Joint International Conference on Theory, Data Handling and Modelling in GeoSpatial Information Science, Hong Kong, PEOPLES R CHINA, 26 - 28 May 2010, vol.38, pp.92-97
In this study, mean annual precipitation and temperature values observed at 225 meteorological observations over Turkey are used to disclose spatial distribution of mean annual precipitation and temperature values. Data components were obtained from the Turkish State Meteorological Service for 34 years period (1970-2003). The basic objectives of the study are: to infer the nature of spatial variation of precipitation and temperature over Turkey based on meteorological observations and to model the pattern of variability of these data components by using secondary variables extracted from SRTM and river network. Modeling the spatial distribution of data sets is implemented with Co-kriging (COK), Ordinary Least Squares (OLS) and Geographically Weighted Regression (GWR) techniques with using secondary variables such as elevation, aspect, distance to river, roughness, drop (elevation differences between station and grid), sd-grid (standard deviation of 5*5 km grid), and plan-profile curvature. Correlations among the listed variables were analyzed and highly correlated ones were removed from the analysis. The study found a presence of high spatial non-stationary in the strength of relationships and regression parameters. The co-kriging interpolation method gave strong relationship for temperature (r(2)= 0.823) but comparatively weak relationship for precipitation (r(2)= 0.542). OLS method resulted with lower relationships for temperature (r(2)= 0.68) and for precipitation (r(2)= 0.3). The highest adjusted r(2) values were obtained with GWR method; 0.96 for temperature and 0.66 for precipitation.