Evolution of the soil surface roughness using geostatistical analysis

The objective of this work was to investigate the decay of initial surface roughness induced by simulated rainfall under different soil residue cover and to compare classical statistical indices with geostatistical parameters. A conventionally tilled loamy soil with low structure stability, thus prone to crusting was placed at 1 m2 microplots. Each microplot received three successive rainfall events which bring about cumulative 25 mm, 50 mm and 75 mm at 65 mm h -1 intensity. Five treatments without replication were tested with different corn straw quantities (0, 1, 2, 3 and 4 Mg ha -1 ). Soil surface microrelief was measured at the initial stage and after each simulated rainfall event. Five treatments and four surface stages were monitored, resulting in 20 data sets. Point elevation data were taken at 0.03 m intervals using a pinmeter. Digital elevation models were generated and analysed using semivariograms. All data sets showed spatial dependence and spherical models were fitted to experimental semivariograms. A very significant relationship was found between the random roughness index, RR, and the sill of the semivariogram (C0 +C1 ). All the treatments showed a clear trend to sill value reduction with increasing precipitation. However, roughness decay was lower in treatments with higher straw cover (3 and 4 Mg ha -1 ). Therefore, residue cover limited soil surface roughness decline. The control treatment, without straw, showed the lowest nugget effect (C0 ), which means the lowest spatial discontinuity of all treatments in this study. The range of spatial dependence (a) also showed a trend to decrease with increased cumulative rain, which was most apparent in treatments without or with relatively low straw cover (0, 1 and 2 Mg ha -1 ). The suitability of using sill variance and range for describing patterns of soil surface microrelief decline is discussed.