If a VES survey has much noise, then it is not possible to smooth the noise and end with a better result. It really depends upon the type of array used. An offset sounding array will allow you to quantify the noise. Other arrays do not give you this advantage. In either case, a noisy survey is really telling you that the ground can not be well approximated to a sequence of uniform plane layers - and therefore, a valid interpretation is not possible for this location. You would be better repeating the survey using the imaging technique.
If a VES survey has much noise, then it is not possible to smooth the noise and end with a better result. It really depends upon the type of array used. An offset sounding array will allow you to quantify the noise. Other arrays do not give you this advantage. In either case, a noisy survey is really telling you that the ground can not be well approximated to a sequence of uniform plane layers - and therefore, a valid interpretation is not possible for this location. You would be better repeating the survey using the imaging technique.
There are several reasons behind VES curves distortions most of them are related to the effect of heterogeneity which produce shifting of MN curve segments , there is also positive and negative cups which are related to gravel lenses or buried metal pipes ... etc. Furthermore , erroneous readings related to resistivity meter or array used.
Zohdy et. al. 1974 , in the reference (Application of Surface Geophysics To Ground-Water Investigations) discussed all resistivity curves distortion cases.
During manual curve interpretation positive and negative abnormal kicks neglected , also shifted MN curves segments treated to insure that the resulted curve is smooth and continuous before applying matching method with Orellana and Mooney, 1966 standard curves.
After manual interpretation the smoothed curve data input to a software like IpI2Win , in order to enhance results by reducing the rms% between the field and calculated resistivity curves as much as possible.
I agree with the answers of the colleagues given above. It is important to known is this a singular noisy curve, or the noisy records are repeated on all VES sites in the area. For one or two noisy curves among a number of measurements the solution is clear - ignore them, repeat the measurements if necessary, or remove some measurements from the affected curves to make them similar to the adjacent VESes. If the noisy effect is recorded in all studied area, be sure that your equipment has to be tested, or you have artificial source of electrical potentials affecting your measurements. If you don't known the measuring conditions, the geological background, the potential sources of random electrical emissions, don't loose your time to make speculations on the VES curves. The result will be most then unacceptable.
I agree with Alkhafaji in using IPI2Win as very good tool to graphically edit VES curves. Furthermore, it is possible to make a statistical analysis of the VES curves. In this case, types of characteristical curves of the site are defined. This will help you to be more effective using IPI2Win. Please check paper where statistical analysis is performed.
I totally agree with Mr. Acworth. If the area is not really inhomogeneous, and, the 1D assumption is almost OK, then the technique which is mentioned by Mr. Jackson, would be a good idea.
I have worked with several VES datasets. Sometimes the noisy data can be pinpointed. However, in most cases, removal of bad data points is very hard and impossible (especially for smaller AB readings). On the whole, try 2D ERT if possible.