My Sense Hat Temp/Press/Hum Readings seem inaccurate

[brainchill]

So I've been trying to make the best of this for a few days but it's just driving me crazy. I know that this has to be my inexperience with python or the sense hat or something so I'm certainly not blaming the hardware but my readings seem to be wildly inaccurate. I would understand if it were just the temperature as I've seen the threads about the cpu thermals interfering with the temp sensor on the sense hat but all of my sensor reedings seem to be really off.

As an example the current outdoor temperature pressure and humidity as read by my davis weather station are
Temperature = 70F, Pressure = 31in, Humidity = 26%

and my current reading from the pi, sitting behind my house without direct sunlight are, converted for matching units
Temperature = 75.7, Pressure=38.5, Humidity = 20.4%

Like I said, the temp I know will be variable (I have a pimoroni black hat to move the sense hat off of the pi itself)
but that doesn't explain the other numbers being skewed unless the warmer temps are directly affecting the other
sensor readings ... it might increase the pressure but it should lower the humidity.

[experix]

I made a C program that lets you measure the temperature via all three sensors on the sense hat, plus the CPU, with ability to exercise most of the options that exist. They all get different results. I have not tried comparing vs. "known" temperature. It will be some time before I can look at this again since I am in the middle of moving across the continent.

[TideMan]

Atmospheric pressure will have an offset that depends on altitude and a few other parameters.
You need to calibrate it against known pressure (your Davis perhaps), find the offset and add or subtract to the sense hat pressure.
Thereafter, you'll get the correct pressure readings.

The attached plot compares pressure measured by the sense hat with pressure from a personal weather station (PWS) sitting alongside the sense hat, and with pressure from a MetService automatic weather station (AWS) in the next street. It is for a couple of days earlier this month when there was a large change in pressure. I have not applied the offsets so that you can compare the curves better. You will see that the sense hat and PWS pressures are duplicated even down to small fluctuations. The AWS is smoother because it is sampling at hourly intervals, whereas the others are sampling at 10 min intervals.

I think this confirms the accuracy of the pressure sensor on the sense hat, once you figure out the offset and apply it.

Temperature and humidity relate more to the local environment around the instrument, so they are not a reliable measure of the air temp and humidity in the room.

[richrarobi]

The temperature of the sense hat will actually be consistently about 5 degrees high, by the experience of many people. See this link.
viewtopic.php?f=104&t=111457&start=50

This seems to be due to radiant heat from the main pi processor.

If the temperature is high, then the relative humidity will be proportionally lower. Best bet is probably to move your sense hat away from the main board using something like a black hat hackr..... If you already did this then I don't have a clue!! Sorry, missed that last paragraph.....You may need to move the pi around to get different micro-climates - comment still applies that the relative humidity goes down when temperature goes up. (read up dew-point...dangerous temperature/humidity for fog, etc)

To add to the comments above ref: pressure, did you have to put altitude into your weather station, when you installed it?
There is a formula that escapes me for the moment....

[jojopi]

If you get pressure of 38.5 inHg then something is wrong.

Are you converting from hPa (millibar) correctly?