A new spatial prediction model and its application to wind records

SummaryContour maps of any meteorological variable cannot give radius or area of influences around the measurement station by considering the records at surrounding sites. The main purpose of this paper is to propose a trigonometric point cumulative semivariogram (TPCSV) concept for deciding on a spatial dependence function and then its use for regional prediction. The TPCSV provides a unique opportunity for the establishment of a regional objective prediction method whereby the radius of influence helps to predict wind velocity at any site by using the weighted averages. The spatial correlations and weightings are obtained through the TPCSV provided that the distance between two sites is known. If the slope of TPCSV is greater than 80° after some distance, then beyond this distance the regional correlation is considered as negligible. The implementation of the proposed methodology is presented for 68 wind velocity measurement stations in Turkey. The proposed method yields the least prediction error compared with other objective methodologies. It is seen that areas of influence at Central Anatolia are generally bigger than coastal areas of Turkey.