پیش‌بینی تغییرات کوتاه مدت و میان مدت شاخص فرسایندگی باران با روش میانگین متحرک خود همبسته‌ی فصلی (SARIMA)

نوع مقاله : مقاله پژوهشی

نویسندگان

1 استادیار گروه خاکشناسی دانشکده کشاورزی دانشگاه شهید چمران اهواز، اهواز- ایران

2 عضو هیئت علمی

10.22055/agen.2022.39147.1621

چکیده

اقلیم یکی از مهمترین عوامل تاثیرگذار بر فرایندهای تشکیل، تکامل و تخریب خاک است که با توجه به تاثیر آن بر رخدادهای فرسایشی، پایش دائمی آن اهمیت دارد. در این پژوهش به منظور ارزیابی تغییرات اقلیمی در قالب سری‌های زمانی چهار ایستگاه هواشناسی شامل اردل، سامان، ایذه و دهدز به‌عنوان ایستگاه‌های مطالعاتی منتخب مورد بررسی قرار گرفت. با بررسی آماری داده‌های بارش در ایستگاه‌های منتخب، محاسبه‌ی شاخص فرسایندگی باران از 1990 تا سال 2017 انجام شد. پس از تحلیل روند داده‌ها و بررسی ایستایی داده‌های بارش، گراف‌های Auto-correlation function (ACF) و Partial auto-correlation function (PACF) ترسیم شد، سپس آزمون دیکی‌فولر (ADF) در سطوح اطمینان 1، 5 و 10 درصد صورت پذیرفت. در مرحله‌ی بعد، عملیات ایستا نمودن داده‌های ناایستا و یافتن پارامترهای مناسب p، r و q انجام شده و مدل Seasonal auto-regressive integrated moving average (SARIMA) تهیه و ایجاد شد. ارزیابی‌های آماری توسط نرم‌افزارهای StataSE، Minitab 18 و SPSS 19 صورت گرفت. نتایج نشان داد که جهت آشکارسازی روند تغییرات بارش، مدل ARIMA (0,0,1)×(1,1,1)12 دارای بیشترین نیکویی برازش است. همچنین روش میانگین متحرک خودهمبسته‌ی فصلی به‌خوبی تغییرات بارش را در منطقه‌ی مطالعاتی واقع در جنوب غرب ایران نشان داد. نتایج حاصل از مدل برازش شده حاکی از احتمال کاهش در میزان بارش سالیانه طی دوره‌های زمانی 5 و 10 ساله پس از سال 2017 می-باشد. با توجه به مقادیر بارش ماهیانه شبیه‌سازی شده، مقدار شاخص فرسایندگی در منطقه به دست آمد که کاهش ضریب فرسایندگی باران طی دوره های تا 10 سال آینده را بیان می‌نماید.

کلیدواژه‌ها


عنوان مقاله [English]

Forecasting of short-term and mid-term variations of rainfall erosivity index using SARIMA

نویسندگان [English]

  • Ataallah Khademalrasoul 1
  • Hadi Amerikhah 2
1 Soil Science Department, Faculty of Agriculture, Shahid Chamran University of Ahvaz, Ahvaz, Iran
2 soil science department
چکیده [English]

Introduction Climate is one of the most effective factors on soil formation, evolution and degradation. It is include different parameters which mainly based on precipitation and temperature. In the recent years the effects of global warming and climate change has extremely enhanced. Climate change as an important phenomenon is effective on precipitation parameters including volume, intensity and concentration which categorized in the temporal and spatial variations. Quantifying the effects of climate change is important for identifying critical regions prone to soil erosion under a changing environment. Land-based ecosystems are influenced by patterns of air temperature and precipitation, which include daily and seasonal changes along with humidity and wind, and the nature of the land surface. Global climate change already has observable effects on the environment. Regarding the importance and effectiveness of climate factor and climate changes during the time, it is essential to focus on climate changes on water behavior at different scales. Indeed, precipitation parameters interacting the soil parameters are influencing on runoff potential in the fields and watersheds. In this regard Rainfall-runoff erosivity (R) is one key climate factor that controls water erosion. Universal soil loss equation (USLE) is the main common equation to predict soil loss, this equation consisting 5 factors which R-Factor (Rainfall erosivity factor) is one of the effective factors in this equation.
Material and Methods Regarding the effect of climate on soil erosion processes therefore, monitoring of climate is really important. In this study in order to evaluate the climate changes based on time series, four climatological stations including, Ardal, Saman, Izeh, and Dehdez were selected. Using the statistical data of precipitation, calculation of eroding index was performed until 2017. The ACF (Auto Correlation Function) and PACF (Partial Auto Correlation Function) for precipitation data were prepared, afterwards the ADF test was performed at confidence level of 1, 5 and 10 percentage. Then the suitable parameters for p, r and q were selected and the SARIMA (Seasonal auto-regressive integrated moving average) model was provided. The statistical analyses were performed with Stata SE, Minitab 18 and SPSS 19. Moreover, the graphical trends of rainfall as an index of precipitation and the rainfall erosivity factor (R-Factor) were presented. Also, the spatial distribution of R-Factor (in the form of GIS-Maps) were provided including three separated maps based on real data, 5 year predicted and 10 year predicted data. So there was a possibility to monitor and compare the spatial distribution of R-Factor at different time periods. Then based on the area, the percentage of rainfall erosivity index was calculated for the study area based on the real data, 5 year predicted and 10 year predicted data. In addition, the statistical parameters including R-square, RMSE, P-value and so on were calculated for the best model (SAR12) regarding all climatological stations.
Results and discussion Our results depicted that to present the trend of precipitation variations as erosive factor the ARIMA (0,0,1)×(1,1,1)12 was the best model. Also, the seasonal autoregressive moving average showed the variation of precipitation in the study area which located in the southwest of Iran. The results of modeling stated that reduction of precipitation for 5 and 10 year periods after 2017. According to amount of monthly simulated of precipitation, the amount of erodibility index was obtained in the area which illustrated the declining trend until 10 year. According to ADF test for all evaluated climatological stations the probability for Ardal was 0.34, for Dehdez was 0.425, for Saman was 0.345 and for Izeh was 0.177, therefore there was difference between climatological stations. Furthermore, the statistical analyses for SAR12 model revealed that the R-square for Ardal station was 0.492, for Dehdez was 0.716, for Saman was 0651 and for Izeh was 0.576. Moreover, approximately 37 % of area has very low rate of erodibility index without previous occurrence.
Conclusion Our results clearly confirmed the importance of climate factors and climate change during the time. As results illustrated regarding the variations of precipitation the R-Factor changed. Moreover, climate change is effective on spatial variations of crop cover in the watersheds. Climate change is capable to alter the crop cover patterns in the watersheds and the changes in crop cover distribution and runoff could change the soil erosion potential. Generally, based on results has to focus on water resources conservation in the study area to preserve soil and water against erosive forces and try to improve the vegetation cover because of decreasing of precipitation. In order to manage the soil resources, we need to monitor the climate changes in the watersheds and try to enhance the vegetation covers in the critical parts on the fields.

کلیدواژه‌ها [English]

  • Rainfall erosivity index
  • SARIMA model
  • precipitation forecasting
  • climate changes
  • Dickey- Fuller test