Davis Weather Station

In early 2004, I purchased and installed the wireless model of the Davis Vantage Pro weather station.  It has been running fairly well since that time, feeding data to the wunderground.com web site every 2.5 seconds via the PERL vanprod script on my main Linux server.  Weather data is also uploaded every ten minutes to the APRSWXNET and to the AnythingWeather web site.  The later is accomplished with a simple PERL script found on the repository page.

To display the vanprod data in real-time, a series of programs was written for my local Apache web server running Slackware Linux.  All current weather conditions, including a graphic display of 14 weather conditions since midnight, dominate wind direction for the last 60 minutes, barometric trends, estimated snowfall (based on liquid equivalent), sunrise and sunset, normal maximum and minimum temperatures, heating and cooling degree days, moon phase, the Vantage Pro text/icon forecasts, and weather advisory comments are displayed on a web page using any modern web browser capable of executing standard AJAX commands.

Color-coded red and blue arrows [] are displayed next to most data, representing values that have changed within the past 60 seconds.  The color of the text will also change to either red for rising, or blue for falling, based on the previous 2.5 second data value.  In addition, a small alarm bell [] is displayed next to weather data that exceeds last years extreme value.  And when it is raining, a small umbrella [] is displayed adjacent to the rainfall rate.

A webcam view of the current outside weather conditions is displayed as a thumbnail image, with the option of opening a full-sized image in a second web page.  A "weather movie", displaying each minute of the previous days webcam in a DivX compatible video, is also available for download.

Other data, such as the temperature and dew point in Celsius, wet-bulb temperature, snowfall possibility, barometric pressure in millibars, Summer Simmer Index, Temperature-Humidity Index, Humidex, WBGT flag warnings, the 15 minute, 60 minute, 24 hour rainfall rates, and the monthly/yearly rainfall departure can be viewed with most browsers by mousing over their corresponding entry.  Mousing over the maximum data conditions since midnight will provide the daily range of that value.  Mousing over the sunrise and sunset values in the weather almanac will display the length of day and night respectively.

On demand data graphs showing wind speed and direction, temperature, dew point, wet-bulb, heat index, wind chill, humidity, barometric pressure, rainfall, and snowfall are also available.  Graph duration is selectable from as little as one hour to as long as 56 days.

A slightly abridged version of the above page displaying live Montevideo weather and data graphs for the last seven days is available here for your viewing pleasure.

Mounting the ISS

In preparation of "Ol' Man Winter", I upgraded the mounting of the outside unit.  The wind speed/direction unit is now guyed, and raised up to about the 25 foot level, which is about 15 feet higher than my previous mount.  In the past, the unit would be partially blocked by the surrounding building and trees.  The increase in height makes the wind speed reading a bit more accurate.

Because of the increased height, I decided to add better bracing to the base mount.  A 1-1/4 x 4 board was added a few feet below the the previous mount, providing a longer mounting area.  Two heavy-duty saddle antenna mounts with one standard saddle mount were bolted to the wood with 3/8" lag bolts.

While not absolutely required for weather watching, I decided to purchase the rain collector heater from Provantage so that I could measure the amount of liquid precipitation of snow and freezing rain.  The unit consists of a 24 ohm, 50 watt wire wound resistor, a 110° thermal cutoff switch, an aluminum foil clad piece of bubble wrap as a heat shield, and a 24 VAC/1.5 Amp wall-wart.  A well stocked junk box could build the same unit for about $10 or less.

Towards the end of the rain heater installation manual, I discovered that "it may be necessary to add a lamp or heater beneath the unit to keep the drains open".  I decided to add a 470 ohm, 2 watt carbon resistors to the each of the drain openings as a means to keep the drain open in colder weather.  The resistors are connected in parallel with the main resistor, and dissipate about 1-1/4 watts of heat.  Bench testing showed that the temperature of these resistors get up to around 155° at room temperature.  Hopefully, this will be enough heat to keep the drains open during the colder temperatures.

ISS Problems

One morning during the summer of 2005, I awoke to find the rain fall rate on the Davis console to be locked at 14.40 inches/hour with a total rainfall of over 53 inches since midnight.  We indeed had received a fair amount of rain the night before, but obviously it was not at the level now showing on the console!

After a few hours of troubleshooting, I determined that the remote ISS sensor was sending the bogus information.  This was done by placing the console in an RF tight environment (a microwave oven) and watching the rainfall rate.  The bogus information stopped incrementing the moment that this was done.

I removed the ISS from my roof and brought it to my test bench.  After opening up the unit, I was surprised to find that it contained some fairly sophisticated circuitry.  Unfortunately, I could not find any sort of "reset" switch.  So, the next best thing was to remove power to the unit.  This is not as easy as it sounds.  My unit consists of a solar panel to power/charge the unit during the day, and a Lithium battery to be used at night.  Removing both of these did not stop the unit from transmitting!  Looking closer at the unit, I found a one Farad capacitor in parallel with the main power source.  Apparently, this can power the unit for hours on a single charge.

I discharged the capacitor through a small 12 V lamp, and verified that the bus voltage was zero before reinstalling the battery.  This apparently was enough to reset the CPU, as the rain gauge has been working fine ever since.  I can only surmise that the unit went into some weird state due to a nearby lightning strike.  Keep this in mind if your ISS does something similar.

Rain gauge calibration

Over the past few years, I have always wondered how accurate the rain gauge was calibrated.  The value always seemed to be a little low when compared to other local stations.  Even the snow to liquid calculation seemed to be a bit off.  The only solution was to find a way to calibrate it.

I started out with a little math.  The Davis rain gauge has a circular opening of 6.5 inches.  This equates to a surface area of 33.1831 inches (3.1416*3.25^2)  One inch of rain on a 6.5 inch circle would be equal to 33.1831 cubic inches.  One cup of water has a volume of 14.4375 cubic inches.  Therefore, one cup of water poured into the rain gauge would be equal to 0.4351 inches of rain (14.4375/33.1831).

Before proceeding, be sure to mark the two small stop screws below the tipping buckets with an indelible marker.  This will make it much easier to see how far you have turned the screws, as well as make it possible to reset the unit back to the factory defaults, if required.  You will also need to clean any dirt from the tipping buckets, as this will affect your calibration.

It is important to limit the rate of the water flow into the gauge.  I did this by cutting the top off of a two liter soda bottle, and poking a small hole in the bottom.  If you make the hole too large, the excessive water flow will overload the gauge.  My hole was just large enough to let water flow at a three inch per hour rate.

Once you are ready, place the soda bottle in the rain gauge and pour in the water.  I used two cups of water to make the results a bit more accurate.  The adjustment screws change the value about 6-8% for each turn.  In my case, I rotated both screws out (counter-clockwise) two turns with a 3/16" nut driver, which increased the reported value by about 15%, or pretty much exactly what I needed to match the calculated value.

When you are finished, you can reset the daily rainfall value that the console has accumulated during your test, and you are all set.  I did my calibration during a period when there was no rain in the forecast, as well as no accumulated rain for the month.  With the exception of the "Storm Total", my rainfall totals were back to normal and ready for the new calibration.