can anyone suggest me the best sampling approach and best interpolation method for creating sound level map for an industrial area (including several refineries and petrochemical units)?
You can collect the sound level as a point data by using some tools with location information. Then you can interpolate collected data . The given link may be helpful to you
Are your samples obtained from measurement or calculation?
Assuming your samples are from measurement, unless:
1. the only sources of noise contributing to the sound level at the receiver location comes from industrial facilities
2. The sources of noise can be distinctively identified
3. intervening terrain is relatively flat
it would be very difficult to use any interpolation technique to create an industrial noise map from spot samples.
If you are considering assessment in an urban area, a road network servicing through traffic would most likely surround the industrial area. As such, other anthropogenic noise sources are likely going to influence sound level at the receiver location.
It may be best to create a noise map (taking into account topographic data and intervening objects) based on known or assumed activities ocurring at the refineries and petroleum units then verify your calculated noise map against noise measurement samples.
The branch standard I would say is to use software like SoundPlan or Cadna or similar. You would need the source strengths of the main noise sources and calculate the noise distribution map following an industry standard prediction method like ISO 1996.
If you do have a grid of measured values a graphic software like Surfer or other graphics program can do the job. Even the Conditional formatting in Excel can be used. I do not know the details in the typical interpolation methodes used in the programs, but several possibilites are usually available. It is pretty common to use 5x5 m grid or similar for the calculation map.