Management of rangelands by using artificial neural network in Nazlouchai rangelands in West Azarbayjan province

If the rangeland forage is used continuously, the important elements such as NPC do not return to the soil, which will cause the rangeland lands to lose their fertility. Therefore, nowadays, in the field of rangeland management, rangeland improvement and rehabilitation has become very important. The use of fertilizers is one of the methods to rehabilitation the rangelands. If the proper fertilizer application is carried out in accordance with the climatic conditions, cover condition, and soil characteristics, it will improve the rangeland. Otherwise, it will increase the concentration of salts, soil toxicity, and surface and groundwater contamination and leads to drying of the plants. The aim of this study is to present a model based on the use of an artificial neural network that expresses the relationships between organic carbon, nitrogen and phosphorus of rangeland soil and plant factors, based on which, it is possible to estimate the mentioned elements in the rangeland ecosystems without statistics to manage fertilization. Based on the results, organic carbon, nitrogen and phosphorus of the soil were estimated in the Nazlocha rangeland of Urmia. Eight factors of electrical conductivity, acidity, clay percentage, silt percentage, sand percentage, lime content, production and canopy cover percentage of rangeland plants were also selected for factor analysis. Therefore, according to the mentioned results, the neural network was able to accurately predict the amount of organic carbon, nitrogen and phosphorus in rangeland soils. According to the results, the vegetation type Astragalus gummifera-prangos uloptera-Bromus tomentellus requires phosphorus and nitrogen fertilizers. Onobrychis cornuta- Festuca ovina-Thymus kotschyanus requires phosphorus fertilizer, and Astragalus macrostachys-Noeae mucronata-Stipa barbata requires organic matter and nitrogen and phosphorus fertilizers.