The amount and distribution of both rainfall and irrigation during these two rice-growing seasons are shown in Fig. 1. During the 2009 season, the total rainfall of 97.4 cm was higher than the long-term average and occurred mainly during the first three months of the crop growth. During the 2008 season,the total rainfall of only about 54.3 cm was lower than the long-term average. During the 2008 and 2009 seasons, 14 and 9 irrigations were applied, resulting in total water depths of 72 and 46 cm, respectively. To prevent long flooding and anaerobic conditions and to control rice insects and weeds, rice fields need to be naturally dried for about 5–10 days during the season, depending on the rice and soil conditions. The soil drying stage in the 2008 season was between August 7th and 12th, while in the 2009 season it was between August 14th and 20th.The daily climate data, including temperature, wind speed, humidity, and daylight hours, were obtained from an adjacent agrometeorological station in the Dangyang region. The amount of irrigation water was measured using flow meters at inlets, and the flooding water depth was recorded every two days at several points in the field. The total amounts of surface runoff at outlets were measured by flow meters to be 5.2 and 22.5 cm during the 2008 and 2009 seasons, respectively . Before rice seeding, trim tray for weed self-made flux lysimeters for measuring vertical water fluxes were installed at a depth of 60 cm below the soil surface.
Each device consisted of an open-cylinder that was connected to a glass bottle. The top surface of an open-cylinder was welded using stainless steel ribbons to avoid soil falling down into the cylinder. Leached water that was collected in the cylinder freely flowed into the outside glass bottle through a flexible tube. The water volume in the bottle was measured once every two weeks. The groundwater table was observed once a week at an observation point near the experimental field. The recorded data show that the groundwater table slightly fluctuated, but remained around a depth of 120 cm below the soil surface. Piezoelectric tube tensiometers were installed at five observation points in the field to measure pressure heads one month before rice seeding. At each observation point, five tensiometers were installed at depths of 20, 40, 60, 80, and 100 cm below the soil surface . The pressure head values were recorded every two days. The averaged data from each of the depths at the five observation points were used for model simulation.As expected, larger differences in flooding water depths between the 2008 and 2009 seasons were mainly found during the first half of the season. The frequent intensive rainfall in the early 2009 season caused the water depth to reach the maximum depth of 10 cm within several days. On the other hand, during the 2008 season, the water depth was always below the maximum controlled depth, because the water input was mainly determined by irrigation. According to simulated values, flooding water depths between 0 and 5 cm and between 5 and 10 cm during the 2008 season were 90 and 7 days, respectively, while during the 2009 season the corresponding values were 72 and 26 days, respectively.
Overall, rainfall was sufficiently utilized by the rice crop during the 2009 season. Although DSR could effectively use early rainfall, a large amount of water was still lost from fields as surface runoff. During the 2009 season, most rainfall occurred during the early part of the season, resulting in five surface runoff events. The biggest surface runoff event was observed on 20 DAS, just after continuous intensive precipitation . During the 2008 season, only two surface runoff events were observed. To prevent long-flooded and anaerobic conditions in 2009, the field was artificially drained on 65 DAS. Simulated and observed total surface runoff during the 2009 season was 24.4 and 22.5 cm, respectively, which was approximately four times more than during the 2008 season, when simulated and observed runoff was 5.8 and 5.2 cm, respectively. The difference between simulated and observed surface runoff can be attributed to several small surface runoff events simulated by Hydrus-1D that were not observed in the field. This could also be due to preferential flow caused by macropores and cracks, which was not considered in the simulation, and which would increase water infiltration and reduce surface drainage to a certain extent .Phogat et al. showed that the Hydrus-1D model can describe root water uptake of transplanted rice very well. In their simulations of water and salt movement in rice fields, they showed that simulated root water uptake closely corresponded with crop performance in their experiment.
Fig. 5 shows daily root water uptake by DSR during the 2008 and 2009 seasons, as simulated using Hydrus-1D. Root water uptake was initially almost zero during the seeds’ germination stage. Root water uptake then gradually increased after about 10 DAS, reflecting the crop growth, and reached its maximum values between 90 and 120 DAS. The maximum daily root water uptake rate was approximately 0.6 cm/day around 100 DAS. During later stages of crop growth, daily root water uptake substantially declined. The cumulative root water uptake only slowly increased during the initial 30 days,then quickly increased and ultimately reached 51.4 and 48.2 cm during the entire growing seasons of 2008 and 2009, respectively. Actual evaporation rates simulated using Hydrus-1D and estimated potential evaporation rates were 17.4 and 15.8 cm during the 2008 and 2009 seasons. Most evaporation occurred during the early part of the season when the crop cover was still relatively small. Similar results have been reported in the literature. For example, Choudhury et al. estimated the seasonal DSR evapotranspiration loss in dry-seeded rice, on a flat land with row spacing of 20-cm in New Delhi, India, to be 55.6–56.0 cm during 2001 and 2002. Sudhir-Yadav et al. reported that the estimated ET values in intermittently irrigated DSR fields were 49.7–70.7 cm during the 2008 and 2009 seasons in Punjab, India.Dynamic changes in water contents of the soil profile of rice fields depend on many natural and artificial factors. Differences in soil hydraulic properties between different soil layers cause characteristic water content profiles. While the plough sole layer was often saturated, unsaturated conditions occurred in the subsoil below the plough sole layer . This is because the subsoil is often more permeable than the plough sole layer and its hydraulic conductivity is 1.5–2.3 times higher than that of the plough sole layer. Actually, cannabis grower supplies the plough sole layer often plays an obstructing or buffering role on vertical water flow . The low hydraulic conductivity of the plough sole controls the vertical water movement during both drying and wetting in rice fields. Simulation results indicated that the soil water storage in the upper 60 cm of the soil profile throughout the 2008 season decreased 1.2 cm, while during the 2009 season, it decreased by about 2.3 cm. The change of soil water storage predominantly depended on the initial water content and water contents at harvest.Downward leaching represented the largest loss of water from the rice field. Leaching predominantly depended on the soil hydraulic conductivity of individual soil layers and the overall pressure head gradient. As shown in Fig. 6, leaching closely corresponded with precipitation and irrigation events. Due to early intensive rainfalls, relatively continuous high leaching rates were mainly observed during 30–60 DAS. During the later growth stages when the water input was predominately provided by irrigation, leaching was relatively regular and reflected particular irrigation events.
