[aprssig] Low-cost weather station: Weather Observation Standards & Sensor Siting

[Ray McKnight]

-----Original Message-----
From: aprssig-bounces at lists.tapr.org
[mailto:aprssig-bounces at lists.tapr.org] 
>> Most likely, it doesn't conform to current NWS/WMO
>> standards.  

>Ev, W2EV asks:
>
>Where would an interested person find those reporting
>standards specified?

Good question.  I'll post the following again for the benefit of those
interested:

To assist Amateur Radio Operators and other individuals using personal
weather stations to collect and report observations, I have compiled
this reference of terms and standards.  The primary resource used was
the Federal Meteorological Handbook No. 1, Surface Weather Observations
and Reports (FMH-1).  FMH-1 complies with international standards
outlined in the World Meteorological Organization (WMO) publication No.
306, Manual On Codes, as well as ICAO Annex 3, Meteorological Services
for International Air Navigation.  Another good source of information is
Federal Standards for Siting Meteorological Sensors at Airports,
FCM-S4-1994.

This information is provided for reference only, and should NOT be
considered as a replacement to official publications or standards.  Its
purpose is to clarify common misconceptions found throughout the Amateur
and Internet communities, with the goal of increasing the accuracy and
usefulness of data collected by amateur meteorologists.

It is important to note that the primary motivation behind most weather
observation programs is to assist forecasters in providing accurate data
for Civil, Commercial and Military Aviation.  Thus, many standards have
originations based upon the needs of the aviation community.

The recent proliferation of inexpensive, hobby oriented weather stations
has created a secondary network of weather observation stations
throughout many areas which formerly may have been substantially
isolated from primary stations operated by the government and industry.
These private weather stations benefit not only the owner, but can also
be integrated into the primary network to assist forecasters in
collecting data over wider areas and from locations previously devoid of
data.  The Private Weather Stations (PWS) may use various means to
supply their data to forecasters, the two most common being automated
transfer using the Internet (TCP/IP) and Amateur Radio using the APRS
Network (which is often also combined with the Internet to expand
coverage).


PWS systems differ from their professional counterparts in several ways.
Primarily, PWS are less accurate than professional instruments.  Also,
the installation of a PWS may introduce substantial anomalies due to the
inability to site individual sensors in the appropriate manner.  PWS are
rarely if ever calibrated to traceable standards, and their routine
maintenance and adjustment often not as frequent.  As an example, a
station installed to NWS standards at an airport usually has no
obstructions for 1,000 ft in all directions, is situated in a clear area
devoid of tall foliage, and is fairly level.  A PWS, however, is often
installed on or near residential buildings, at lower than optimal
height, and is usually shadowed by trees and nearby structures.

We must also consider the differences between individual PWS
manufacturers and station models.  Because PWS are designed mainly for
the hobby market, data may not necessarily be sampled or formatted to
established guidelines.  The most common example of this is in how PWS
measure and report wind.  Each model has slightly different methods for
reporting gusts, averages and trends.  Those of us who use and rely upon
PWS must be aware of these considerations and base our reports and
forecasts accordingly.  

It is important to understand that data reported by a PWS which is known
to be in error, or may vary substantially from accepted standards MAY
STILL BE USEFUL!  Obvious bad data can be flagged and ignored, and
reports from specific stations known to have consistent errors can be
either adjusted by a correction factor or the data given less "weight"
in its influence in computer analysis and forecast models.

THE MOST IMPORTANT PRECAUTION IS THAT NO ONE SHOULD EVER USE A PWS TO
BASE DECISIONS IN SITUATIONS POSING IMMEDIATE DANGER TO LIFE OR
PROPERTY.  ALWAYS rely on recognized professionals and seek your weather
information from established sources.

PWS owners are urged to also use common sense and caution in the
installation and use of their systems.  Do not blindly accept the advise
included in installation manuals or on the Internet, as it may be based
upon false assumptions or inaccurate information.  One excellent example
I can provide is from the installation manual of a PWS highly regarded
by many Hams and widely used all over the world.  Its manual instructs
the owner to install the wind sensor aligned to magnetic north.  This is
absolutely wrong, yet consulting their own technical support personnel
failed to convince them of their error.  Because the manufacturer says
to do it this way, some owners of this PWS routinely report inaccurate
winds, yet often cannot be compelled to reconsider, choosing to believe
the PWS manufacturer instead of established protocols.

