نوع مقاله : مقاله پژوهشی

نویسندگان

• حلیمه پیری
• مجتبی مبارکی

دانشگاه زابل

چکیده

روش‌های مختلفی بـرای تخمـین خـصوصیات خـاک در منـاطقی کـه نمونه‌برداری نشده‌اند وجود دارد که روش‌های زمین‌آماری و هوش مصنوعی از جملـه آن‌هـا می‌باشند. در ایـن پژوهش از روش زمین‌آمار و مدل‌های هوش مصنوعی (شبکه عصبی مصنوعی، درخت تصمیم و ماشین بردار پشتیبان) برای پیش‌بینی شوری خاک اراضی قرقری شهرستان هیرمند استفاده شد. برای انجام کار تعداد 130 نمونه خاک از عمق 30-0 سانتی‌متری خاک بر اساس شبکه‌‌بندی منظم با ابعاد 1500 ×1500متر برداشت شد. نمونه‌های برداشت شده به آزمایشگاه منتقل گردید. در آزمایشگاه هدایت الکتریکی با استفاده از دستگاه هدایت الکتریکی‌سنج اندازه‌گیری شد. در ابتدا با استفاده از روش زمین آمار روند شوری مورد بررسی قرار گرفت. با استفاده از روش زمین‌آمار و روش‌های هوش مصنوعی مقادیر شوری خاک برآورد شد. مدل‌های زمـین‌آمـاری و هوش مصنوعی برازش و بهترین مدل انتخاب و با استفاده از اعتبارسنجی مستقل دقت روش‌ها با هم مقایسه گردید. نتایج نشان داد روش‌های هوش مصنوعی نسبت به روش زمین‌آمار شوری خاک را بهتر برآورد می‌کنند. بین روش‌های هوش مصنوعی مدل درخت تصمیم با توجه به ضریب تبیین بالا (99/0) و آماره‌های خطای پایین (26/0 و 18/0) به‌عنوان مدل برتر انتخاب شد. نتایج روند شوری نشان داد شوری خاک منطقه از از غرب به شرق ابتدا کاهش و سپس افزایش و از شمال به جنوب کاهش می‌یابد. با توجه به نتایج به‌دست آمده پیشنهاد می‌شود برای حفظ محیط زیست منطقه، نقشه شوری خاک و مقادیر شوری با استفاده از مدل برتر ترسیم شود و زمینه کاشت گونه‌های گیاهی سازگار با منطقه، منطبق با شوری خاک فراهم شود.

کلیدواژه‌ها

• درخت تصمیم
• ماشین بردار پشتیبان
• شبکه عصبی مصنوعی
• روند شوری

موضوعات

• فیزیک و حفاظت خاک

عنوان مقاله [English]

Comparison of Artificial Intelligence and Geostatistical Methods in Soil Surface Salinity Prediction (Case study: Ghorghori of Hirmand city)

نویسندگان [English]

• Halimeh Piri
• mojtaba mobaraki

University of zabol

چکیده [English]

Soil salinity is one of the factors limiting production in agricultural lands. Preparing accurate soil salinity maps can improve land management practices. There are various methods for estimating soil properties in areas with limited observational data, Ggeostatistical and artificial intelligence methods. In this study, Ggeostatistical methods and artificial intelligence models (artificial neural network, decision tree and support vector machine) were used to simulate the soil salinity of Ghorghori lands in Helmand city. To do the work, A total of 130 soil samples were callected from 0-30 cm layer of the soil from a network with dimensions of 1500 × 1500 meters. The samples electrical conductivity was measured using an electrical conductivity device in the laboratory. At first, salinity trend was studied using the Geostatistical method. Soil salinity values were estimated using Geostatistical methods and artificial intelligence methods. Geostatistical and artificial intelligence models were applied and the best model was selected and the accuracy of the methods were compared using independent validation. The results showed that the artificial intelligence methods outperformed Geostatistical method in estimating the soil salinity of the artificial intelligence methods, the decision tree model as the superior model due to its high coefficient of determination (0.99) and RMSE and MAE statistics (0.26 and 0.18) respectively. The results of salinity trend showed that the salinity of the soil of the region decreases from west to east first and then increases and decreases from north to south. Increasing wind intensity in this part and severe wind erosion has caused destruction and instability of soil structure and has greatly reduced vegetation and increased soil salinity. According to the obtained results, it is suggested to draw a soil salinity map and salinity values using the superior model to preserve the environment of the region and pave the ground for planting species compatible with the area soil salinity.

Soil salinity is one of the factors limiting production in agricultural lands. Preparing accurate soil salinity maps can improve land management practices. There are various methods for estimating soil properties in areas with limited observational data, Ggeostatistical and artificial intelligence methods. In this study, Ggeostatistical methods and artificial intelligence models (artificial neural network, decision tree and support vector machine) were used to simulate the soil salinity of Ghorghori lands in Helmand city. To do the work, A total of 130 soil samples were callected from 0-30 cm layer of the soil from a network with dimensions of 1500 × 1500 meters. The samples electrical conductivity was measured using an electrical conductivity device in the laboratory. At first, salinity trend was studied using the Geostatistical method. Soil salinity values were estimated using Geostatistical methods and artificial intelligence methods. Geostatistical and artificial intelligence models were applied and the best model was selected and the accuracy of the methods were compared using independent validation. The results showed that the artificial intelligence methods outperformed Geostatistical method in estimating the soil salinity of the artificial intelligence methods, the decision tree model as the superior model due to its high coefficient of determination (0.99) and RMSE and MAE statistics (0.26 and 0.18) respectively. The results of salinity trend showed that the salinity of the soil of the region decreases from west to east first and then increases and decreases from north to south. Increasing wind intensity in this part and severe wind erosion has caused destruction and instability of soil structure and has greatly reduced vegetation and increased soil salinity.

کلیدواژه‌ها [English]

• Artificial neural network
• Decision Tree
• Salinity Trend
• Support Vector Machine