|Local forecast by
DATA ACQUISITION, EQUIPMENT,
Automated and manual data networks and equipment had performance problems during
this flood event; sporadic equipment problems and many critical outages of automatic
and manual gage networks were noted during this event that impacted the acquisition of
data. Sporadic equipment problems and some critical outages of automatic and manual
gage networks were noted during this event which did impact the acquisition of data.
Staffing at each office seemed adequate to meet the challenges of this event. In most
cases, overtime was used to keep the Service Hydrologist (SH), Science Operations
Officer (SOO), Warning Coordination Meteorologist (WCM), Meteorologist-in-Charge
(MIC), and one or two additional operational members on duty throughout the event.
Most offices acted quickly in developing a contingency staffing plan to ensure the event
would be well covered. Most offices had a member of the management team available
throughout the event to answer questions from the media and public safety officials,
conduct interviews, and fill in as needed, operationally.
3.2 DOPPLER WEATHER SURVEILLANCE RADAR (WSR-88D)
All offices are equipped with a WSR-88D which provides an areal assessment of
rainfall amounts based on the returned power processed through a series of
hydrometeorological algorithms. The radars at Binghamton, New York; State College,
Pennsylvania; and Philadelphia, Pennsylvania, did a good job of providing reasonably
accurate rainfall amounts.
Most forecasters who worked this event were pleased with the precipitation products
generated by the WSR-88D. While rainfall estimates were not perfect, they provided a
good representation of the rainfall patterns when compared to rain gage measurements.
Rainfall amounts provided by these radars, along with critical rainfall reports from
cooperative observers and amateur radio people, were crucial to Quantitative
Precipitation Forecast (QPF) revisions late Friday morning, January 19, and to the
issuance of early Flood and Flash Flood Warnings. The revised QPF forecast provided
a more accurate picture of the rainfall and snowmelt situation. This action allowed the
Middle Atlantic RFC (MARFC) to revise their river forecasts for more accurate stages.
WSR-88D rainfall amounts were also used to compose Flood/Flash Flood Warnings
and provide longer lead times than would otherwise have been possible.
WSR-88D precipitation estimates at longer ranges were less accurate, but higher values
could be inferred along the defined linear area observed. Bright-band contamination
began to occur by Friday afternoon, January 19, in the colder air behind the cold front
creating erroneously large totals. The accuracy of radar estimated rainfall information
at Baltimore, Maryland/Washington, DC, was hampered by beam blockage to the west.
Thus, critical precipitation estimates were not available across several counties which
sustained serious river and flash flooding.
At State College, Pennsylvania, a brief problem occurred shortly after midnight on
Friday, January 19, when the WSR-88D wideband connection between the Radar Data
Acquisition (RDA) antenna site and the Radar Product Generator (RPG) failed at 1
a.m. EST. The forecaster tried, unsuccessfully, to reestablish the wideband connection
using the methods listed in the operations handbook. After consulting with the
Electronic Systems Analyst (ESA), the decision was made to switch the unit to
auxiliary power and then back to commercial power. This completely reset the RDA,
and the wideband connection was reestablished. A second similar outage occurred at
7 a.m. EST as high winds, lightning, and heavy rains crossed the area. High winds
may have also caused the initial outage.
The Ohio RFC (OHRFC) and MARFC do not have the capability to mosaic data from
multiple WSR-88D sites within their forecast area but must call each WSR-88D
individually. This proved to be rather time consuming in a fast-breaking event such as
3.3 RIVER AND RAIN GAGE NETWORKS
Most automated river gages are owned and maintained by the U.S. Geological Survey
(USGS). These gages utilize several ways of communicating information to systems
that process these data. The most common communications method is through Data
Collection Platforms (DCP), electronic devices connected to the data sensing units
which report through a geosynchronous satellite communications system at set times
during a 24-hour period, usually at 3-, 4-, or 6-hour intervals. Some DCPs are capable
of reporting data at random times. The satellite tracking station that collects DCP
information is located in Wallops Island, Virginia. From there, data are transmitted to
the National Weather Service (NWS) Telecommunication Gateway and the NOAA
Central Computer Facility (NCCF) where they are ingested into the
Hydrometeorological Automated Data System (HADS) then sent to the Automation of
Field Operations and Services (AFOS). Through AFOS, all NWS offices have access
to these data.
