In order to conduct Inverse Distance Weighting (IDW) of EMF from cell phone towers in an area of about 50 sq. km. and at the same time having 12-15 cell phone towers, should one opt for grid wise GPS locations or is it advisable to take random locations?
Inverse Distance Weighting (IDW) is just a heuristic method, it is not based on any theory so choosing data locations randomly is irrelevant
IDW will not give you any measure of the interpolation errors
IDW is based on the idea that there is a "value" at each point in the region of interest. In your case it appears that the "value" is the EMF which is a composite of the EMF's generated at each of the cell towers hence there is an interaction between the spatial patterns of the cell tower locations and the spatial pattern of the GPS locations (presumably where the EMF measurements are to be made). A further complication is the spatial pattern of any buildings or structures or topography in and around the cell towers.
IDW is quick and dirty, i.e. easy to use and requiring no theory or understanding but likewise the results may not be very good
I suggest you look at alternative interpolation methods, e.g. kriging. However you will also want to consider characteristics of specific problem
Thank You Dr. Donald Myers.
As regards the topography and surroundings, it is an agricultural landscape in rural India with small settlements in between. The terrain is flat and there exists little spatial barrier with respect to the towers.
I carried out a study on "Comparative Interpolation Methods for creating Digital Elevation Models (DEM) (A case study at Chorin Agricultural Landscape, Brandenburg-Germany)"
objective: to create and compare digital elevation models for the agricultural landscape of Chorin while demonstrating appropriate choice of different interpolation techniques for different forms of datasets
Methods & dataset: IDW and Ordinary Kriging were selected and used for equal number of regular (grid-wise) and randomly selected resolution sampled points
Results suggested that: kriging with randomly distributed dataset (sampled points) produced more accurate estimations than the IDW method for either random or grid-wise datasets based on the Root Mean Square (RMS) values.
will attach report later
- Badges
- Science topic
- Similar topics
- Biological Science
- Biochemistry
- Biochemical Techniques
- Analytical Chemistry
- pH