St. Bernard, Nova Scotia
Denman Island Weather Station
The Denman Island weather station is owned and operated by me, Keith Walker. I do it because I am a science nerd and I like keeping statistics. My claim to be the "official" Denman Island weather station is simply because I write the monthly Weather Retrospective column for the Flagstone.
I had kept temperature records since long before moving to Denman Island in 2005. The original version of the weather station consisted of a thermistor connected to the game port of a PC. PCs no longer have game ports, so that incarnation was retired when I bought my most recent computer, replaced by the One-Wire network described below.
Within two months of moving to Denman Island, I had my weather station up and recording, using my home-made temperature sensor. In August 2006, when Graham Brazier retired from writing the Flagstone weather column, I stepped in to take it over. Graham graciously gave me copies of his weather records going back more than ten years.
When I began writing the Flagstone column, I added a home-made rain gauge, since upgraded to an electronic one. I have gradually added more instruments to the station over time.
The weather station is located at: 49° 32' 16.34"N, 124° 48' 11.72"W, at 300 feet above sea level.
The station is located on Denman Ridge, at one of the highest points on the island. The altitude means that it sometimes gets more precipitation than lower parts of the island. Being at the top of the ridge also means that summer daytime temperatures and winter overnight lows tend to be higher than in lower-lying areas.
The station is located in a 1.5 acre clearing in a forest of 120-foot douglas fir trees, far from an ideal observing site. It means that my recorded wind speeds are a fraction of what would be recorded in open air. The wind could be howling at 90 km/h overhead, and my anemometer will show less than 10 km/h. So, take my wind speed observations with a grain or two of salt.
Wind direction measurements also suffer because of the forest. The wind tends to swirl and be quite turbulent within our clearing. As a result, the recorded direction tends to vary rapidly and randomly.
The weather station is fully automatic and runs 24 hours a day. The various instruments are connected to the computer by a One-Wire network, a simple and reliable low-cost instrumentation network system. Most of the components came from Hobby Boards.
The temperature and humidity module from Hobby Boards is located in a Stevenson screen in a small meadow. The Stevenson screen is home-made and has double louvers for maximum thermal radiation protection. The centre post inside the screen has a regular thermometer on the front side for calibration and the Hobby Boards module on the back side. The temperature module actually has two temperature sensors. I read both of them and report the average.
I used one half of the Hobby Boards case as a mounting bracket, but I left the other half of the case off. I found that, when the module is powered (as opposed to operating on parasitic power), the voltage regulator throws off enough heat to raise the temperature inside the case by a couple of degrees. Leaving the case open lets the heat dissipate. The module is mounted with the voltage regulator at the top to ensure that it does not heat the temperature sensor by convection. The module is connected to the computer by about 150 feet of underground cable.
The rain gauge is a Rainwise tipping-bucket type gauge, located on the railing of the deck. The height of the gauge with respect to the railing is not ideal, since splash from the railing could end up in the funnel. However, this is a compromise with my wife's need for an esthetically pleasing line of the deck and railing. The splash issue is not likely to make a big difference in reported rainfall amounts.
The rain gauge is reliable in above-zero operations. At and below the freezing point, it freezes and will not register precipitation until it thaws. For light snowfalls followed by above freezing temperatures, this causes a delay in counting the precipitation until the snow melts. For heavier snowfalls that more than fill the funnel of the gauge, and for temperatures that stay below freezing for a long time, I have to take manual measurements. At some point, I may provide heat for the rain gauge.
The AAG wind instrument is located on a mast at the east end of the house roof. It contains an anemometer (wind speed) and a wind direction vane. (There is also a temperature sensor, but I do not use it, since the anemometer is necessarily exposed to full sunlight.) As noted above, the wind speed tends to under-read, and the wind direction tends to be fairly random, due to the proximity of trees. The gable end location places the instrument about 20 feet above ground, and as far from trees as is practical in this forest location.
The gray box on the anemometer mast houses a solar radiation sensor and an ultraviolet sensor, both from Hobby Boards. Both modules are mounted inside the weatherproof enclosure using their cases as mounting brackets. The cases are glued to the inside of the lid of the box, and their windows are sealed to the case with caulking. Cables enter the box through weatherproof clamps on the underside.
The Hobby Boards barometric pressure sensor is located indoors in my office. It is calibrated for our altitude and provides a reading that is corrected to sea level. As a result, it usually reads the same as the pressure at the Comox airport, 20 km away.
The instruments are connected to the comnputer by a One-Wire network. This simple instrumentation network typically uses standard network (Cat-5) cable, and can use ordinary telephone wiring. This greatly simplified connecting some of the more remote sensors, since I was able to utilize an unused pair of conductors in the existing telephone cable. The network connects to the computer via a small USB interface module.
The One-Wire system is quite simple, but one must still take into account the electrical characteristics of the wiring. Using telephone wire instead of CAT-5 network cable reduces the maximum length of cable that one can use. In addition, the overall topology of the network must be a single line; it does not like to operate over a Y- or star-shaped configuration. My initial configuration was Y-shaped, causing reliability problems with the rain gauge.
I have since installed a network hub (visible in the barometer photo) that allows each branch of the network to operate separately in the desired single-line configuration. This has improved the reliability of communication with the various devices. As a consequence of this change, I was able to move the wind and temperature instruments to more favourable locations, as described above.
Though not strictly a weather instrument, the station includes a sensor to measure well depth. This sensor is entirely my own invention. Its data are recorded using the same software that records the weather observations, and it uses the same OneWire network to communicate with the computer. More details.
Weather readings from the various sensors are collected and recorded every five minutes by my computer, using software I wrote myself. Once an hour, the readings are uploaded to this website.
The software was written in VB.Net, and runs on a Windows Vista platform. It is loaded as a service, which enables it to start automatically after a reboot without operator intervention, important when Windows decides to upgrade itself in the middle of the night.
The temperature that is reported on the website at any time is an average of two temperature readings at the time indicated. Wind speed is an average over the previous five minutes. Gusts are not recorded separately. Wind direction is a snapshot at the moment the five-minute observation is taken. Daily averages are true averages of all 288 five-minute readings, not simply the mean of the day's high and low.
If rainfall readings are missed, whether due to power failures or computer or network malfunctions, the rain gauge keeps counting regardless, using battery power. When the software eventually gets a reading, it will record the entire accumulated amount in the next five-minute slot. However, the time of the last reading is also recorded, allowing me to work out how much of the amount to allocate to which day, in the event that the outage spans more than one day.
The hardware is now in a stable long-term configuration. I am always doing minor software upgrades. I may at some point add a soil moisture sensor and have considered monitoring temperatures at other locations, particularly in the garden.
Copyright © 2010 Keith Walker
Last modified: 25-Jul-2010