基于GIS、模糊逻辑和专家知识的土壤制图及其在中国应用前景

详细的土壤空间与属性的信息已成为环境模型和土地管理的基本参数,传统的以类别多边形和手工编制为基础的传统土壤制图效率低精度也较差。本文基于GIS、模糊逻辑和专家知识,建立了土壤一环境推理模型(SoLIM),通过基于土壤一环境关系模型的土壤相似度模型与对该模型进行赋值的推理技术来编制土壤图,从而克服了传统土壤制图中的简化。通过两个小区的研究表明,与传统土壤制图相比,通过SoLIM得出的土壤信息在空间详细度和属性精确度都有较大的提高,也能够大量减少调查的时问和经费,从而大大提高土壤调查的效率。SoLIM方法在我国推广十分必要且具有一定的条件,但仍需要进一步完善。

PREDICTIVE SOIL MAPPING BASED ON A GIS, EXPERT KNOWLEDGE, AND FUZZY LOGIC FRAMEWORK AND ITS APPLICATION PROSPECTS IN CHINA

Detailed soil spatial and attribute information are now basic parameters for environmental modeling and land management applications. The accuracy and efficiency of conventional soil surveys, based on the polygon model and the manual mapping practice, are quite low. A geographical information system (GIS) and expert knowledge based-fuzzy soil inference scheme (soil-land inference model, SoLIM) was developed to overcome the problems faced by the conventional soil survey. The scheme consists of three major components: (i) a model employing a similarity representation of soils, (ii) a set of inference techniques for deriving similarity representation, and (iii) application of the similarity representation. According to case studies conducted in Wisconsin, U. S. A, SoLIM improves the accuracy and efficiency of soil survey. Soil type and properties maps based on SoLIM are better than these based on conventional techniques in term of both spatial detail and attribute accuracy. The accuracy of the soil series maps based on SoLIM is about 80% . Moreover, soil mapping by means of SoLIM is about ten times faster than by conventional ones and saves about 2/3 in cost. However, how the SoLIM works highly depends on the availability and quality of environmental data and the quality of soil, environmental relationship model for the study area. The Second Soil Survey in China was conducted 20 years ago. Recent intensive land use activities may have greatly impacted the soil conditions. It' s quite necessary to update the soil data in China for agricultural purposes. The SoLIM framework can provide great assistance in updating soil resource inventory. At the same time, the availability of spatial data, spatial information processing technology and human resources related to GPS, GIS and RS also make it possible to apply the SoLIM approach in the Chinese soil survey. However, the degree of success of the SoLIM highly depends on the quality of knowledge on soil environmental relationships in the study area. The knowledge about soil-environmental relationships used for the SoLIM in USA comes mainly from local soil survey experts. Although experts in soil taxonomy can provide much help for soil survey in China, the number of experts who focus on soil mapping in local areas is small, which will pose a major problem in using the SoLIM approach in China due to its reliance on human expertise for soil-environmental relationships. If we try to discern the soil-environmental relationships by field sampling, we will need a large amount of samples, which is too expensive and too slow. Thus, we must develop new approaches for extraction of knowledge on soil-environmental relationships. These new approaches could include new sampling methods, such as purposive sampling to reduce fieldwork and data mining for extracting knowledge from existing soil maps.