Cropland has increasingly occupied large areas in Southern Amazonia since the 1990s, yet few direct field-based evapotranspiration (ET) measurements are available. We used the eddy covariance method to measure ET on a farm with multiple cropping cycles per year over September 2015 through February 2017. This period covered two soybean crops and associated rotations in each of two adjacent fields: a rain-fed field planted to soybean, maize, brachiara, and soybean, and an irrigated field planted to soybean, rice, bean, and soybean. Total ET during these crop cycles was 1265 ± 294 mm and 1414 ± 181 mm respectively for rain-fed and irrigated fields compared to 3099 mm of precipitation. Soybean ET was similar in both rain-fed and irrigated fields and ranged from 332 ± 82 mm (October 2015–February 2016) to 423 ± 99 mm (October 2016–February 2017), while rain-fed maize ET was 313 ± 68 mm (February–July 2016). The main difference between the fields was observed in the dry season (June–August 2016) when irrigation was applied for bean production. A crop model validated with the field data, showed that (1) crop transpiration represented 48–52% of ET for both soybean and maize, and (2) that irrigation for soybean planted at the end of the dry season resulted in marginal improvements
to crop water productivity. Results provide insight into the use of irrigation to increase cropping frequency rather than yield with regional land and water implications at the Brazilian agricultural frontier.
