Abstract:
Advanced information technologies in modern decision support system enables to
improve the decision-making process in agricultural management for land identification.
Agriculture is the key pillar of the economy and ensures food security. Managing soil
fertility for improved production is significantly important. In this regard knowledge
based approach to land evaluation for selecting suitable agricultural cereal crops for a
land unit is necessary. This thesis presents the design and development of a prototype
knowledge based cereal crop land identification (KBCCLI) for evaluating land resources
and choosing suitable agricultural crops for a farm unit. The study was conducted using
empirical research design. The developed prototype system is powered primarily by
human and laboratory experimented data collected from Jimma agricultural and research
center (JARC). Domain knowledge was acquired using interview and questionnaire.
Purposive and stratified sampling technique was employed to get right expert
respondents. The total sample taken was 17 experts and 87 farmers. Decision tree was
used as a knowledge modeling tool and forward chaining method to infer the rule and
provide appropriate direction. The knowledge base consists of representative rules to
reflect the inherent physical and chemical property of soil that affect the choice of land
use. Soil texture is the relative proportions of sand, silt and clay in a soil and is a feature
used to classify land for cereal crop cultivation. The system is developed using SWI
Prolog programming language. The knowledge-based approach to land evaluation is built
on the land evaluation framework designed by the united nation food and agricultural
organization (FAO). The KBCCLI model of land evaluation suggests a strategic land use
plan considering soil physiochemical property which eliminates non-feasible land use or
crop choice. According to the system evaluator the prototype system achieved 87.76% of
overall performance in identifying suitable land for cereal crop cultivation. It is believed
that the prototype system achieved good performance and has potential to use as a
decision tool for land identification for cereal crop.
