چکیده:
فرسایش شدید خاک، تهدیدی جدی برای مدیریت پایدار سرزمین و استفاده از منابع آب و خاک در بسیاری از نقاط جهان است. بهمنظور کنترل فرسایشهای ورقهای، شیاری، خندقی و آبراههای و کاهش رسوب تولیدی ناشی از آنها در خروجی حوضههای آبخیز، لازم است به شناسایی سهم منابع تولیدکننده رسوب آنها پرداخت تا اقدامات حفاظتی با موفقیت بیشتری انجام شود. یکی از متداولترین روشهایی که در سالهای اخیر از آن بهمنظور تعیین سهم منابع مختلف رسوب استفاده شده، روش انگشتنگاری رسوب است. هدف از این پژوهش، بررسی سهم منابع تولیدکننده رسوب ناشی از فرسایشهای ورقهای، شیاری، خندقی و آبراههای با استفاده از این روش در حوضه آبخیز نیریز واقع در شرق استان فارس به کمک نمونهبرداری از رسوب نهشته شده در بستر است؛ بنابراین از هر نوع از رسوبات فرسایشهای ورقهای، شیاری، خندقی، آبراههای، آبراهه اصلی درون حوضه و منطقه خروجی حوضه آبخیز، ده نمونه (در مجموع شصت نمونه) برداشت شد. بهمنظور تعیین ردیابهای بهینه نیز از دو آزمون دامنه و تحلیل تشخیص چند متغیره استفاده شد و با استفاده از مدل کولینز و همکاران، سهم هر یک از منابع مختلف رسوب بهدست آمد. سپس فقدان قطعیت مرتبط با سهم منابع بالقوه رسوبات، با استفاده از روش شبیهسازی مونتکارلو با اطمینان 95 درصد در نرمافزار MATLAB محاسبه شد. بهمنظور ارزیابی نتایج حاصل از مدل چند متغیره ترکیبی، از نکویی برازش (GOF) پیشنهادی توسط کولینز و همکاران استفاده شد. یافتههای این پژوهش نشان داد که چهار ردیاب (Zr، Al، Sn و Lu) بهعنوان ردیابهای بهینه نهایی انتخاب شدند. بهعلاوه میزان سهم فرسایشهای خندقی، ورقهای، شیاری و آبراههای بهترتیب برابر با 21/45، 07/3، 16 و 72/35 درصد از کل فرسایشهای اتفاق افتاده در این حوضه آبخیز بود. در این پژوهش، کارایی روش انگشتنگاری رسوب بهعنوان روشی موفق و مؤثر در تعیین منابع رسوبات اثبات شد؛ زیرا چهار ردیاب بهینه توانستند 95 درصد منابع رسوب را به درستی طبقهبندی و جداسازی کنند. همچنین با توجه به مقدار 8869/0 GOF نیز دقت بالای مدل را تأیید کرد.
1- Introduction
Severe soil erosion is a serious threat to the sustainable management of land and the use of water and soil resources in many parts of the world. In order to control erosion of sheet, rill, gully, and stream bank erosions and to reduce the resulting sediment at the outlet of watersheds, it is necessary to identify the share of sources that produce their sediment to make protective measures more successful. One of the most common methods that has been used in recent years to determine the share of different sources of sediment is the sediment fingerprinting method.
2- Methodology
The purpose of this study is to investigate contribution of sheet, rill, gully and stream bank erosions in sediment production by using sediment fingerprinting method in Neyriz watershed, located in East of Fars province, with the help of sampling of sediment deposited in the bed. From each type of sediments, sheet, rill, gully and stream bank erosions, the main waterway within the basin and the outlet area of the watershed, 10 samples (60 samples in total) were collected. In order to determine the optimal tracers, two tests of "domain" test and "multivariate detection analysis" were used. Furthermore, by using the model of Collins et al., the share of each of the different sources of sediment was obtained. Then, the uncertainty related to the share of potential sources of sediments was calculated using the Monte Carlo simulation method with 95% confidence in MATLAB software. In order to evaluate the results of the hybrid multivariate model, the Goodness of Fit (GOF) proposed by Collins et al. was used.
3- Results
Based on the range test, among the 51 tracers measured in the samples, twelve tracers (Ag, Ba, Be, Eu, Mn, Ni, Ta, Tb, Th, Tm, W, and Zn) are found as tracer’s non-conservative variables were identified, and these detectors were discarded in other statistical tests such as Kruskal-Wallis and discriminant analysis function. The results of the Kruskal-Wallis test showed that among the 39 tracers that passed the range test, sixteen tracers (Al, Ca, Co, Cr, Er, Fe, Gd, Lu, Mo, Na, Nd, Pb, Pr, S , Sc and Zr) with significance at one percent level (p ≤ 0.01), and 9 tracers (Cu, Ga, Hf, Ho, La, Sn, Sr, Y and Yb) with significance at five percent level (p ≤ 0.05) is that in total, these 25 detectors had a significant level and could separate sources; while fourteen tracers (As, Ce, Cs, Dy, K, Li, Mg, Nb, P, Rb, Sm, Te, Ti and V) were not statistically significant, these tracers were deleted from the DFA statistical test. In the first step of the DFA test, the Zr detector, the second step of Zr and Al detectors (with Wilkes lambda from 0.717 to 0.244), the third step of Al, Zr and Fe detectors (with Wilks lambda from 0.39 to 0.057), the fourth step of Zr, Al, Fe and Sn detectors (with Wilkes-lambda 0.362 to 0.04), the fifth step of Zr, Al, Fe, Sn and Lu detectors (with Wilkes-lambda 0.233 to 0.03) and the sixth step Zr, Al, Sn and Lu tracers (with Wilks lambda 0.289 to 0.045) were entered into the model. Based on the obtained results, among the 25 tracers that passed the Kruskal-Wallis test, five tracers (Al, Fe, Lu, Sn and Zr) were entered into the DFA test step by step. In the third stage, iron tracer (Fe) was added to the model and in the sixth stage, it was removed from the DFA test. In general, four Zr, Al, Sn and Lu tracers were selected as the final optimum tracers. These four detectors were able to correctly classify 95% of sediment sources. The findings of this research, which were obtained by using Monte Carlo simulation and the combined multivariable model and evaluating their results using GOF, showed the contribution of gully, sheet, rill and stream bank erosion to the order is equal to 45.21, 3.07, 16 and 35.72% of the total erosions that have occurred in this watershed. Also, considering the GOF value of 0.8869 and mentioning that the closer this value is to one, the more accurate the results of the model is true in this research and this analysis also confirms the high accuracy of the model.
4- Discussion & Conclusions
In this study, the efficiency of sediment fingerprinting method was proved as a successful and effective method to determine sediment sources because the first and most important stage of the sediment source method is to choose a suitable combination of tracers that can isolate sediment sources, and this was done correctly in this research. Also, Monte Carlo uncertainty confidence levels showed that the scope of this uncertainty is large (0.8869) and therefore, it shows a greater lack of certainty on different sources of sediment production. Determining the share of four types of erosion in the Neyriz watershed and placing the share of gully erosion as the most important type of erosion in the production of productive sediments in it shows the importance of controlling erosions, especially gully erosion, with emphasis on biological plans.