Dynamic within-season irrigation scheduling for maize production in Northwest China: A method based on weather data fusion and yield prediction by DSSAT

Authors: CHEN, SHANG - Northwest A&f University; JIANG, TENGCONG - Northwest A&f University; MA, HAIJIAO - Northwest A&f University; HE, CHUAN - Northwest A&f University; XU, FANG - Northwest A&f University; Malone, Robert - Rob; FENG, HAO - Northwest A&f University; YU, QIANG - Northwest A&f University; SIDDIQUE, KADAMBOT - University Of Western Australia; DONG, QIN'GE - Northwest A&f University; HE, JIANQIANG - Northwest A&f University

Submitted to: Agricultural and Forest Meteorology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/31/2020
Publication Date: 2/9/2020
Citation: Chen, S., Jiang, T., Ma, H., He, C., Xu, F., Malone, R.W., Feng, H., Yu, Q., Siddique, K.H., Dong, Q., He, J. 2020. Dynamic within-season irrigation scheduling for maize production in Northwest China: A method based on weather data fusion and yield prediction by DSSAT. Agricultural and Forest Meteorology. https://doi.org/10.1016/j.agrformet.2020.107928.
DOI: https://doi.org/10.1016/j.agrformet.2020.107928

Interpretive Summary: With more efficient agricultural water management needed to ensure food security, tools are needed to improve irrigation management. We developed a new algorithm and incorporated into the CERES-Maize crop model to provide dynamic irrigation scheduling for maize (Zea mays L.) production in arid regions. Historical 50-year (1968-2017) weather data were obtained from four sites in Northwest China. Each year of these records was used to develop 50 complete climatic series as input to the CERES-Maize model to forecast the maize yield for a given growing season. The actual daily weather data for a growing season were combined with historic weather data as the simulated growing season progressed. Forecasted maize yields were adjusted for each advancing day of the growing season as the actual weather data replaced the historic data. Using this combination of the actual annual and historic weather data for a site, an irrigation with a user defined amount was triggered if the trend of forecasted maize yield decreased continuously for a user defined number of days. Compared with irrigation schedules based on actual field management and the automatic irrigation option in the CERES model, irrigation scheduled by the new algorithm generally resulted in higher water use efficiency (WUE=corn yield divided by the sum of irrigation and precipitation). The new algorithm will help improve maize irrigation scheduling in arid and semiarid areas. This research provides a new tool to optimize irrigation management and may help agricultural scientists and the agriculture industry design more efficient systems.

Technical Abstract: With more efficient agricultural water management needed to ensure food security, tools are needed to improve irrigation management. This study used the CERES-Maize model in DSSAT with a new algorithm to provide dynamic irrigation scheduling for maize (Zea mays L.) production in arid regions. The study included field experiments conducted in four arid and semiarid sites in Northwest China: Changwu (2010 and 2011, rainfed), Yangling (2014 and 2015, irrigated), Jingyang (2015, irrigated), and Shiyanghe (2015, irrigated). Historical 50-year (1968-2017) weather data were available for the four sites. Daily weather data used in model simulation were divided into two parts (actual and historic) during a given maize growing season. Actual weather data were observed from local weather stations, while historic data supplemented the continuous weather series. Therefore, 50 complete climatic series were created and used to run the CERES-Maize model to forecast the maize yield for a given growing season. As the growing season advanced, historical weather data were gradually replaced by the actual weather data. Forecasted maize grain yield adjusted each day were obtained, which was calculated as the median of the 50 predictions. Dynamics of daily forecasted yield were then used to schedule irrigation based on a new automatic irrigation algorithm. When the trend of forecasted maize yield decreased continuously for a user defined number of days, an irrigation event with a user defined depth was triggered. The new algorithm considers the actual weather conditions before starting an irrigation event in a given growing season. The results showed that the uncertainty of forecasted maize yield was large before tasseling but rapidly converged to the actual yield about one month before harvest with an absolute relative error (ARE) less than 8%. Compared with the irrigation schedules based on actual field management and the automatic irrigation option in the DSSAT model, irrigation scheduled by the new algorithm had higher water use efficiency (WUE) for the experimental sites and years investigated, except for Yangling in 2014 due to the heavy rainfall during the grain filling stage. The new irrigation scheduling algorithm will help improve maize irrigation scheduling in arid and semiarid areas.