Water is vital for the cultivation of fresh produce, which in turn is integral to global food security. The demand for water resources is projected to increase rapidly due to global population growth and the impacts of climate change, leading to heightened competition between agricultural requirements and other water uses, both human and environmental. Effective water management practices in agriculture, coupled with precise monitoring of water consumption, will be essential for achieving the Sustainable Development Goals (SDG) related to efficient water use. Despite this, reports highlight the issue of overuse of water in irrigated areas, particularly in semi-arid and arid regions. This study reveals only a weak relationship between gross primary productivity (GPP) and transpiration (T) in arid, irrigated farmlands of the Heihe River Basin in China, using daily T and GPP data from 2000 to 2020. Transpiration data were calculated using a Two-Source Energy Balance (TSEB-SM) model driven by thermal infrared and microwave soil moisture satellite data. The model's estimates of daily evapotranspiration (ET) showed strong agreement with observations from eddy covariance flux measurements in grassland and cropland areas, with root mean square error (RMSE) below 1.0 mm/day, and good agreement in Tamarix shrub and desert steppe environments where the vegetation may access deeper water (RMSE < 2.0 mm/day). The increased T but steady GPP, especially in the middle to late growing season of the irrigated crop areas, indicating low plant productivity per unit of water transpired. These findings align with the concept of irrigated crop decoupling, whereby under intense heat stress and an abundant water supply, plants can maintain transpiration rates significantly higher than those associated with photosynthesis. This study highlights the need to improve irrigation practices and increase real-time monitoring of water usage.