Looking at the graph data it seems that the sensor overestimate every time there was the sun
since the sensors are in the same radiation shield
It was strange since the probe is inside a metallic case so I though that maybe there was an influence by the UV rays. Since I have an UV led flashlight, a UV spectrometer and a similar sensor I made some experiments.
I have an SHT30 in a metallic case like the Ivan13's one
I got my UV LED flashlight anch cheched it with the spectrometer. It is an Ocean Optics USB2000+ that I used with the OceanART V1.01 software.
This is the LED spectrum
The peak is around 365nm, I have a common UV LED.
Then I connected the SHT30 to an Arduino board and used the IDE plotter and put the flashlight over the sensor tiny hole and I got this
As you can see the temperature increased dramatically and the humidity decreased accordingly. When I removed the light the temperature decreased more slowly and the umidity returned to the previous value. Temperature is in °C and since i plotted 4 data every seconds the time in the X axis must be divided by four to obtain seconds. Every 100 values are 25 seconds.
Then I use a white LED flashlight but nothing happened.
So I tested the plastic of the Ecowitt WH31 using the battery cover.
Without plastic the spectrum is the one above but when I put the plastic over the spectrometer hole I got this
The signal is near zero.
So I put the flashlight just over the spectrometer hole with the platic battery cover in the middle and I got this.
No UV light but there is a light with longer wavelenght, maybe due to fluorescence effect
Snow, an user of Meteonetwork forum, suggested me to use the flashlight with the WX31 so i put it just in front of one of the two small slits that are on the WH31 case and I got this.
Then I put the flashlight over the SHT30 metallic case and used agai Arduino and the IDE plotter getting this one.
The increase is small and the return to the previous temperarute take about one or two minutes
Then I checked how the plastic battery case reflected the UV light and I got this
So there is an UV reflection added to another light that is the same that is transmitted across the plastic.
It seems that the sensor is sensible to UV light but I am not sure if it is a photoelectric effect or there is a real temperature increase.