To estimate the mean value of discontinuities trace length by sampling window method is the key step of joint geometry modeling, sampling bias existed during the field surveys relatively to infinite two dimensions rock mass. Therefore, field sampling bias of trace length has to be corrected for the actual rock engineering to get the corrected mean value of trace length. Sampling window method proposed by Kulatilake in 1984[10] was simply reviewed, some modification to sampling window method was proposed in this paper, which makes the method more objective, more perfect. It is easy to simulate the corrected mean value of trace length in infinite two dimensions by means of sampling window method computer program.
The stability of rock mass or rock engineering is dominated by discontinuities developed in the rock mass. Three basic geometry parameters of discontinuities are the orientation distribution, size and size distribution and densities of discontinuities, these three basic parameters were emphasized to be studied in the research field of rock mechanics, engineering geology and geological disasters evaluation, it is a very important study for the evaluation of the stability of rock engineering and for the numerical modeling method in the research field of rock mechanics, such as 3-D numerical network modeling for the randomly distributed discontinuities in the rock mass (Kulatilake et al. 1995), the distinct element method (Lorig. 1984) and the discontinuous deformation analysis (G. H. Shi and R. E. Goodman, 1984). A large amounts of the former research paper on the orientation, size and density have been published, for example, an research on the orientation of discontinuous rock mass proposed by Mahtab, et al (1972, 1984), a study on the spacing of discontinuities was provided by S. D. Priest and J. A Hudson (1976). There were also a lot of results on the research of trace length in the past. A study on the trace length sampling bias correction of discontinuities was emphasized in this paper. The main methods to estimate the mean value of trace length of discontinuities are scan lines and sampling window, no matter which method to be used for trace length surveys in the field, the sampling errors can occur due to the following reasons:
Size bias: large joints have a greater probability of being sampled than small joints. This bias affects the results in two ways: the first one is the difference between the actual frequency and the frequency on an infinite outcrop. The second is the difference between the frequency of the sample on the infinite outcrop and that of a sample on a finite area of the outcrop.
Truncation bias: Joint length s shorter than some known cutoff length is not recorded.
Censoring bias: Lengths of the trace are longer than the sampling windows provided, That cause one or two ends of the trace length couldn't be seen. From The point view of engineering rock mass, truncation has a very lower affect.