Dear Naser ahmadi sani , it is associated with the experimental design itself. That is to say, before talking about sensor networks, you should analyze the measurement process and the experimental design underlying. It is essential to ensure the repeatability, consistency, and extensibility of the measurement process. Also, a good measurement process should ensure comparable measures for performing different evolution analyses.

I suppose that in this case, you could carry forward a stratified sampling, considering each kind of particularity into the heterogeneity of the sensed region. From the statistics, there are different formulas to estimate the sample size for each region.

I mean, in each case, the measurement techniques, sensors, and requirements could vary sensibly according to the experimental design. I suggest taking a look to the measurement frameworks, ISO 25012 (Data Quality), and how to automatize the experimental design through the measurement process.

For your reference, here you have a review about how the particulate matter is measured today through IoT sensors:

Article IoT-Based Approaches for Monitoring the Particulate Matter a...

Also, here you have a reference about a methodology for defining a measurement project using a measurement framework:

Chapter The Impact of the Measurement Process in Intelligent System ...

I hope that it is useful for you!

Dependent on the reflection of the spectrum in the layer