DCPs are generally owned, operated, and maintained by the USGS and the U.S. Army
Corps of Engineers (COE). These systems are capable of observing and transmitting
all types of data such as precipitation amounts, river stages, pool elevation at
reservoirs, and temperature. Data is monitored frequently, as often as every 15
Although most river and precipitation gages were operational, many did not provide an
accurate assessment of the true environmental condition just prior to the flood event.
Unheated tipping-bucket precipitation gages became blocked with snow due the heavy
snowfall from the previous weekend. River gages were affected by ice jams, and in
some cases there was flooding of several gage houses which washed out or
contaminated the antifreeze. The rapid warmup prior to the flood event and subsequent
quick freeze resulted in some gages remaining at one level from several hours up to
days or even weeks in some cases.
For the most part, LARC-equipped sensors performed well during the event except at
sites that had inoperative sensors or impaired communication circuits. Several gage
sites were flooded resulting in damaged manometers (actual river stage measurement
sensors). At some locations, the NWS employed backup observers to provide critical
information during the event.
Automated Local Evaluation in Real Time (ALERT) and Integrated Flood Observing
and Warning System (IFLOWS) networks are systems which provide additional rainfall
and river stage information. These systems are generally developed and funded by
local communities or counties. Several of the communities impacted by the flood were
equipped with ALERT or IFLOWS networks. IFLOWS was the most common
network available in all states inundated by flood waters. However, several ALERT
networks were also available such as those located in Steuben and Chemung counties in
New York. NWSO at Binghamton, New York, had only intermittent direct access to
New York IFLOWS data due to ongoing radio communications problems. Many
IFLOWS gages also became blocked by snow, leading to erroneous data.
3.3.1 SOME SPECIFIC PROBLEMS WITH GAGES (ALL TYPES)
Since most offices had similar outages, only several examples are provided. The
stream gage networks performed well given the magnitude of the event. In spite of
some outages and malfunctions in the automated gage network, there were only minor
impacts on river forecasts and warnings. Data from the manually read gages were
sparse during the event because of flood problems encountered by the observers.
Gage houses on Lycoming Creek and Trout Run Creek in Pennsylvania were damaged
during the event. Both gage houses were inundated by flood waters, and data
transmissions from DCPs ceased. Highwater marks indicated that floods of record
occurred at both of these locations. The USGS was able to put the Trout Run gage
back in service on Monday, January 22. The last data transmission from Trout Run
was at 5:30 a.m. EST on Friday, January 19, at a stage of 6.62 feet; flood stage is 15
feet. The DCP from Loyalsockville on nearby Loyalsock Creek was of little value
because of long delays in receiving the data.
The river gage at Marietta, Pennsylvania, on the Susquehanna River failed to measure
the actual crest. This was due to the river level rising above the level of the gage house
floor, causing the gage equipment to measure a constant stage until the water level
receded. The gage readings were erroneously low for over 12 hours due to this
condition. The USGS measured a crest from a highwater mark after the water receded.
The gage at Milton, Pennsylvania, on the West Branch of the Susquehanna River began
to transmit erroneous readings the afternoon of Friday, January 19. This was due to
ice damaging the orifice of the gage.
Cold temperatures late on January 19-20 resulted in a freeze-up in a number of gage
houses. Several of these freeze-ups occurred at locations which remained above flood
stage. The loss of current stage readings hindered the forecasts as to when water levels
would fall below flood stage at these locations. USGS personnel worked diligently to
return those gaging stations to service and were largely successful.
The Geostationary Operational Environmental Satellite (GOES) DCP network was also
utilized for precipitation. Timeliness of the GOES precipitation data was a problem.
For instance, the MARFC forecast run at 1200 Coordinated Universal Time (UTC)
could not include data from a number of the DCPs because the data was several hours
Precipitation amounts from the IFLOWS rain gage network had to be used subjectively
since there was some question about residual snow and ice that might have remained in
the gages from the early January snowstorms. This was especially true in the central
and eastern areas of Pennsylvania, central New York, and northern Virginia and
Maryland where only short periods of above freezing temperatures occurred in the days
prior to the event.
In Pennsylvania, the LARC precipitation gages functioned flawlessly during the event.
Data from these gages were invaluable in the analysis of the situation and as input into
the hydrologic models. The GOES DCP network was also utilized during the event.
