Improving agricultural water management in the semi-arid region of Brazil: experimental and modelling study

Abstract

Agricultural practice in the semi-arid region of Brazil is highly dependant on irrigation. As access to water is limited in the region, there is a need to guarantee its efficient use, especially in small-scale farming schemes. Models adequately calibrated for semi-arid conditions and for typical crops are useful tools for analysis of on-farm strategies to improve water use efficiency. A physically based agrohydrological model, SALTMED, is calibrated and validated for carrots (Daucus carota L., Brasilia variety) and cabbage (Brassica oleracea var. capitata), two of the main crops in small-scale irrigated agriculture in the northeast of Brazil. SALTMED is also calibrated for castor beans (Ricinus communis L.) under rainfed conditions. The results demonstrated the suitability of the SALTMED model for simulating soil water dynamics and crop yield in the area. Predicted time series of soil water content and matric potential of root zone showed good agreement with observed values, as shown graphically and statistically. Using the calibrated and validated model, management scenarios were analysed in terms of applied irrigation water, irrigation frequency, soil and crop types. Impact of climate change on the irrigation water requirements was also briefly analysed for possible changes in annual temperature using two different emission scenarios. Analysis of possible impact of climate change on temperature related to two emission scenarios on the area showed an increase around 11% of the crop water requirement for carrots and cabbage, for the low emission scenario, and around 17% for the high emission scenario. The analysis of management scenarios indicated possible over-irrigation in the area. The simulation showed that the deficit irrigation was a useful water-saving strategy for the region. The simulations also indicated that irrigation frequency affected crop water use and differed according to the soil type.