In this study, optimum combinations of available rainfall gauging stations are selected by a model which is consist of geo statistics model as an estimator  and an optimized model. At the  first,  watershed  is  approximated  to  several  regular  geometric  shapes.  Then  kriging calculates  the  variance  of  the  estimation  error  of  different  combinations  from  available rainfall gauging stations using inside and outside stations of watershed. In each combination, n is number of considered stations and N is number of available stations (N>n). At the end, the best combination is selected by genetic algorithm (the  error variance of this combination is minimum). For optimal set with one sample point (station) estimator model and optimize model select station that locates near to center of watershed. While for two stations case, these models select two stations that l ocate in boundaries face to face. Also for combination n  stations  of  N  stations,  selected  stations  have good  and  proportional  distribution  in watershed. These results show correctness of research methodology.
In this study, effects of variations of paramet ers of theoretical variogram and number of blocks  in  block  estimation  of  kriging  method  are  evaluated  too.  The  variance  of  the estimation error from block estimation with 8*8 blocks has showed the acceptable results.
This research shows a linear relation between variations of error variance and scale of variogram. Optimum combination does not vary with variations of scale of variogram but it varies with variations of range of variogram. Increasing of nugget effect of variogram would raise the variance but does not vary optimum combinations.

Digging of geotechnical boreholes and soil resistance tests are time-consuming and expensive activities. Therefore selection of optimum number and suitable location of boreholes can reduce cost of their drilling and soil resistance tests. In this research, a model which is consisting of geo statistics model as an estimator and an optimized model is selected. The kriging calculates the variance of the estimation error of different combinations from available geotechnical boreholes. In each combination, n is number of considered boreholes and N is number of available boreholes (N>n). At the end, the best combination is selected by genetic algorithm (the error variance of this combination is minimum). Also the Kean Shahr of Ahvaz city (in Khuzestan province, Iran) is selected as case study in this research. Location of selected boreholes is in points that soil resistance of these points represents mean soil resistance of total region. Optimum number of boreholes is 15. Also results show that location of selected boreholes depends to soil resistance and diameter and length of applied piles are not important for this purpose.