For several reasons, a number of the GOES gages were nonfunctional during the event.
NWSO Binghamton reported that 9 LARCs and 27 Data Collection Platforms (DCP)
were not operating properly or were out of service. The gage outages did not prove to
be a major detriment, however, for the NWSO Binghamton staff. Reports from their
network of amateur radio and precipitation readers, and reasonably accurate WSR-88D
precipitation measurements proved to be valuable in making timely and accurate
warning and forecasts decisions.
Other gage problems were also evident concerning the availability DCP gage data.
This event illustrated the inadequacy of relying solely on timed transmissions from
DCP gages. In many cases, water levels rose in excess of 1 foot per hour on Friday,
January 19. Data updates received only once approximately every 4 hours were not
enough to track stage and rainfall during this event. Random transmissions were
received from some of the DCPs during the event and were very useful; overall,
however, there was a general lack of random reports. On a final note, DCP data was
unavailable from Saturday evening, January 20, to Sunday morning, January 21, due to
an NAS 9000 outage. Fortunately, this outage occurred after most locations had
already crested. Had this situation occurred on Friday, January 19, it would have had
a serious impact on the river forecasting efforts.
The locations currently served by manual equipment need to be surveyed to determine
the possibilities for improvement. Obviously, the possibility of automated equipment
should be explored for each location. If automated equipment is not practical or
affordable, then the need for additional manual equipment, i.e., staff gages, should be
examined, particularly in those locations which became inaccessible during this event.
Forecast problems were also compounded by some river and precipitation gage outages.
One forecast point, Shepherdstown, West Virginia, has not been operational for almost
2 years. Another key gage, the Great Cacapon, West Virginia, was eliminated in early
January 1996, due to USGS budget cuts.
3.4 COOPERATIVE OBSERVERS AND SKYWARN
Nationwide, the NWS maintains a network of approximately 10,000 Cooperative
Observers. However, this network has been declining in numbers in recent years.
These observers manually collect data such as high and low temperatures, precipitation
amounts, snowfall and water equivalent, river stages, and other data. These data are
usually reported to the local NWS office. Some observers report information daily
depending on the weather that is occurring, such as significant rain or snowfall based
on a set criteria. The data is converted to SHEF by the local NWS office and entered
into AFOS for transmission. These data are then used in forecast models at the RFCs.
These data are also formatted into products used by the public, media, and other users,
and used in determining local climatology.
Many observers in West Virginia report their data through ROSA, a telephone keypad
data entry system which allows observers to enter their reports automatically into a
central computer system. Observers without ROSA must telephone their reports to
NWS personnel who must encode the information into SHEF and transmit it through
AFOS. During this event, many of the ROSA reports were collected by the NWSFO at
Charleston, West Virginia. However, some observers had problems collecting and
sending information because they were impacted by the flood or there were local
communication problems. These problems caused late cooperative reports to be
received Friday morning, January 19. Routinely, ROSA and cooperative observations
that are phoned in generally are received between 7-8 a.m. EST. During the early part
of the flood event, many reports were not received until after 10 a.m. EST.
SKYWARN is a network of NWS-trained volunteers, mainly licensed amateur radio
operators who serve as hazardous weather spotters. SKYWARN spotters also include
police, fire personnel, emergency managers, and other public safety officials. During
times of hazardous weather, these volunteers are activated by the NWS or other public
safety officials to provide hazardous reports from their location. These volunteers
played a key role in reporting to the local NWS offices significant rain and snowfall
information, flood conditions, and damage reports. Several SKYWARN personnel also
maintained equipment to provide the NWS with temperature, wind speed and direction,
and precipitation measurements.
NWSOs Binghamton, New York, and Pittsburgh and State College, Pennsylvania,
made full use of their SKYWARN observers since many of them are trained to measure
rain and snowfall amounts and relay that information via amateur packet radio.
Cooperative observers were extremely valuable during this event. When the
Petersburg, West Virginia, river gage on the south branch of the Potomac became
inoperable, a nearby observer provided readings from a staff gage. When the water
exceeded the height of the staff gage and citizen evacuations were requested, he
continued to estimate the river stage until he was evacuated. A day later he assisted the
USGS in deriving an estimated crest height and time of occurrence. This critical
information will be included in an updated E-19. Another observer near the
headwaters of the Rapidan River provided river reports throughout Friday. With no
gage at the site, the observer did a very good job in providing other types of
information that was used for forecasting downstream river rises.
