www.seipub.org/des Development in Earth Science (DES) Volume 3, 2015 doi: 10.14355/des.2015.03.002
Modelling of Surface Air Temperature Element in Malaysia F. Yunus*1, N. K. Chang2, F. J. Fakaruddin3, M. K. Mat Adam4, J. Jaafar5, Z. Mahmud6 1‐4
Malaysian Meteorological Department, Jalan Sultan, Petaling Jaya, Selangor, Malaysia
5‐6
Universiti Teknologi MARA, Shah Alam, Selangor, Malaysia
fariza@met.gov.my; 2nursalleh@met.gov.my; 3fadila@met.gov.my; 4matkama@met.gov.my; 5jasmee@salam.uitm.edu.my; 6zamalia@tmsk.uitm.edu.my *1
Abstract In spatial interpolation approaches, attribute data assumes continuous over space and spatially dependence. The surface air temperature is a collection of data, which are observed at discrete locations. Normally, spatial interpolation analysis applies to produce a continuous surface of this discrete data. There are a number of spatial interpolation techniques available to create continuous distribution of surface air temperature. To generate map of surface air temperature, this study examined the interpolation technique of inverse distance weighting (IDW). In applying the IDW technique, two different types of main data were assessed, i.e. mean monthly temperature data of T and estimation error of T – T’, where T was observed mean monthly surface air temperature and T’ was estimated mean monthly mean surface air temperature from a multiple regression. The multiple regression model was developed based on eight independent variables of elevation, location of latitude and longitude, distances of a station to nearest coastline, four types of land use, which included water bodies, forest, agriculture and build up areas. Cross validation analysis was conducted to calculate five different measured of errors. Inverse distance weighting (IDW) spatial interpolation of T ‐ Tʹ main data was produced acceptable errors and reliable map for mean monthly mean surface air temperature element in Peninsular Malaysia. Keywords Temperature; Malaysia; Spatial Analysis; Interpolation Technique
Introduction Spatial interpolation technique plays a significant role in many environmental studies. The technique was used to generate continuous surface based on point data. There are a number of spatial interpolation technique established, such as inverse distance weighting (IDW), kriging, spline and trend surface regression (Myers, 1994). Although various types of kriging are among the best known spatial interpolation technique in earth science (Brown & Comrie, 2002; Myers, 1994), IDW is frequently used in spatially located of climate data (DeGaetano & Belcher, 2007; Dodson & Marks, 1997; Kurtzman & Kadmon, 1999; Price, McKenney, Nalder, Hutchinson, & Kesteven, 2002; K. Stahl, R.D. Moore, J.A. Floyer, M.G. Asplin, & I.G. McKendry, 2006a). Furthermore, previous studies (Jarvis & Stuart, 2001a; K. Stahl, R. D. Moore, J. A. Floyer, M. G. Asplin, & I. G. McKendry, 2006b; Valley, Drake, & Anderson, 2005; Yunus, 2005) have shown that the performance of kriging and IDW are almost the same. In relatively homogeneous flat surface, direct interpolation analysis of surface air temperature element produced acceptable errors (Serbin & Kucharik, 2009). However, in heterogeneous terrain surface, relationship of surface air temperature with elevation and others environmental elements preclude the direct interpolation of point‐based surface air temperature observation (Civerolo, Sistla, Rao, & Nowak, 2000; DeGaetano & Belcher, 2007; Dodson & Marks, 1997; Stahl et al., 2006a; Ustrnul & Czekierda, 2005). Elevation consistently influenced the distribution of surface air temperature element, where it has directed the spatial interpolation of surface air temperature in many studies (Johnsons, Daly, & Taylor, 2000; Kurtzman & Kadmon, 1999; Price et al., 2002). Other than main influence by elevation, location and distances to coastline and build up areas, are also significant variables in determining surface air temperature values (Jarvis & Stuart, 2001b). Stahl et al. (2006) were included meteorological stations location as independent variable in modeling surface air temperature over British Columbia, Canada. By considering elevation and urbanization, errors in interpolation of surface air temperature can be reduced up to 30% (Choi, 2003). Previous studies (Civerolo et al., 2000; Gallo, Owen, & Easterling, 1999; Shudo, Sugiyama, Yokoo, &
10