I used DHT22 sensors for a while. I'll continue to use it in some project, and i'll keep supporting those for this project, because they're cheap and easy to come by. However, those are not really good sensors. Why? Because of their weird communication protocol, low accuracy, and variance between multiple sensors. They are good enough to know if it's hot, wet, etc, but not so much to measure those parameters.

We're trying to build a meteorological station/network of sensors here, so let's up the standards. I have bought an MCP9808 from Adafruit, that is given to have a good precision, and a guaranteed 0.5 max error at common temps. It's also I2C, a standard and and widespread protocol. For 5$ it should be more reliable than what we get from the DHT22/DHT11, and we'll see how it compares to the DS18B20.

I also ordered a BME280 from Adafruit for 20$, and it is supposed to be much more accurate about temperature, pressure and hydrometry. I also ordered this version of the BME280 from Aliexpress for 7$. When i get all those sensors, i'll make a measurement box with data logging and compare the results of all sensors (including the DHT11, DHT22, DS18B20, etc).

That's for the basic sensors. I don't have a way to compare loads of anemometers, rain meters, or wind direction sensors. I'll use what i have, and what can be easily found : the sparkfun kit.

I decided to make this more of a environnemental measurement station than just a meteorological station, meaning that we won't restrict ourselves to classic sensors.

I will measure the amount of dust in the air, using a cheap sensor with the GP2Y1010AU0F, that i bought from aliexpress at 8$ for a 2-pack. Again, i have nothing to compare those to, but at least i can check for consistency over time and among multiple sensors.

I also plan to include lux sensors, UV radiation sensors, and even a radioactivity sensor.

So, we'll make everything possible to support the most common sensors, but also try to use higher quality sensors if available.