9681

8925

كارشناسي ارشد

بيابان زايي

اصفهان: دانشگاه صنعتي اصفهان، دانشكده منابع طبيعي

1393

پانزده،149ص.: مصور،جدول،نمودار

ص.ع.به فارسي و انگليسي

رضا جعفري، حسين بشري

93/12/23

حسين مرادي، سعيد پور منافي

مهندسي منابع طبيعي

ID8925

11 5 1 3 مزيت هاي مدل بيزين 721 5 1 4 معايب مدل بيزين 721 5 2 نتيجهگيري كلي در هر بخش از اين تحقيق 821 5 2 1 مدل مدالوس 821 5 2 1 مدل شبكه هاي باور بيزين 121 5 3 پيشنهادات 121 پيوست ها 131 مراجع 141 چكيده انگليسي 151 يازده

151 Mapping and Assessment of Desertification Using MEDALUS Model and Bayesian Belief Networks Abdolhossein Boali Email address hossien boali@yahoo com Department of Natural Resources Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc language Farsi Reza Jafari reza jafari@cc iut ac ir Hossien Bashari hbashari@cc iut ac ir Abstract Desertification refers to the conditions that soil loses its fertility forever due to ecologicaland management changes This phenomena has widely occurred in different regions withvarious climatic conditions throughout the world Till now scientist have utilized variousmodels to assess desertification processes including FAO UNEP MEDALUS ICD MICDand IMDPA In this study for the first time in Iran we used Bayesian Belief Networks BBN along with MEDALUS to assess desertification processes While a desert manager facesuncertainty in results and predictions obtained from different assessing models it iscompletely cumbersome for him to take a proper action in order to truly manage a desertregion BBNs can accommodate uncertainties in its prediction properly This study aimed tointegrate BBNs and MEDALUS models to assess desertification level as well as introducing adecision support tool in 30000 ha areas of Seghzi region located in Isfahan province Regarding literature review and field assays six criteria including climatic condition vegetative cover soil condition underground water quality wind erosion and managementpolicies were selected to build the models Then we defined the most important indicesaffecting each criteria and weighted based on the MEDALUS procedure The status of all theused indices and criteria in MEDALUS were mapped using geometric mean in Arc GIS9 Allthe layers were then combined to produce the risk of desertification in the study area BBNsapproach was used to sketch cause and effects models for soil condition underground waterand wind erosion status criteria The conditions of the geomorphologic units were inserted inthese models and the soil condition underground water and wind erosion status of theseworking unites were evaluated Eventually we introduced the final desertification model forthe study area In order to have a comparison between results obtained from MEDALUS andBBNs we employed the same variables to construct the models The results of BBNs modelswere verified and compared with MEALUS and sensitivity analysis were also used to evaluatethe BBN models performance MEDALUS showed that 13 and 78 87 of the study area isintensively and severely desertified respectively The sensitivity analysis of the modelsrevealed that climatic conditions and wind erosion status are the two most influencing factorsto lead the region through desertified conditions This study showed that BBNs due to theirabilities to take the cause and effects relationships between desertification variables intoaccount as well as accommodating uncertainty of the predicted system can work as a valuabletool to predict desertification conditions Therefore it could be considered as a decisionsupport tool for decision makers to manage desert regions Keywords desertification MEDALUS Bayesian belief Networks modeling sensitivityanalysis decision support tool

رضا جعفري، حسين بشري

حسين مرادي، سعيد پور منافي