Publication: Samsun Bölgesi Yeraltı Suyu Kalitesi Verileri Üzerine Zaman Serileri Analizinin Uygulanması
Abstract
Ill ÖZET Bu çalışmada ilk olarak > Samsun bölgesi yeraltısulanmn kalitesini belirlemek amacıyla Mayıs 1991 -Ekim 1992 dönemleri arasında yapılan aylık kalite ölçümlerinden elde edilen veriler üzerine zaman serileri analizi (ZSA) uygulanmadan önce, eksik değerler enterpolasyonla tamamlandı. Analizler öncesinde ilk olarak veri setlerinin zaman serisi grafikleri çizildi. Serinin genel davranışı, mevsimsellik içerip içermediği ve salınım periyodu (mevsimsellik varsa mevsim uzunluğu) tespit edildi. Bu amaçla serinin durağansızlığının giderilmesine çalışıldı. İlgili hesaplamalar ve grafikler yardımıyla otokorelasyon analizi (otokorelasyon ve kısmi otokorelasyonların yorumu) yapıldı. Bu analizler sonucunda Box ve Jenkins tarafından geliştirilen ARIMA modellerinden uygun olabilecekler seçilerek uygunluk testine tabi tutuldu. Serinin karakteristiğine en uygun modelin tespitinden sonra öngörü (tahmin) çalışmaları yapıldı. % 95 güven aralığı sınırlan içindeki öngörü sonuçlarına ulaşıldı. Öngörü çalışmaları neticesinde oldukça yüksek güvenilirliğe sahip (birkaç değer dışında çoğunlukla %90 civarlarında) değerlere ulaşıldı. Öngörüsü yapılan değerler ile öngörü sonuçları birbiriyle oldukça paralel ilişkide bulundu ve kayda değer sonuçlar elde edildi. Öngörü değerleri yeraltısuyu kirliliği kontrol ve yönetiminde kullanılacak düzeyde uygun bulundu. İşlemler esnasında yapılan kabuller ve bazı kuyulara ilişkin verilerin iptal edilme nedenleri açıklandı. Kuyulara ilişkin kalite değerlendirmeleri grafik ve sınıflandırmalar yardımıyla yapıldı ve elde edilen bulgular değerlendirildi.
IV ABSTRACT In this study, insufficient groundwater quality test data for Samsun district which had been performed in between May 1991 and October 1992 have been completed for time series analyses by using interpolation techniques. During these analyses firstly, time series graphics of obtained data were plotted. Then general behavior characteristics of time series, especially seasonal variations on the measured data were determined. If there is a seasonal variations, time intervals of the affected seasons were resolved. In order to perform these analyses, nonstationarity of the series were tried to be eliminated. Related calculations and graphics were used to autocorrelate the analyses (for autocorrelation analyses and partial autocorrelation analyses). Best fit ARIMA models which were developed by Box and Jenkins were selected and the models were tested for their suitability to the test data. Forecasting calculations were then accomplished after the determination of the model which fits with the general characteristics of the time series. The results of the forecasting had been reached with 95 % confidence interval (except a few values, they are generally more than 90 % confidence interval). Forecasting analyses' results and the data used for these analyses were showed parallel relations therefore, outstanding conclusions were determined from these analyses. The values obtained from forecasting calculations were detected that they were at required level to use in groundwater contamination analyses and its management methods. In addition to these studies, the assumptions used in the analyses and the reasons of certain well data elimination were explained. At the end, groundwater quality analyses were discussed by using related graphics and classifications. The results of the analyses were then concluded.
IV ABSTRACT In this study, insufficient groundwater quality test data for Samsun district which had been performed in between May 1991 and October 1992 have been completed for time series analyses by using interpolation techniques. During these analyses firstly, time series graphics of obtained data were plotted. Then general behavior characteristics of time series, especially seasonal variations on the measured data were determined. If there is a seasonal variations, time intervals of the affected seasons were resolved. In order to perform these analyses, nonstationarity of the series were tried to be eliminated. Related calculations and graphics were used to autocorrelate the analyses (for autocorrelation analyses and partial autocorrelation analyses). Best fit ARIMA models which were developed by Box and Jenkins were selected and the models were tested for their suitability to the test data. Forecasting calculations were then accomplished after the determination of the model which fits with the general characteristics of the time series. The results of the forecasting had been reached with 95 % confidence interval (except a few values, they are generally more than 90 % confidence interval). Forecasting analyses' results and the data used for these analyses were showed parallel relations therefore, outstanding conclusions were determined from these analyses. The values obtained from forecasting calculations were detected that they were at required level to use in groundwater contamination analyses and its management methods. In addition to these studies, the assumptions used in the analyses and the reasons of certain well data elimination were explained. At the end, groundwater quality analyses were discussed by using related graphics and classifications. The results of the analyses were then concluded.
Description
Citation
WoS Q
Scopus Q
Source
Volume
Issue
Start Page
End Page
170
