The field approach was designed to interfere as little as possible with the cultivation
systems, and to minimize the possible errors introduced by selecting small
erosion plots and by extensively bounding plots on land with locally complex topographies.
Erosion plots have been criticized because they simplify topography and
ignore storage of runoff and sediment (Boardman, 1996). This is why, in this study,
the units of measurement were either whole terraces or substantial terrace segments.
Use was made of the agricultural ditches at the foot of the terrace risers for channel
runoff and eroded soil to collection drums. This also meant that artificial
bounding of the plots was kept to a minimum. Where artificial metal sheet boundaries
were emplaced, these were located wherever possible along natural drainage divides
on the terraces. Plots were thus arranged to accommodate natural drainage routes.
The riser ditches were left unlined and continued their normal function of trapping
some of the eroded soil from the terrace above. At the time of soil preparation for
the next crop, the ditch is re-excavated and the soil added to the adjacent terrace.
Thus the monitored terraces include the ‘natural’ storage components, both on the
terraces and in the ditches. Land management regimes on the terraces were largely
unaffected as the erosion ‘plots’ were, in general, sufficiently large to accommodate
ploughing.