INFLUENCE OF RAINFALL ENERGY FLUX-DENSITY AND ANTECEDENT SOIL-MOISTURE CONTENT ON SPLASH TRANSPORT AND AGGREGATE ENRICHMENT RATIOS FOR A HAWAIIAN OXISOL

Erosion process information from aggregated tropical soils is under-re presented in the literature. Thus, an Oxisol was used in laboratory ra infall simulation experiments to examine the importance of antecedent moisture content (AMC) and rainfall energy flux density (EFD) on splas h. Ten splash experiments were conducted at EFD-values ranging from 0. 064 to 0.53 W m(-2) for 1 h dry and 1 h wet runs. Results indicated th at AMC had no significant influence on total splash flux or splash flu x per aggregate size fraction. However, splash increased linearly with EFD. The Oxisol exhibited a very high resistance to splash detachment (whole-soil basis), and the 2000-4000 mu m aggregate size fraction wa s the most resistant to splash detachment. There was no significant di fference for total splash flux between dry AMC-low EFD and wet AMC-low EFD events; or between dry AMC-high EFD and wet AMC-high EFD events. However, time trends differed significantly, and these reflected the i nfluence of variable water content on near-surface soil strength, part icle shielding and availability of material. Splash enrichment ratios (ER) for all AMC and EFD data indicated that the 425-2000 mu m aggrega tes were preferentially transported (i.e., ER significantly > 1.0). Ag gregates of 250-500 mu m were splashed in similar proportion to their content in the original soil, and aggregates < 250 mu m and > 2000 mu m were significantly depleted in splashed material (i.e., ER < 1.0). T he reasons for depletion include either increased cohesion with fine g rained aggregates or the inability of simulated rainfall to transport the most massive aggregates. Our data show that splash is a time-depen dent process influenced primarily by rainfall energy flux and to a les ser extent by interactions with antecedent soil moisture status. Data of this type are essential for incorporation into state-of-the-art ero sion prediction models if they are to be truly universal.