KAJIAN KERAWANAN LONGSORLAHAN MENGGUNAKAN LOGISTIC REGRESSION MODEL DI DAS KODIL PROVINSI JAWA TENGAH

ELOK SURYA PRATIWI, Dr. Danang Sri Hadmoko, M.Sc

2013 | Skripsi | GEOGRAFI DAN ILMU LINGKUNGAN

DAS Kodil yang terletak di Provinsi Jawa Tengah adalah salah satu kawasan rawan longsorlahan yang patut untuk diwaspadai. Tingginya jumlah kerugian yang ditimbulkan dari kejadian bencana longsorlahan terdahulu membuktikan bahwa upaya mitigasi dan manajemen bencana di daerah tersebut masih belum optimal. Kajian mendalam mengenai longsorlahan sangat penting dilakukan untuk tujuan pengurangan risiko bencana. Tujuan penelitian ini adalah: (1) Menganalisis dinamika persebaran kejadian longsorlahan secara temporal di DAS Kodil; (2) Menyusun peta kerawanan longsorlahan menggunakan metode Logistic Regression Model. Grafik jumlah curah hujan dan jumlah kejadian longsorlahan dibuat untuk menganalisis persebaran longsorlahan secara temporal bulanan dan harian di DAS Kodil. Peta kerawanan longsorlahan di susun menggunakan Logistic Regression Model (LRM). Delapan variabel yang digunakan untuk merekonstruksi peta kerawanan longsorlahan adalah kemiringan lereng, jarak dari sungai, jarak dari jalan, geologi, permeabilitas, tekstur, ketebalan material longsorlahan dan penggunaan lahan. Hubungan antara curah hujan dan kejadian longsorlahan menunjukkan bahwa curah hujan memiliki peranan penting dalam kejadian longsorlahan di DAS Kodil. Longsorlahan hanya terjadi pada bulan bulan basah saja. Hujan yang tidak terlalu lebat namun terjadi secara terus menerus dapat memicu terjadinya longsorlahan di daerah tersebut. Peta kerawanan yang dihasilkan melalui tahapan logistic regression dan eliminasi variabel yang tidak signifikan memiliki tingkat akurasi sebesar 69,50%. Jarak dari jalan merupakan variabel yang paling signifikan mempengaruhi longsorlahan di daerah tersebut.

Kodil Watershed where located in Central of Java Province is considered as landslide prone area. High numbers of loss from the previous landslide hazard proved that hazard management and mitigation in that area is still low. In-depth study about landslide susceptibility is very important for risk reduction purpose.. The aim of this research are: (1) to analyze the dynamic of landslide’s temporal distribution in Kodil Watershed; (2) to build landslide susceptibility map by using Logistic Regression Model. Graphic of rainfall and number of landslide occurrence was made to analyze temporal landslide distribution (monthly and daily) in Kodil Watershed. Landslide susceptibility map was built by Logistic Regression Model (LRM). Eight variables were used to build landslide susceptibility map, i.e. slope, distance from river, distance from road, geology, permeability, soil texture, landslide material thickness, and land use. The relationship between rainfall and landslide occurrence showed that rainfall plays an important role to the landslide occurrence in Kodil Watershed. Landslide occurred only in the wet months. Rainfall with a low intensity and long duration can trigger the landslide occurrence in that area. The landslide susceptibility map generated with logistic regression and insignificant variable elimination step reached an accuracy as high as 69,50%. Variable of Distance from road performs as the biggest influence for landslide occurrence in that area.

Kata Kunci : Kerawanan Longsorlahan, Logistic Regression Model, Dinamika Temporal Longsorlahan