Castilla-La Mancha in Central Spain is a semi-arid area of extremely high i
nterannual and seasonal rainfall variability. Average annual rainfall for t
he catchment of the Upper Guadiana using data from 60 rain gauges for Octob
er 1956-September 1991 varied from a minimum of 326 mm in October 1982-Octo
ber 1983 to a maximum of 642 mm in October 1968-September 1969. The mean an
nual rainfall for the period was 495 mm with a coefficient of variation for
annual rainfall of 26.4%. In addition to this the spatial variability of r
ainfall is particularly high. For example total annual rainfall varied from
200 to 1200 mm for the hydrological year October 1968-September 1969 over
a distance of only 50 km. The mean annual rainfall for all 60 stations for
the period 1956-1991 was 460 mm with a coefficient for spatial variation of
15%. Dryland farming which relies on these uncertain precipitation inputs
is a high risk activity [Tarjuelo, J.M., de Juan, J.A., Valiente, M., Garci
a, P., 1996. Agric. Water Manage. 31, 145-163] and over recent decades inte
nsive irrigation has increased dramatically in order that precipitation inp
uts can be supplemented as required thereby allowing more stable agricultur
al productivity. The paper uses a coupled hydrology and vegetation growth m
odel - PATTERN [Mulligan, M., 1996. Modelling hydrology and vegetation chan
ge in a degraded semi-arid environment. PhD. Thesis, University of London]
to explore the relationship between irrigation and productivity for differe
nt soils typical of the Upper Guadiana catchment. Analysis of the model res
ults shows that irrigation efficiency is highly sensitive to both soil text
ure and irrigation volume. Optimally efficient irrigation in terms of water
losses occurs at the lowest volumes of applications. Fine grained soils ar
e better suited to irrigation than coarse grained soils as losses to rechar
ge are minimised. Coarse grained soils have large recharge losses and irrig
ation is also more sensitive to small changes in texture for coarser soils.
Irrigation losses through recharge are also very sensitive to the interann
ual and spatial variability of rainfall. (C) 1999 Elsevier Science B.V. All
rights reserved.