Comparison of different interpolation techniques to map temperature in the southern region of Eritrea

Temperature and rainfall vary markedly throughout Eritrea, from hot desert in the east to a mild, subhumid climate in the highlands (Wolfe et al., 2008). Prediction and understanding of the spatial variation of climate data, particularly temperature, is important to many agricultural and economic sectors for planning and management activities (Moral, 2009). This is especially important in Eritrea where agriculture provides 12.4% of the gross domestic product, and 80% of the population are involved in farming and herding (Wolfe et al., 2008).
Several studies have demonstrated that various spatial interpolation techniques perform differently depending on the type of attribute, geometrical configuration of the samples, spatial resolution, world region, etc. (Martínez-Cob, 1996; Goovaerts, 2000; Haberlandt, 2007). Hence, selecting the best interpolation technique for each particular situation is a key factor. The major objective of this study is to assess the spatial variability of annual average temperature in the southern region of Eritrea by comparing different interpolation procedures. The temperature data were interpolated using a deterministic method (Inverse square distance) and three geostatistical methods (Ordinary, Universal and Simple kriging). The performance of the different techniques was compared
through error statistics computed using Jackknife cross-validation