GIS based Spatial Sampling Procedure for Environmental Studies in Agriculture
by
Tauqueer Ahmad
Indian Agricultural Statistics Research Institute,Library Avenue, N. Delhi-110012, India
Coauthors: Anil Rai (Indian Agricultural Statistics Research Institute, Library Avenue, N. Delhi-110012, India)

The Geographic Information System (GIS) is a powerful tool for storing , retrieving, analysis and integrating spatial and non-spatial geographical data apart from drawing any kinds of maps. In last few decades there has been substantial developments in the field of GIS and spatial statistical techniques (Ripley, 1981, Griffth, 1988, Haining, 1990). Unfortunately, the level of integration between these two rapidly growing fields is at very low profile. However, recent advances of computer hardware and GIS software have a great potential to change substantially in statistical approach to the study of geographical reality. The ability shown by GIS to handle various kind of information through their geographical coordinates has a vast capability, particularly, for survey design and processing.

In classical sample surveys number of sampling techniques are proposed, which are applied to agricultural surveys with some modifications, depending on its feasibility in field. However these designs do not consider spatial parameters while, assigning the probability of selection to sampling units of the target population while in most agricultural surveys the parameter of interest is geographical in nature. The neighboring sampling units tend to be homogeneous, when, the parameter of interest is geographical in nature. If a particular unit is selected in the sample, the neighboring units are not likely to provide enough additional information of the target population and hence there is a need to modify the classical approach of sampling technique. This modification should not only ensure better representation of the target population, but also, likely to provide efficient estimation procedure under the classical randomization framework. The explicit purpose of this study is to see that GIS can improve traditional sampling methods for agricultural surveys in a substantial way.

Sampling design for spatial data have a long tradition starting from Mahalanobis (1940), Quenouille (1949) and Das (1950). Hedayat et al. (1988) suggested Balanced Sampling design Excluding Contiguous unit technique (BSEC) with the help of BIBD approach, which excluded contiguous units to be sampled, thereby, resulting in second order inclusion probabilities being zero corresponding to pair of contiguous units. Extending this idea, Arbia (1993) proposed Dependent Unit Sequential Technique (DUST), a draw-by-draw sampling procedure for more efficient area sampling design. Measuring the environment is an awesome challenge. Environmental studies in agriculture continue to be challenging issues to be tackled by the Agricultural Scientists. Some of the important environmental concerns in agriculture are soil, ground water, and air pollution due to excess application of chemical fertilizer, spray of insecticides and pesticides, spraying of weedicides etc.. To study the impact of different chemicals used in agriculture and identifying the environmental " Hot Spots" which are created in making efforts to provide sufficient food for our fast growing population classical as well as usual spatial sampling designs are not sufficient.

In this article an attempt is made to suggest a stratified sampling design in which not only the spatial nature of these environmental variables has been taken care by incorporating spatial correlation based on auxiliary character in selecting the sample but also the effect of clustering properties of environmentally polluted neighboring areal units is considered. It is proposed to stratify the study region based on neighborhood properties of the sampling units.

Date received: November 17, 2000