat and blowing sand—from ease of mounting and maintenance to durability
and long-term performance—from simple analog output signals to polled digital
serial data streams.
1 BACKGROUND
The Wind Monitor was created in 1979 as a small development project to try to
satisfy the requirements of the National Data Buoy Center (NDBC) at Stennis Space
Center, Mississippi. Now, approximately 23 years later, seven standard models plus
several additional special models of the Wind Monitor are being manufactured to
serve different customer requirements. Over 50,000 units have been produced (as of
February, 2002) and are in worldwide service in more than 85 countries as well as
Handbook of Weather, Climate, and Water: Dynamics, Climate, Physical Meteorology, Weather Systems,
and Measurements, Edited by Thomas D. Potter and Bradley R. Colman.
ISBN 0-471-21490-6 # 2003 John Wiley & Sons, Inc.
763
the Arctic and Antarctic regions and ships and buoys at sea. New applications are
still being discovered.
From our company’s early days we were advocates of the use of helicoid
propellers for measurement of wind speed. Several products, which origi nally
were categorized as ‘‘sensitive wind instruments’’ combined a helicoid propeller
with a wind vane to provide measurements of both wind speed and wind direction
in a single instrument. We manufactured propellers from very lightweight polysty-
rene foam for sensitivity and also from injection-molded polypropylene for greater
durability. The typical wind sensor matched one of these propellers with a wind vane
of comparable sensitivity. Typical fabrication materials were aluminum castings,
machined aluminum and stainless steel, standard metal and plastic component
parts, with stainless steel fasteners used for assembly. The wind speed transducer
was typically a miniature tachometer generator, which required slip rings and
brushes, and the typical wind direction transduc er was a wirewound or conductive
plastic potentiometer.
2 DEVELOPMENT
In the fall of 1979 we entered into a development contract with the National Data
Buoy Center (NDBC; then referred to as the Nationa l Data Buoy Office) to design a
propeller-vane-type sensor that would address its future needs for deployment on
offshore buoys and remote stations. At the time the operational wind sensors for
these applications were the J-Tec, which measured wind speed by means of a vortex
shedding technique (utilizing a detector that counted the number of vortices shed
from an obstruction in the tail assembly), and the Bendix Aerovane, a rugged sensor
that utilized a propeller and vane combinati on. The main problem encountered with
these sensor s was the difficulty in mounting at sea due to their size and weight plus
the need to install mounting bolts between the base and the mounting fixture on the
buoy. Many of the buoys were a long distance offshore, which sometimes meant
servicing the sensors in difficult weather conditions.
The design goals of the development contract were: small size (maximum 36 cm
overall height; 46 cm overall length), light weight (2.5 kg maximum; 1.5 kg desired),
simple mounting allowing one-hand replacement, corrosion resistant, no electrical
slip rings, three- or four-blade helicoid propeller (maximum diameter 20 cm), work-
ing range 0 to 60 m=s and 0 to 360
, cost reduction, and reliability (18-month
MTBF). The contract called for delivery of three prototype sensors for testing and
evaluation.
The first prototype, which was delivered to NDBC in November, 1980, was
machined aluminum with a detachable tail assembly (Fig. 1). It was intentionally
designed for testing flexibility. Many different size and shape tail fins were tested to
determine the opti mum trade-off between overall size and dynamic performance.
The wind speed sensor was an injection-molded polypropylene four-blade propeller,
18-cm diameter by 30-cm pitch, which had been previously developed. The wind
speed transducer was a magnetically operated Hall effect switch, actuated by a two-
764 COMMERCIAL RESPONSE TO MEASUREMENT NEEDS
pole magnet on the propeller shaft, which provided an output of one pulse per
propeller revolution. The wind direction transducer was a conductive plastic 1 kO,
352
function angle, potentiometer which was mounted concentrically in the sensor
housing and coupled to the movable vane above. Replaceable ball bearings and
sleeve bearings were provided to conduct torque and speed tests as well as to
assess long-term reliability. Early wind tunnel testing showed that sleeve bearings
would overheat at the higher wind speeds and would not be suitable. The most
difficult part of the design was actuation of the wind speed transducer that, to
eliminate the need for slip rings and brushes, had to be mounted on the nonrotating
inner part of the main body. Since the potentiometer was also located concentrically
in the same area and being actuated by a coupling from the vane assembly, it was
necessary to mount the Hall switch slightly off-center to clear the coupling. To
accomplish this, the Hall switch was sandwiched between two soft iron circular
plates. The circular magnet attached to the propeller shaft would then always be
the same distance from the edge of the plates and the magnetic flux would be
directed through the Hall switch as the magnet rotated.
Figure 1 First prototype wind sensor, November, 1980.
2 DEVELOPMENT 765