Statistical Downscaling Modeling Through K-means Clustering
Keywords:
K-Means, Partial Least Square, Principal Component Regression, Statistical DownscalingAbstract
Statistical downscaling (SD) is a technique used to describe the relationship between data on a global grid (predictor) with data on a local scale grid (response) to translate global-scale anomalies into anomalies of some local climate variables. SD modeling using GCM outcomes involves many dependent variable (high correlation). In this study used the principal component analysis ( PCA ) and the partial least squares (PLS) to overcome the multicollinearity problems that occur in the dependent of GCM output data, and using K-means clustering as mediator to minimize variety that occurs in GCM data and local rainfall. SD modeling was applied on each group formed for each local rainfall station using principal component regression (PCR) and PLS. This study aims to determine the best model between PCR and PLS in stastistical downscaling by using K-means as mediator for monthly rainfall at four rainfall stations in West Java, Indonesia. The dependent variables used in this study are monthly rainfall data from 2011 to 2018 of four rainfall stations in West Java Province Indonesia and independent variables (global scale) are climate forecast system reanalysis v.2 (CFSRv2) data. This study resulted that the PCA model using K-means clustering as mediator resulted the smaller RMSEP values and the higher correlation values for monthly rainfall in each rainfall station used in this study. RMSEP value has range from 86 to 122 and correlation value has range from 0.82 to 0.94.
References
- Brun M, Sima C, Hua J, Lowey J, Carroll B, Suh E, and Dougherty ER, "Model-based evaluation of clustering validation measures", Pattern Recognition Society, Vol:40, 807–824, 2007.
- Estiningtyas W, Wigena A.H., " Teknik Statistical Downscaling dengan Regresi Komponen Utama dan Regresi Kuadrat Terkecil Parsial untuk Prediksi Curah Hujan pada Kondisi El Nino, La Nina, dan Normal ", Journal of Meteorology and Geofisika, Vol. 12, No. 1, 65–72, 2011.
- Irvan M, “Statistical downscaling modeling using linear regression with percentile L1 and percentile L2 to predict rainfall”, IPB, 2017.
- Johnson R.A., Winchern D.W., “Applied Mulivariate Statistical Analysis”, 6th ed, United State of America, Pearson Education, Inc., 2007.
- Marcus G.L., Wattimanela H.J., Lesnussa Y.A., “Principal component regression analysis for solving multicorrelation in multivariate regression analysis ( A study case: rainfall in Ambon 2010)”,Journal of Barekeng, Vol. 6, No.1, 31–40, 2012.
- Narang B, Verma P, Kochar P, “Application based, advantageous K-means clustering algorithm in data mining: a review”, International Journal of Latest Trends in Engineering and Technology, Vol. 7, No. 2, 121–126, 2016.
- Nurhasanah S, “Climate regionalization and agro-climatic zone: planting calendar in subang district”, IPB, 2017.
- Nurhayati, “Principal component regression, partial least square, and lasso methods on poverty data result of susenas 2012”, IPB, 2014.
- Sahriman S, “Statistical downscaling model with time lag of global circulation model to forecast rainfall”, IPB, 2014.
- Soleh A.M., “Gamma and generalized pareto distribution linear modeling with L1 regularization to predict montly rainfall in statistical downscaling”, IPB, 2015.
- Trzaska S, Schnarr E, “A review of downscaling methods for climate change projections”, United States Agency for International Development by Tetra Tech ARD, 1–42, 2014
- Wigena A.H., “Multi response partial least square for statistical downscaling”, Prociding Scientific Journal Club, Vol. 16, No. 2, 12–15, 2011.
- Zorita E, Storch H, “The analog method as a simple statistical downscaling technique:comparison with more complicated methods”, Journal of Climate,Vol. 12, 2474–2489. 1999.
Downloads
Published
Issue
Section
License
Copyright (c) IJSRSET
This work is licensed under a Creative Commons Attribution 4.0 International License.