Manipulating water depths and timing is a key management practice in rice cropping systems, but rice models have not simulated water-depth effects on plant growth. A new plant growth model, DSRICE1, simulates most cultural and weed management practices except fertility. Water-depth effects on plant light capture are mechanistically simulated as reflection and attenuation of light by water. Light attenuation by water in the model depended on water depth and the light extinction coefficient of water, k(H2O) (m(-1)). DSRICE1 was validation tested for prediction of monoculture growth, and specifically for early-season water-depth effects. Analyses revealed that attenuation by water limited irradiance and reduced seedling growth, and indicated which plant traits contributed to growth during submergence. Reflection did not affect rice growth and may be ignored. In empirical validation tests, DSRICE1 predictions were compared with data from 14 independent data sets. For growth up to 30 days after seeding (DAS) (11 experiments), simulations without water-depth effects failed validation tests by of overpredicting rice shoot dry mass (DM). In contrast, DSRICE1 simulations with water-depth effects included were accurate with k(H2O) from 1 to 4, k(H2O) = 3 was best overall, and predictions improved with experiment-specific k(H2O) values. DSRICE1 also accurately predicted differences in shoot DM per plant in a 1985 water-depth study, while again simulations without water-depth effects did not. Thus, water-depth effects were required to simulate early rice growth accurately. In whole-season empirical validation tests (14 experiments), DSRICE1 simulated shoot, stem, leaf, live leaf, and live stem DM accurately, especially considering the range of data used and the fact that no calibration was needed. DSRICE1 was also corroborated by the fact that it accounts for many physical factors and plant traits that affect submergence tolerance. The model or the techniques it uses may be useful in analyses of cultivar tolerance to submergence. In some respects, DSRICE1 was subjected to more rigorous validation testing than previous rice models, and potentially explains more interactions between rice and weed growth and management. This approach may broaden and improve simulation analyses of integrated weed management in direct-seeded rice systems. (C) 1999 Elsevier Science B.V. All rights reserved.