3.5 COMMUNICATION SYSTEMS
3.5.1 AUTOMATION OF FIELD OPERATIONS AND SERVICES (AFOS)
AFOS is the main operational communications system of the NWS and is based on
1970s minicomputer technology. However, recent improvements such as upgrading the
old CPU, the addition of a front-end processor, and higher speed modems have updated
the equipment in recent years.
During this event, AFOS systems proved to be very reliable except for several unstable
periods at State College, Pennsylvania. At the MARFC, two AFOS computer boards
had to be ordered during the event. That order was delayed but products were still
disseminated through backup systems.
3.5.2 NOAA WEATHER RADIO (NWR)
Nationwide, low-powered NWR transmitters provide continuous weather information to
specially designed radio receivers. Many of these receivers are equipped with a
warning alarm feature which, when activated by NWS personnel for the issuance of a
hazardous weather alert bulletin, will sound an alarm. NWRs are programmed by
NWS personnel via telephone or ultra high frequency communication systems.
At this time, about 75 percent of the contiguous United States is covered by NWR
broadcasts. All major metropolitan areas are covered by NWR broadcasts, but many
rural areas lack adequate coverage.
No major outages of NWR transmitters were noted during the event, but the coverage
pattern of several areas of each CWA is not complete, especially over northwest New
Jersey and the Delaware basin. Programming on most NWRs became lengthy during
the event due to the vast amount of hydrologic information which was placed in the
broadcast cycle. However, except for calling local NWS offices, this was the only
source many emergency managers had to obtain stage information.
3.5.3 TELEPHONE SYSTEMS
Most NWS field offices involved reported receiving and making over 1,000 calls
during the period from Thursday, January 19, through Sunday, January 22. In
addition, all offices made follow-up phone calls to the county communications centers
relaying most Flood Warnings issued by the NWSOs and NWSFOs. These calls were
also used to obtain flood information from county and city officials and to provide them
with additional forecast details. Overall, telephone service was satisfactory, but some
rural areas did lose communications impairing the ability of the NWS to access some
river and rain gage information.
Until the event's conclusion on Tuesday, January 23, the Baltimore,
Maryland/Washington, DC, office took 1,500 phone calls and made numerous
outgoing phone calls and National Warning System (NAWAS) contacts. Many of the
contacts made were with public safety officials, the media, other local, state, and
Federal government representatives, and cooperative observers. Telephone recordings
were kept up to date with the most current flood information.
3.5.4 NATIONAL WARNING SYSTEM (NAWAS) AND STATEWIDE COMMUNICATIONS
NAWAS is a telephone hotline system operated nationally by the Federal Emergency
Management Agency (FEMA). It is used to coordinate and disseminate warning
information to local officials with NAWAS lines. Although NAWAS sites have been
eliminated at the county and local level, all NWS offices and many states still maintain
NAWAS lines at major locations such as state warning points, highway patrol, or state
police offices, and in major metropolitan areas. The NWS uses NAWAS to confirm
that warning products issued have been received by state and local agencies, and to
receive storm reports from those same agencies. There were no known failures of the
NAWAS system during the flood event. All watches, warnings, and other necessary
information were promptly disseminated.
Along with NAWAS, each state maintains its own statewide emergency communication
system, usually through a state law enforcement agency such as the state police or
highway patrol. The Commonwealth of Pennsylvania probably maintains one of the
nation's most advanced communication systems through satellite. This system provides
an almost instant relay of all NWS watch and warning products to each county
emergency manager across the state.
As effective as this system is, it does not relay NWS products by Federal Information
Processing System (FIPS) codes, but rather by product header. As a result, counties
receive selected watch and warning products issued by the NWS in Pennsylvania.
Many of the statewide communication systems work well, but unfortunately many
counties have only one computer linked to the statewide system. This data link is
usually located at a sheriff's office, not at the emergency manager's location. One
solution for emergency managers to receive timely weather information at their office
would be for each county emergency management office to obtain access to the new
NWS Emergency Management Weather Information Network (EMWIN) or subscribe
to a commercial service.
Return to Table of Contents