(I'm not quite sure about all this "@Joe Huang" twitter stuff, but I did get an e-mail message from UnmetHours informing me of this question :-) )

In answer to the original question, i.e.., "the best way to find the type of sky (whether it is overcast, sunny clear, clear, uniform cloudy) in EnergyPlus", I have the following comments:

1) EnergyPlus *.epw is just a derivative file format, so that the data contained depends entirely on the original provider of the weather file, such as NREL for the US TMYs , ASHRAE for the international IWECs, etc.

2) 99% of the weather files currently available are processed from standard weather station reports, although they don't carry over all the reported data, and also calculate a number of parameters, i.e., solar radiation and illuminances, that are not reported but needed for simulation programs.

3) Standard weather station reports contain a field called "Present Weather", which uses a 2-number code for the weather conditions that hour (00 = clear, 01 = clouds dissolving, 04 = haze, etc.). Unfortunately, most processed weather files do not include Present Weather, and even when they do (TMYs, IWECs) I'm not sure if they are retained when converted to epw.

4) I'm not familiar what is "Daylighting Model Sky Clearness". There is no such term in either the source or processed weather data of which I'm aware.

5) The closest weather fields I can think of are the TotalSkyCover or OpaqueSkyCover (always missing in my experience), which range from 0 to 10 as tenths of the sky dome covered by clouds. This field (TotalSkyCover) is almost always reported, as well as retained in most *.epw files, since it's a primary component used to calculate Solar Radiation.

6) Although there's clearly some correlation between TotalSkyCover and PresentWeather, it's not an exact match (e.g., does 5 indicate that half the sky is covered, or the whole sky is covered to 50% opacity?) and I don't know of anyone who's attempted to correlate the two.

I don't want to venture any suggestions, as I'm not sure of the objective or need for the weather condition.

To answer the original question, Output:Variable Site Daylighting Model Sky Clearness is calculated as:

    Zeta = 1.041 * pow_3( SunZenith );
    SkyClearness = ( ( DifSolarRad + BeamSolarRad ) / ( DifSolarRad + 0.0001 ) + Zeta ) / ( 1.0 + Zeta );

This is used in two ways. First it is used to calculate the sky luminous efficacy. This calculation uses binned empirical coefficients which have 8 ranges, the last for SkyClearness > 6.2 (thus >6 is a clear sky).

The other use of SkyClearness is to determine the sky luminance at each time step. From the Engineering Reference

Time-Step Sky Luminance The sky luminance distribution, ψ, for a particular time step is expressed as a linear interpolation of two of the four standard skies — ψcs, ψts , ψis and ψos — described above under “Sky Luminance Distributions.” The two sky types that are interpolated depend on the value of the sky clearness.

Here there are three ranges for SkyClearness that are meaningful:

SkyClearness <1.2 uses the overcast and intermediate sky luminances

1.2 < SkyClearness <= 3 uses the intermediate and clear turbid sky luminances

SkyClearness > 3 uses the clear turbid and clear sky luminances.

Another method for teasing out the sky conditions from weather file data is to look at the ratio of direct normal to diffuse horizontal flux (can use radiance, irradiance, or illuminance). If you have a really high direct normal and low diffuse horizontal, you have a clear sky, and vice versa -- which follows, since a low direct normal means there are clouds between the sun and the detector, and all that flux is getting scattered by the clouds. The following images show these distributions, using Seattle, WA and Boulder, CO as exemplars for cloudy and clear:

These are falsecolor visualizations of June 21st at noon, using the Perez sky model; gendaylit was used to make the skies. The commands to create the two skies are underneath the images, and the values after the "-L" are the direct normal and diffuse horizontal illuminance (lux), respectively. This is admittedly a somewhat indirect means to get what you're looking for, but generally speaking, weather files have those two pieces of info in them...