The basic idea of environmental sampling is to estimate some characteristic of an environmental population by observing only part of it. For example, to estimate the total number of a rare species of fishes in the marine water, a sample of units is selected for observation. But in a trawl survey for fish or other marine species, it is unlikely that every fish in a selected plot will be detected.

In this thesis, considering the roaming pattern a sampling design has been proposed for environmental roaming population. The detection probability has been considered is a distribution function of the roaming pattern of that particular species.

The Horvitz-Thompson method is also used in this thesis. The selection probability for particular individual is suggested, and the population total, population mean and population variance has also been estimated using the roaming pattern of the population. It has been observed that the estimated population mean of the proposed design is same as the population mean obtained from Simple Random Sampling design. Two-stage sampling procedure has also been discussed for the environmental population.