Spatial variation of infiltration rate and related soil properties in a central Gezira clay soil. I. conventional statistics

Two experimental sites, Kubur and Talha, in the central Gezira clay plain, were selected for investigating the spatial variation of infiltration rate (IR), clay content, EC, SAR, CaCO3 and PH (paste). In each site, IR was measured and soil samples (0-50 cm) were collected for determining other soil properties, at every 5 m- interval along S-N and W-E transects, each 500 m lomg and intersecting in the middle. The soils of the two sites belong to Wad EL Hadad series (vey fine, montmorillonitic, isohyperthermic Typic Chromustert). The results showed that clay content, PH (paste) and CaCO3 were nearly similar in the two directions, whereas EC, SAR and IR were not. Testing of normality indicated that the distribution of EC and SAR were lognormal, whereas those of the other properties were normal. There was a highly significant correlation (square R=0.926) between the coefficient of determination of the fractile diagram and the coefficient of skewness. In general, the average data for the two transects in each site, showed that the soil has high clay content (51.3%), low IR (1.9 cm/hr), very low EC (0.46 ds/m) and SAR (2.3) and slightly alkaline PH (paste) (8.1). Average data showed that the variation of PH (paste) was very low (CV=2-3), that of clay content was low (CV=6-10), that of IR was intermediate (CV=31-36), those of CaCO3 (CV=27-43) and EC (CV=27-51) were intermediate to high, whereas that of SAR (CV= 44-67) was high. The number of samples or measurements required for estimating PH (paste), clay content, IR CaCO3, EC and SAR within 10% of the correct value at the 5% probability level are 1, 4, 50, 71, 100 and 172, respectively. Although the soil belongs to one series, the relatively high variations of the salt parameters prompted the use of a large number of samples for their reliable determination and for that of IR. In view of the lack of significance of the correlation coefficients in 4 out of 6 cases, and the extremely low accountability of SAR (square R 5%) for IR variation in the two cases, regression analysis was ruled out as a tool for predicting IR