A Bayesian decision model for drought management in rainfed wheat farms of North East Iran

Document Type: Research Paper

Authors

1 Associate Professor, Watershed Management Department, Gorgan University of Agricultural Sciences and Natural Resources, Gorgan, Iran.

2 Post Graduate Student, Arid zone Management Department, Gorgan University of Agricultural Sciences and Natural Resources, Gorgan, Iran.

Abstract

Drought is a feature of climate that can occur in virtually all climates. Therefore, it is an
inevitable global but site-specific phenomenon which requires tools to predict and strategies and
options to cope with it. In this research, the ability and effectiveness of the Bayesian Decision
Networks (BDNs) approach in decision-making and evaluating drought management options for
rainfed wheat production in the eastern region of Golestan Province, Iran are demonstrated. The
results revealed that during drought conditions, the Koohdasht cultivar had higher yield than
other cultivars of wheat. Two management scenarios have been specified for the forecasted
period on the basis of wheat cultivars adopted in the region. The results of scenario analysis
with a BDN model indicate that the probability of low, medium and high yield levels in scenario
2 (Koohdasht 70%, Zagros 20% and the other cultivars 10%) has a better status compared with
scenario 1 (current condition). The paired t- test indicates that there is a significant difference
between the two scenarios for wheat yield in low and medium states (P<0.05). Adopting
appropriate cultivars in the region with favourable yield and adaptability to drought conditions
proved to be an effective management action. The BDN approach implemented in this research
serves as a valuable tool to represent the system as a whole, to integrate outputs from models
and expert judgment, to evaluate the outcomes necessary for decision-making and to
communicate uncertainty of the parameters in the model.


Keywords