Modelling Mean Surface Temperature Of Nigeria Using Geostatistical Approach

Understanding spatial variability of mean surface temperature (MST) of Nigeria isnecessary for ecological restoration and national planning toward effects of unstableclimate conditions. This study developed models for MST derived from two geostatisticalprocedures and Multiple Linear Regression (MLR) using measurements ofmonthly MST in Nigeria.The two geostatistical approaches were Ordinary Kriging(OK) and Regression Kriging (RK).The Ordinary kriging (OK) was developed intwo dimensions with isotropy and in three dimensional plane with anisotropy andregression kriging (RK) that employed both correlation with independent variablesand spatial autocorrelation simultaneously. Four statistics were considered to evaluatethe performance of the approaches used. For the fitted MLR model, some of thepredictors were significant at p-value of 0.05 with R2 equal to 88 percent. The oneleave-out cross-validation indicated that RK produced minimum errors compared toOK model . The OK with zonal anisotropic shown that the spatial continuity inthe directions of North and North East were stronger than the directions of Eastand South East. The kriging weights for OK and RK were similar as shown inthe maps.The RK model outperformed the OK and the MLR and was thereforerecommended for Long term prediction of mean Surface Temperature of Nigeria.

Get Full Work

Report copyright infringement or plagiarism