It has been standard practice since the 1800’s to take temperature readings in the shade. The reason is that direct UV radiation from direct sunlight can skew the temperature reading by a large margin and create false data. The original Stevenson screen was designed in the U.K. in the 1800’s to eliminate errors in temperature readings by standardizing the direct environment around the thermometer, regardless of it’s location. Since these shelters are very expensive to purchase, I decided to build my own shelter using established guidelines. Here’s how I built it…
In the U.S. a Cotton Region Shelter was designed as a variant of the original Stevenson Screen design. The two shelters use the same principles in operation, while the CRS is usually a bit larger inside to account for the hotter temperature ranges encountered in the cotton growing regions of the Southern U.S. At this time, I know of just two vendors that still sell these shelters but one has recently discontinued selling it, while the other looks to be made of plastic or resin and is likely imported. Regardless, both are very expensive to purchase.
I came upon the idea to build my shelter while looking around at a used building supply store. I found some exterior wooden shutters that were for sale. I quickly devised a plan to cut the shutters to a suitable size and construct a CRS with minimal effort. Due to the fixed size of the original window shutters, my shelter is not exactly the same size as specified for a true to spec. CRS as used by the National Weather Service. It’s close enough though, and has worked superbly for many years.
After sawing the shutters to a suitable length, I glued and screwed them together using epoxy glue and pocket hole screws. To speed up construction, I used an old base assembly from an earlier shelter design. This base has 1/2″ gaps between the floor boards to allow for good ambient airflow inside the shelter.
I also built a double roof from plywood. This allows for airflow under the roof which is important because the direct sun will overheat a single layer roof which will cause errors in temperature inside. By having an 1″ air gap between the two roof panels, the temperature inside the screen is much more consistent and steady.
I built a wooden H-frame assembly inside the shelter to hold various thermometers and instruments and sensors.
After I finished construction, I painted the interior and exterior with white latex paint. There seems to be much debate about painting the instrument shelters. The commercial Stevenson screens available for sale in the U.K. are nearly always painted white on the outside and black on the inside. This is presumably to absorb any UV radiation, while the white paint outside reflects it off the exterior surface of the shelter. Below is an example of a commercially made Stevenson screen for sale in the U.K.
The Cotton Region Shelter design in the U.S. is painted white on both the inside and outside of the shelter. The next issue is whether to use gloss paint or flat paint. Studies have been conducted and published that show the direct effects of both types of finishes. The general consensus is that neither gloss or flat paint makes a significant difference in temperature readings, taking into account all of the other variables inherent in the wooden construction of CRS design. I chose to use a semi-gloss white latex paint as that’s what I already had in my shop. Case closed. 🙂
The first shelter that I had built many years ago was destroyed in an extreme hail and downburst wind event. I had built a wooden stand that was insufficient to withstand the force of a strong windstorm and it was always tipping over and moving around.
For this newer version, I built a strong and steady stand for the shelter by using treated fence boards that are screwed together. To eliminate tipping and moving out of place during high wind, I secured the stand using sturdy metal t-posts driven into the ground and attached them to the wooden legs of the stand.
According to published guidelines, the shelter must be mounted with the doors facing a northerly direction. This is to keep any direct sunlight off of the thermometers inside while the doors are open, thus skewing the actual readings. In my particular case this was not practical, so I am forced to mount my shelter with the doors facing a south east direction. I have noticed significant spikes in temp readings from my Davis sensor whenever I open the doors even for just a few seconds and the direct sunlight floods the interior. Therefore I only open the doors of the shelter after mid-day when the sun is in a better position and the sensor is not directly affected. If possible, I do recommend siting a shelter as originally specified with the doors facing north.
Finally, a few words about maintenance and use. Here in Texas we have a problem with wasps building nests. My new shelter is a virtual paradise for the pests and so I am constantly removing wasp nests inside the shelter. Also I have found that the outside of the shelter gets surprisingly dirty from exposure to the wind and elements. About twice a year I wipe down the exterior using a wet rag taking care not to damage the sensors inside the shelter.
If you decide to build your own version of a Cotton Region Shelter, I hope these tips will help you. It was a fun project to build and has proved very useful for mounting various instruments, cameras, etc. The even and consistent ambient air flow throughout the interior really helps keep the thermometers from wildly fluctuating and giving false readings, thus improving the overall accuracy of the station temperature records.