Publication: Kümeleme Analiz Yöntemleri İle Karadeniz Bölgesi Meteorolojik Verilerinin Değerlendirilmesi
Abstract
Son yıllarda küresel iklim değişikliği etkilerinden kaynaklanan afet sayılarında hissedilebilir bir artış görülmektedir. Bu kapsamda iklim değişikliği etkilerini azaltmak amacıyla ülkemizde ve dünyada çeşitli çalışmalar yapılmaktadır. İklim değişikliğinden etkilenen bölgelerin iklim parametreleri bakımından benzer sınıflara ayrılması bu bölgelerde yapılacak olan çalışmalarda benzer yöntemlerin uygulanması açısından önemlidir. Böylece iklim değişikliğinin etkilerini azaltmak amacıyla yapılacak olan çalışmalarda doğru bir stratejinin belirlenmesi sağlanacaktır. Çalışma kapsamında değerlendirilen gözlem kayıtları Meteoroloji Genel Müdürlüğüne ait Karadeniz Bölgesinde yer alan 31 istasyondan periyodu 1982-2020 yılları arasını kapsayacak şekilde yıllık toplam yağış, yıllık ortalama sıcaklık ve yıllık ortalama rüzgâr hızı serileri olarak temin edilmiştir. Çalışmada öncelikle Mutlak Homojenlik Testleri (SNHT, BST, PT ve VNOT) ile verilerin homojenlikleri tespit edilmiştir. Sonrasında homojenlik sağlamayan istasyonlarda eğilim olup olmadığını araştırmak amacıyla Mann-Kendall ve Spearman'ın Rho testleri ile trend analizi gerçekleştirilmiştir. Daha sonra maksimum küme sayı 5 olarak tespit edilen çalışmada veriler yağış, sıcaklık, rüzgâr hızı ve bu üç veri birlikte matris oluşturacak şekilde 4 faklı parametreye göre 2, 3, 4 ve 5 küme sayısı için Bulanık C-Ortalamalar ve K-Ortalamalar yöntemleri kullanılarak kümeleme analiz çalışması gerçekleştirilmiştir. Çalışma sonucunda optimum küme sayısı Siluet indeks analizi ile tespit edilmiştir. Sonuç olarak yağış serileri için en uygun sınıflandırma, küme sayısı 4 seçilerek Bulanık C-Ortalamalar yöntemi ile elde edilmiştir. Sıcaklık serileri için en uygun sınıflandırma, küme sayısı 5 seçilerek K-Ortalamalar yöntemi ile elde edilmiştir. Rüzgâr hızı serileri için en uygun sınıflandırma, küme sayısı 5 seçilerek K-Ortalamalar yöntemi ile elde edilmiştir. Yağış, sıcaklık ve rüzgâr hızının birlikte değerlendirildiği veri matrisi için en uygun sınıflandırma, küme sayısı 4 seçilerek K-Ortalamalar yöntemi ile elde edilmiştir.
In recent years, there has been a noticeable increase in the number of disasters caused by the effects of global climate change. In this context, various studies are carried out in Turkey and in the world in order to reduce the effects of climate change. The classification of regions affected by climate change into similar classes in terms of climate parameters is important for the application of similar methods in studies to be carried out in these regions. Thus, a correct strategy will be determined in the studies to be carried out to reduce the effects of climate change. Observation records evaluated within the scope of the study were obtained from 31 stations of the General Directorate of Meteorology in the Black Sea Region, covering the period between 1982 and 2020, as the series of annual total precipitation, annual average temperature and annual average wind speed. Primarily in the study, the homogeneity of the data was determined by Absolute Homogeneity Tests (SNHT, BST, PT and VNOT). Afterwards, trend analysis was carried out by Mann-Kendall and Spearman's Rho tests in order to investigate whether there is a trend in the stations that do not provide homogeneity. Then, in the study, the maximum number of clusters was determined as 5; Clustering analysis study was carried out by using Fuzzy C-Means and K-Means methods for 2, 3, 4 and 5 cluster numbers according to 4 different parameters in a way that precipitation, temperature, wind speed and these three data together form a matrix. The determination of the optimum cluster numbers was carried out by Silhouette index analysis. As a result, the most appropriate classification for the precipitation series was obtained by using the Fuzzy C-Means method by choosing the number of clusters as 4. The most suitable classification for temperature series was obtained by K-Means method by choosing the cluster number as 5. The most suitable classification for the wind speed series was obtained by the K-Means method by choosing the number of clusters as 5. For the data matrix where precipitation, temperature and wind speed are evaluated together, the most appropriate classification was obtained by K-Means method by choosing the number of clusters as 4.
In recent years, there has been a noticeable increase in the number of disasters caused by the effects of global climate change. In this context, various studies are carried out in Turkey and in the world in order to reduce the effects of climate change. The classification of regions affected by climate change into similar classes in terms of climate parameters is important for the application of similar methods in studies to be carried out in these regions. Thus, a correct strategy will be determined in the studies to be carried out to reduce the effects of climate change. Observation records evaluated within the scope of the study were obtained from 31 stations of the General Directorate of Meteorology in the Black Sea Region, covering the period between 1982 and 2020, as the series of annual total precipitation, annual average temperature and annual average wind speed. Primarily in the study, the homogeneity of the data was determined by Absolute Homogeneity Tests (SNHT, BST, PT and VNOT). Afterwards, trend analysis was carried out by Mann-Kendall and Spearman's Rho tests in order to investigate whether there is a trend in the stations that do not provide homogeneity. Then, in the study, the maximum number of clusters was determined as 5; Clustering analysis study was carried out by using Fuzzy C-Means and K-Means methods for 2, 3, 4 and 5 cluster numbers according to 4 different parameters in a way that precipitation, temperature, wind speed and these three data together form a matrix. The determination of the optimum cluster numbers was carried out by Silhouette index analysis. As a result, the most appropriate classification for the precipitation series was obtained by using the Fuzzy C-Means method by choosing the number of clusters as 4. The most suitable classification for temperature series was obtained by K-Means method by choosing the cluster number as 5. The most suitable classification for the wind speed series was obtained by the K-Means method by choosing the number of clusters as 5. For the data matrix where precipitation, temperature and wind speed are evaluated together, the most appropriate classification was obtained by K-Means method by choosing the number of clusters as 4.
Description
Citation
Source
Volume
Issue
Start Page
End Page
112