3.1 Descriptive statistics of the soil properties
Descriptive statistics of the properties of SAS are presented in Table 1. For the full dataset, the coefficient of variation varied from 16.75-118.30% with the highest observed for Fe (118.30%) and lowest for soil pH (16.75%). For the calibration and validation dataset coefficient of variation varied from 18.46 to 122.91% and 15.44 to 110.85%, respectively. The highest coefficient of variation (%) was recorded for Fe (122.91%) and lowest for soil pH (18.46%) for the calibration dataset whereas in the case of validation dataset it was highest for Mn (110.85%) and lowest for soil pH (15.44%). The soil pH and EC of the full dataset varied from 3.47-8.04 (mean 5.42) and 0.10-24.11 (mean 6.91) dS m-1, respectively. It indicates co-existence of the salinity and acidity which is the unique property of these soil . The SOC ranged from 0.02% to 3.20% with a CV of 56%. A physical property, BD, varied between 0.89-2.17 Mg m-3 with an average of 1.44 Mg m-3and CV of 18.56%. The variability with respect to the macronutrient elements N and P was 42.76% and 93.00%. The cations, K, Na, Ca and Mg had coefficient of variation ranging from 65.97% to 83.44%. The variability of the micronutrients (Fe, Mn, Cu, Zn and B) varied from 63.93% to 118.30%. The results of the descriptive statistics revealed presence of adequate variability in most of the parameters measured. All the soil properties were positively skewed for full, calibration (except pH) and validation dataset. To ensure the random selection of calibration and validation dataset, test of equal means, variance, distribution and coefficient of variation was carried out and the results are presented in Table 2. The results of all the test were non-significant (p ≥ 0.06) except F-test (equal variance) for S (p=0.05) and Zn (p=0.00004). This indicates that the calibration and validation dataset effectively represent the variability existed in the full dataset.