Quality of soil maps is a prerequisite for rational land use and soil management. To assess the accuracy of the existing reconnaissance soil map (1:50,000) of the Chaharmahal Bakhtiari province, Iran, a second-order soil survey was conducted. Using aerial photo interpretation, profile description, and field observations, a detailed soil map (1:20,000) was prepared. Then, 85 pedons were described and samples were taken for laboratory analyses. The location of pedons was determined with a Global Positioning System (GPS) and then marked on both soil maps using ILWIS software. Properties of the surface horizons including percentage of gravel, clay, sand, silt, CaCO3 equivalent, organic matter, and the thickness of A horizon were determined. Although detailed soil map units were more homogenous, purity of map units at family and series levels was lower than those expected in American Soil Survey Manual. Taxonomic purity at series level was 18.75% to 33.3% and 0% to 1.3% for detailed and reconnaissance soil maps, respectively. With respect to similar soils, interpretive purity could improve to 75% in detailed soil map. For all the surface soil properties except organic matter, the detailed soil map had within-map unit variances that were smaller than the total variance, whereas reconnaissance soil map variance remained high. Soil classification in both the detailed and the reconnaissance mapping units accounted for 5% to 45% and 1.1% to 27% of the total variance for these properties. Sampling demand to predict mean values was much larger in the reconnaissance soil map for most of the topsoil properties. Laboratory studies showed that the detailed soil map units in similar physiographic units could have different soils and vice versa. It appears that the definition of mapping units based on guidelines by Soil Survey Manual is rather optimistic and does not have sufficient reliability for using in sustainable agriculture.

## Introduction

There has been an increasing need for soil information in recent years. Soil maps are used in many studies such as land evaluation, land suitability analysis, land use planning, and geographic information system (GIS) applications. The most common soil survey is second-order soil, surveying at a scale of 1:20,000 (Dent and Young, 1981; Soil Survey Staff, 1993). According to Western (1978) and Bregt et al. (1992), the quality of a map is a function of reliability, relevance, and presentation of the information. Purity and homogeneity characterize the reliability of information in a map (Beckett and Webster, 1971; Bie and Beckett, 1973; Marsman and De Gruijter, 1986). Soil surveys vary greatly in accuracy, detail, complexity, and the type of output (Coover et al., 1975). Purity of soils in mapping units heavily depends upon mapping scale, intensity of sampling, quality of soil description, and presence of soil landscape relationships (Oberthur et al., 1996).

Dent and Young (1981) suggested that the differences between mapping units should be both statistically significant and relevant to land use or management. Perhaps, the most serious limitation of the current soil survey process is the assignment of properties derived from typical or modal soil profiles to the entire map unit regardless of the inherent spatial and temporal variability of field soils (Beckett and Webster, 1971; Baker, 1978; Bouma et al., 1980; Breeuwsma et al., 1986).