In most part of the world the information on the thematic soil maps (soil erosion, soil degradation, soil organic matter content etc.) are developed as a tool for policy and management support. This information are typically derived through expert interpretation or empirical modeling approaches using typically decades old soil information originating from field investigation, laboratory analysis, reports etc. In recent period, there is a strong emphasis to update the existing soil information in a cost-effective and accurate manner. The advancements in the emerging Geographical Information System (GIS) and digital soil mapping techniques are found to be handy to derive tools addressing the above mentioned problem. In this study, we propose a novel innovative approach to address the issues on evaluating the traditional soil maps and updating the existing soil information based on the principles of digital soil mapping i.e. deriving objective soil information by reformulating the relationships between soil and its environment using ancillary and minimal datasets. The new approach is called as "SEIMS network" (Soil and Environment Interaction based Mapping System). The SEIMS network is a Data Mining and Knowledge Discovery (KDD) based method derived by fusing GIS, DSM techniques along with working principle of the Diagnosis and Recommendation Integrated System (DRIS) approach. The SEIMS network based approach provides a scheme to transform and update the less detailed, discrete and subjectively derived soil information into a continuous, non-subjective and quantitative spatial datasets. The study was tested on the soil erosion map of Tamil Nadu region, southern part of Indian Peninsula. The SEIMS network attempts to reformulate the soil erosion map of Tamil Nadu region and redefine the contours by spreading back the knowledge acquired from the relationship among the soil and its environmental variables used as predictors in the system. The variables like temperature, rainfall, potential evapotranspiration, rainfall seasonality, land cover percentage (derived from MODIS spectral bands), soil crusting, soil erodibility, top soil organic carbon content, altitude and slope that are having major influence on soil erosion were chosen as predictors for characterizing the relationship among the variables on the context of soil erosion process. The test on the efficiency of the SEIMS network's capability to extrapolate and derive the target soil information using the ancillary datasets was performed over established soil erosion map of Europe. The erosion index value derived through SEIMS network scheme exhibited a better correlation with PESERA soil erosion estimates (r2 = 0.81), thereby proving its ability to mimic and characterize the soil erosion process by studying the complex interrelationship among the environmental variables. The weights derived for prediction are mathematically unique, thereby holds scope for further elaboration on its application to derive a tool, addressing the upscaling and downscaling issues in digital soil mapping. The flexibility and reproducibility are the main advantages visualized for this approach. Moreover, the results are objective and easy for interpretation. This study demonstrated that the SEIMS network holds some potential features that can be exploited in digital soil mapping to evaluate and update the existing soil information with minimal ancillary datasets.
