Abstract

Study evaluated predictive accuracy of USDA Soil Taxonomy Classifications of Soils of University of Nigeria, Nsukka. Data from 0 – 20cm and 30 – 60cm depths of 9 profiles, each representing a map unit, were used to determine coefficients of variation (CV) of soil properties over whole area sampled (control), within Great group class and series. There was progressive reduction in CVs from high to low categories, with the properties doing so irregularly. Average CVs for the various levels were 59.58% (over whole area), 56.97% (Great group), 50.77% (series) at topsoil, while at subsoil they were 38.15% (whole area), 31.53% (Great group), 25.19% (series). At topsoil, predictions of K & OC improved by 36.16% on the average at Great group, while it did for Clay, K, OC by 43.71% at series. At subsoil Silt, Mg, CEC, OC, TN improved by 34.17% at Great group on the average, while Clay, Silt, Mg, CEC, OC, TN, av.P did by 47.49% at series. Predicted properties, which were found to correlate with others, influence soil productivity. Sand and pH were virtually unaffected by classification. Study highlights a technique for evaluating predictive accuracy of soil classification using small sample size as well as the essence of detailed characterization of the soils.

This work is licensed under a Creative Commons Attribution 4.0 License.