I guess that if refers to this isolated circles that are common there, but what is the meaning of them, and why it occurs? is it a product of the equation and the weight distance system?
I had never heard that term before, but if I understood what duck eggs are (isolated circles/dome features) I think you are right.
This happens when your IDW power is too high and the importance that further pixels from the central one (the input points, with measured values) is too low.
The effect is a limited "impact area" of your measured value and when the IDW is run, and you visualise your results with classification methods, the input values are ' hot spots" or "duck eggs" in your map.
Obviously it is a matter of sampling design and point density, but if you have for example few points aligned and greatly spaced to one another, IDW will produce duck eggs in a raster.
Hint 1:reduce the power of IDW or change interpolation method (more scientifically correct)
Hint 2: change classification methods when visualising results (stretched for example should avoid sharp differences in classes, recognizable as duck eggs).
Obviously Hint 2 will not eliminate the problem, duck eggs are still there but stretched so much that are almost invisble. Not recommended.
thanks a lot for your help. Actually this is the first time also for me that I heard this term. I read it in the "Principles of Geographical information Systems" by Burrough and Macdonnell. I guess that was one of the disadvantages of the method, and I guess that this problem appears also in the ordinary krigging interpolation. The book suggest something like change the method, but I am not really sure if that is just a visual problem, or if that can lead to a wrong interpretation afterwards
I think it is just a consequence of your point location, the IDW equation and the power value.
If you wanna stick with IDW but you cannot go in the field and collect some more data to densify your network of points, duck eggs will occur anyway. Either in circular shape (if points are so far from one another that the influence of the nearest point is null) or elliptical (if one point is close enough to have an influence with its value).
If you wanna change the interpolation method, give a look to the attached publication.
It is not wrong what you created with IDW, it is just a consequence of the IDW principles, your sampling design (density, number and quality of sampled data) and the power value associated to IDW.