The maximum leaching rate at a depth of 60 cm occurred on 42 DAS during the 2008 rice season, immediately after an intensive rainfall on 41 DAS. During the 2009 season, the maximum leaching rate at a depth of 60 cm occurred on 48 DAS after a continuous intensive rainfall . Chen and Liu reported similar results when an increase in the flooding water depth from 6 to 16 cm produced a 1.5 fold increase in the infiltration rate in a paddy field. As shown in Fig. 6, the differences in leaching rates between two seasons were mainly caused mainly by different water managements. While the water input in the 2008 season was predominately accomplished by irrigation, in the 2009 season, it was mainly by rainfall . Average simulated flux rates at a depth of 60 cm before 66 DAS in the 2008 season and before 72DAS in the 2009 season were 0.36 and 0.41 cm/day, respectively, while those during the entire 2008 and 2009 seasons were 0.37 and 0.34 cm/day, respectively. More frequent irrigations in the late 2008 season resulted in more water leaching compared to the late 2009 season. Similarly, frequent and intensive rainfall during the early half season of 2009 resulted in much more water leaching. The leaching rate in paddy field soils in the area is about 0.3–0.5 cm/day, which is considered to be an optimal leaching rate for good rice growth in this region . Soils with a higher leaching rate are considered “water-wasting” soils, while soils with a lower leaching rate often have Eh too low for rice growth . Lian et al. also reported that the water leaching rates measured by lysimeters in adjacent paddy fields were between 0.49–0.56 cm/day.The measured and simulated components of the water balance of the upper 60 cm of the soil profile are presented in Table 4. The total measured water inputs during the 2008 and 2009 seasons were 126.3 and 143.4 cm, respectively. Simulated evapotranspiration during the 2008 and 2009 seasons accounted for 54.6% and 44.6% of corresponding volumes of water input, respectively, while leaching at a depth of 60 cm accounted for approximately 42.7% and 34.9%, respectively. The total water balance errors were −3.5 and 2.7 cm during the 2008 and 2009 seasons, respectively, which account for about 2.8% and 1.9% of the total water input. This indicates that the Hydrus-1D model is a good tool for simulating the water balance components in the DSR fields. The average grain yields in the 2008 and 2009 seasons were 8980 kg/ha and 8530 kg/ha, respectively. Water productivity , also known as water use efficiency, is a measure that characterizes the crop production per unit of water used. Three measures of water productivity were computed: irrigation water productivity —the ratio of grain yield to the amount of irrigation water; input water productivity —the ratio of grain yield to the amount of irrigation water plus rainfall; and evapotranspiration water productivity —the ratio of grain yield to crop ET . The ET value for these calculations was taken as the sum of cumulated root water uptake and evaporation during the entire season, as simulated by the Hydrus-1D model. Measured seasonal values of precipitation and irrigation were used in the above calculations as the total water input . Using this definition, the WPs in the 2008 and 2009 seasons are listed in Table 4. The differences between WPI and WPIR values in the 2008 and 2009 seasons can be attributed mostly to different amounts of irrigation water. Sudhir-Yadav et al. estimated the WPI for intermittently and daily irrigated DSR fields in India to be 0.7–1.6 kg/m3 and 0.2–0.3 kg/m3, respectively. The WPIRs in our experiments are close to a value reported by Ginigaddara and Ranamukhaarachchi in Thailand for DSR with one week irrigation, followed by three weeks without irrigation. Ye et al. reported that the WPIRs of intermittently irrigated TPR in the Taihu Lake Basin were between 0.48 and 1.06 kg/m3 for different fertilizer managements, with average values of 0.88 and 0.77 kg/m3 during the 2010 and 2011 seasons, respectively. The WPETs in our experiments fall in the range of globally measured values for rice , and are higher than the average value but slightly lower than the median value of 1.52 kg/m3 reported by Roost et al. for TPR of the Zhanghe Irrigation District in Central China. The evaluated water productivities indicated that DSR in the 2008 season had a better water use efficiency compared to DSR in the 2009 season, and that the DSR in the 2009 season sufficiently used rainfall. Similar water productivities in the two seasons, with respect to the simulated evapotranspiration, signified that the crop yields were closely associated with the root water uptake.Cultivation is a practical means to incorporate organic matter and fertilizers into a soil at various depths.