The primary observation code used in the United States for reporting
surface weather observations is METAR (Aviation Routine Weather Report).
Unfortunately, METAR is often confusing to many people, and not easily
understood without practice.  FEW pWS PROVIDE METAR COMPATIBLE REPORTS.
Consequently, the lack of any standard for PWS reports introduces
potential problems to those who receive them, as reports from different
stations are often formatted in widely varying ways.  GREAT CARE must be
used to ensure the proper format is understood to ensure the data is
interpreted correctly.

SOME STANDARDS APPLICABLE TO ALL REPORTING STATIONS

1) ROUNDING of NUMBERS:  - The fractional part of a POSITIVE number
greater than one-half will be rounded UP.  For instance, 1.5 becomes 2,
1.49 becomes 1.  For NEGATIVE numbers, the fractional part LESS THAN
ONE-HALF is dropped.  For example, -1.5 BECOMES -1, and -1.51 is rounded
DOWN to -2.  This is slightly different than methods common in most math
problems!

2) REPORTING OF TIME: - All reports should include a time.  This
contradicts the widely accepted practice in use on APRS where time
stamps are ignored, and are archived according to the time provided by
the server.  TIME SHOULD BE IN UTC, AND ACCURATE TO WITHIN ONE MINUTE OF
OFFICIAL TIME (U.S. Naval Observatory atomic clock, GPS time IF RECEIVER
IS LOCKED, if receiver is in DR mode time should be suspect).  Time is
assigned at the time the last element of the observation is taken.  For
instance, you may be using a sling psychrometer which requires several
minutes of preparation to obtain an accurate reading. The report time is
NOT the start time, nor the scheduled broadcast time.  All times should
be in a 24-hour format, using 0000Z as the start of the day, and 2359Z
as the end of day.

3) MANUAL OBSERVATIONS: - Should be within 10 minutes of the reported
time.  In no case shall observations begin more than 15 minutes PRIOR to
the scheduled observation time.  You should always make your
observations in the same order so each observation will be roughly at
the same interval during the reporting period.

4) DATA AVERAGING: - Manual observations may require a human to perform
"spatial averaging" which can be defined as the predominant condition
observed over a specified interval (clouds=30 minutes, wind=2-10
minutes, etc).  AUTOMATED SYSTEMS will substitute an average of
collected data, USING ACCEPTED ALGORITHMS AND TIME FACTORS.  Gusts and
squalls should be reported IF OBSERVED WITHIN 10 MINUTES of the actual
observation.

5) TIMELINESS OF REPORTS: - If the transmission of a report is delayed
it should NOT be transmitted if the report overlaps the next scheduled
observation.  Only the latest observation should then be transmitted.
NOTE: reports of volcanic eruption should be transmitted by any
available means and as soon as possible, regardless of the age of the
observation.

SITING AND ACCURACY OF SENSORS

1) It is recognized that PWS are generally incapable of being installed
and calibrated to the same strict standards as professional systems.
However, owners should be familiar with the following guidelines and
STRIVE TO ACHIEVE THE GREATEST ACCURACY POSSIBLE, within the limits of
their equipment and physical installation conditions. NOTE:  I have also
produced another reference aimed at installation guidelines located at:
http://www.beals5.com/wx/accuracy.htm   which includes information
contained in Federal Standards for Siting Meteorological Sensors at
Airports, and Federal Standard Algorithms for Automated Weather
Observing Systems.

2) Once your station is installed, you should observe data for 2-4 weeks
and compare readings to known sources, such as the nearest airport, NWS
office, or other nearby stations.  This will help establish a baseline
useful for recognizing known anomalies in your system.  Here are some
links to the APRS FINDU weather database and individual CWOP reports.

3) You should also periodically review your data (QC - quality control)
to ensure consistency.  Often, sensors may be installed in remote
locations, making adjustment and maintenance difficult.  It may be
better to turn off the system or discontinue reporting if known problems
cannot be readily corrected.

STATION INFORMATION FILE
To be developed

MEASURING WIND

1) Wind is measured in terms of VELOCITY which is a VECTOR of direction
and speed.  For most purposes, we should be concerned only with the
HORIZONTAL COMPONENT of the wind vector (one important exception to this
is dealing with thunderstorms and other severe weather).

