An integrated knowledge-based system for grasslands ecosystems

Abstract

In this study, an integrated knowledge-based system (KBS) for the hilly, sandy grassland ecosystems was developed. The system is a component of the Inner Mongolia Grassland Ecosystem Research Information Systems (IMGERIS). IMGERIS is an important subsystem of Chinese Ecological Research Network (CERN). The system was designed to manage land use, planting species/variety, and optimal coverage of plants. The system integrated a KBS for ecosystem with a model system that includes theoretical models of soil water dynamics. The KBS in the system was coded in Prolog and the model system for soil water dynamics was coded in C. The model system was integrated with the KBS.

Introduction

Knowledge-based systems (KBS), as applications of artificial intelligence theory, have been recently adapted in agriculture as well as ecosystem resource management as an effective way to make use of both empirical knowledge of human experts in relevant fields as well as theoretical research results [1], [2], [3], [4]. A KBS is a computer program that emulates problem-solving processes of human experts in a specific field. At its simplest, a KBS consists of a knowledge base and an inference engine [5]. The knowledge base stores rules and facts derived from human experts and are analogous to the memory of human brain. On the other hand, the inference engine performs logical deduction reasoning and knowledge synthesis, and also generates solutions to a specific problem. KBS are especially useful when solution of problems mainly relies on the empirical knowledge of human experts (or in other words, there is not much theoretical knowledge available), when multiple-state solutions may exist, and when solutions involve many aspects or attributes of objects.

Maowusu sandy grasslands are located in the middle and south part of Ertos plateau, at 106°27′–111°28′ east longitude and 37°38′–40°52′ northern latitude. This area has typical continental semi-arid climatic conditions. Grasslands generally receive between 25 and 75 cm of precipitation per year [6]. Annual precipitation of the area ranges from 401.6 mm in the southeast to 162.4 mm in the northwest, with more than 60% of precipitation concentrated in two summer months. Measured annual pan evaporation in the area varies from 2047 to 3085 mm. The annual mean temperature is about 6°C, with monthly mean temperature of 22°C in July and −11°C in January [7], [8], [9]. It is not difficult to postulate a severe hydrological limitation for vegetation development and to realize the importance of water resources for local agricultural management.

The problem of insufficient water supply is further complicated by sandy soil, which has poor water retention capability and large hydraulic conductivity, and the diversity of hilly landscapes reflecting great spacial heterogeneity. The geographical area is meshed with basically four types of landscape components. The components consist of hard hills resulting from erosion and aging of bedrock, soft hills consisting of sediments accumulated during the quaternary period, lower wetlands resulting from cuffing on the quaternary sediments by rivers and streams, and bare sand dunes. The first three types of landscape components can be covered with sands of variable thickness.

In the last few decades, inappropriate agricultural management has caused much degradation of the sandy grassland, with decreased biomass production, vegetation coverage, and increased coverage of bare sand dunes. In recent years, the implication of ecological systems to sustainable agricultural development has begun to be recognized. As a result, a number of research projects have been implemented to investigate the ecosystem dynamics of the hilly sandy grasslands [10], [11], [12], [13], [14], [15].

While observational studies on soil water dynamics in the geographical area were intensively carried out for various particular situations, investigation on seasonal dynamics of the biological aspects of the sandy grassland ecosystems were sparse. Empirical knowledge of local agricultural experts is still the main resource to utilize in grassland management. The complicated site conditions of four different landscape components, coupled with the lack of quantitative mechanistic understanding of adaptability of plants to a specific site, raise an opportunity to build a KBS for local agricultural management.

This paper presents a KBS approach that addresses the above-mentioned issues through an integrated knowledge-based system. Our objective was to develop a KBS that emulates the problem solving processes of human experts by integrating the empirical knowledge of agricultural experts with the models of soil water dynamics. The system was designed to provide policymakers, agricultural advisers and agricultural enterprise managers in Erdos area with decision support in land use, plant species/varieties selection, and optimal coverage of plants.