Irrigation experiments with wetland rice (Oryza sativa L.) paddies for 2 years under shallow water table conditions showed that intermittent submergence 3 days after water vanished from the surface of the soil produced grain yields similar to continuous submergence. Whereas irrigation was necessary every 3 days in clay loam (S1) and silty clay loam (S2), and every 2 days in loam (S3) under continuous submergence of 5 ± 2.5 cm, it was necessary only every 7 days in S1 and S2, and every 6 days in S3, under intermittent submergence of 7.5 cm 3 days after water vanished from the soil surface. The intermittent submergence led to a 34–43% saving of irrigation water. Percolation losses were reduced by 36, 31 and 25% in S1, S2 and S3, respectively, as compared to those under continuous submergence. Average percolation rate was 6.1–6.4 mm day−1 in S1 and S2, and 10.1 mm day−1 in S3. Yields were higher in the better drained S3 than in the poorly drained S1 and S2.
Average ET ranged from 7.4 to 9.2 mm day−1 under continuous submergence and from 6.1 to 7.1 mm day−1 under intermittent submergence. The intermittent submergence produced 7.2–14.7 kg more grain per cm of ET than continuous submergence. Regression of seasonal water requirement (WR) on percolation (Pc) and ET, and grain yield on WR and ET was highly significant. Yield was better correlated with ET (r = 0.92) than with WR (r = 0.86).