2) Wind direction is always referenced to TRUE NORTH, not magnetic, and
should be reported to the nearest TEN degree interval.  IT IS ALWAYS
MEASURED AS THE DIRECTION FROM WHICH IT IS BLOWING, NOT THE DIRECTION
THE WINDING IS HEADING TO.  As an example, for a NORTH wind, you would
be facing true north, the wind would be in your face and it would be
reported as COMING FROM direction 00.  It is measured as an AVAERAGE OF
A TWO-MINUTE OBSERVATION INTERVAL.  It is considered VARIABLE if the
wind speed is 6 knots or less, or if the direction changes by 60 degrees
or more if the average speed is 6 kts or greater during the observation
interval.

3) Wind speed is defined as the rate at which the wind passes a given
point, and is should be measured in KNOTS (knots is defined as a speed
vector of "nautical miles per hour").  It is measured as an average of a
two-minute observation interval.

4) A gust is merely a description of the VARIABILITY of the wind speed.
It is defined as a rapid fluctuation in wind speed, with variations of
10 knots or more between peaks and lulls.  It is measured by evaluating
the MOST RECENT 10 MINUTES of an observation, and is reported as the
peak wind speed observed during that period.

5) A PEAK is defined as a maximum instantaneous observed wind speed.  It
should be measured by automated recorders.  It is the highest
instantaneous observed peak within the last routine reporting period.
It is usually only reported if 25 kts or greater.

6) a wind shift is a change in the direction of the wind of at least 45
degrees in less than 15 minutes, and the sustained (average) wind speed
during the shift interval should be at least 10 knots.

VISIBILITY

Although few if any PWS are sophisticated enough to automatically
measure visibility, I will include the standards here for those who
chose to make manual observations.

1) VISIBILITY is a measure of the "opacity" of the atmosphere.  For
manual observations, it is obtained using the concept of "prevailing
visibility".  It is what is considered representative of the prevailing
conditions at the observation point or station, and is the greatest
distance that can be seen throughout at least half of the observer's
horizon circle.  "Half" need not be a continuous arc.  It is usually
observed at the surface, however, it can be reported from an elevated
point, but that point should be consistent throughout all observations
unless identified by a descriptive indicator such as "sector visibility"
or "tower visibility".

2) SURFACE VISIBILITY is the normally the point of usual observation,
whether or not is is actually taken at the surface.

3) During visibility observations, the observer should move to the
nearest point or points which provides the greatest unobstructed view of
the nautral horizon.  Natural obstructions such as trees or hills are
NOT considered obstructions for visibility determination, as they define
the "nautral horizon".  Buildings are the most common "obstruction".

4) You should adapt to ambient light before making a visibility
determination.  Choose dark objects as your reference points.  At night,
use unfocused lights of moderate intensity as reference points (25
candela).

5) Visibility is always reported in statute miles.

6) When determining the distance of maximum visibility, the observer
should use the following guidelines:  choose visible reference points
with known distances that can normally be clearly distinguished.  If the
object appears in sharp focus and the colors do not appear blurred,
visibility is greater than that distance.  If they can barely be seen,
then the visibility equals their distance.  Use the most distant
reference points that are visible.  Report the farthest distance seen
within the prevailing visibility.

7) Prevailing visibility is considered variable if it is less than 3
miles and increases or decreases rapidly by 1/2 mile during the
observation period.

8) As a general rule, distances are reported to 1/16 mile under 1 mile
visibility, 1/8 mile under 2 miles visibility, 1/4 mile under 3 miles
visibility, to the nearest mile under 15 miles, and in 5 mile increments
above 15 miles.  

9) DO NOT confuse visibility with Runway Visual Range (RVR).

10) The average observer standing on level ground can see 3-5 miles IF
THE REFERENCE POINT IS NOT ELEVATED.  It is therefore imperative to
choose references with sufficient height to permit observation at the
maximum prevailing distance.  If no reference points are available to
permit determination of the maximum distance, on clear days using the
naked eye prevailing visibility is usually reported as 7-10 miles.


A work in progress...
Ray - WB3ABN
Kingston, WA