The variability of soil wetness was studied in the rainy season based
on 12 monitoring series in a layout including 18 tubes and 5 depths. T
he distribution of point volumetric wetness was found to be non-normal
, including a few relatively wet or dry sites. By introducing a simple
microrelief index based on the identification of peaks and depression
s in the furrows at a horizontal resolution of 50 cm, it was found tha
t 40% of the variation in site wetness could be explained by variation
s in microrelief, influencing the surface storage and subsequent absor
ption of water. Then, defining a representative subarea as an area whi
ch includes the full range of variation in microrelief index (at a pro
bability of 95%), a minimum plot-length of 12.5 m in the furrow direct
ion was estimated for a plot to be representative. The change in profi
le (1 m) wetness was found to be normal distributed, with the standard
deviation (3.6 mm on average) linearly increasing with the length of
the monitoring time interval (T) while the error of estimated areal av
erage change in profile wetness, relative to the period evapotranspira
tion, increased linearly with the frequency of monitoring (T-1). As a
result, the number of access tubes required to estimate change in prof
ile wetness at a particular error level was inversely related with the
length of the monitoring time interval. Thus, in terms of precision a
nd network practicality, a minimum monitoring time interval of about 3
days was found for estimating changes in profile wetnes. Then, 20 tub
es were required to ensure a precision of 1.5 mm, equivalent to a maxi
mum error of 10% relative to the period evapotranspiration.