Stephen,
The models I'm using work on emissivity calculations. One model predicts the clear sky emissivity and the other one back calculates the cloud cover based on the observation of actual emissivity (calculated from ambient and sky temperature measurement). I've used the form of equations from literature but fitted the coefficients of the literature models to my local observations.
I also heard the DHT humidity sensors can be a bit hit and miss so ended up changing to an Adafruit SHT31 to get a more accurate humidity measurement.
https://www.adafruit.com/product/2857