Rain-fed lowlands are major agricultural ecosystems used for rice production in Northeast Thailand. Developing a tool to assess the effects of variable water conditions on the regional scale yield is also important to predict the effects of climate change on food supply. To estimate regional yields, we need a simple but accurate measure of the crop calendar (i.e., the distribution of planting dates), which has a strong influence on grain yield. In this article, we modeled the dependence of the crop calendar on rainfall patterns based on a survey of the region's farmers as a part of an effort to provide a stronger basis for regional yield estimates. Our survey, conducted in 11 provinces for 2 years, confirmed the existence of large windows for sowing and transplanting versus narrow windows for heading and harvesting for rain-fed lowland rice culture in all the provinces. Variable water, soil, and toposequential conditions in the paddy fields were responsible for the large sowing and transplanting windows, whereas the use of photoperiod-sensitive varieties explained the narrow windows for heading and harvesting. The crop calendar was well expressed as a function of cumulative precipitation from June onward. When the crop calendar model was combined with a simple phenologybased model that uses growing degree-days adjusted by a day-length factor, we could estimate the rice crop calendar under rain-fed lowland conditions with acceptable accuracy. The model described in this article will be combined with a crop growth model to improve regional yield estimates for rain-fed lowland rice.

Climate change will have significant impacts on the rain-fed rice production ecosystem, and particularly on the ecosystem's hydrology and water resources. Under rain-fed lowland conditions, substantial variations among fields in grain yield are commonly observed, but a method that can account for field-scale yield variability to produce regional-scale yield estimates is lacking, thereby limiting our ability to predict future rice production under changing climate and variable water resources. In this study, we developed a model for estimating regional yields of rainfed lowland rice in Northeast Thailand, by combining a simple crop model with a crop calendar model. The crop model incorporates the effects of two important resources (water and nitrogen) on crop growth. The biomass accumulation is driven by water use, whereas the nitrogen supply determines canopy development and thereby constrains crop water use. Accounting for the wide range of planting dates and the strong photoperiod-sensitive characteristics of rice varieties through the calendar model is an essential component in determining regional yield estimates. The present model does not account for the effects of mid-season drought or flooding, but was nonetheless able to explain the spatial and temporal yield variations at the province level for the past 25 years. Thus, it can be used as a prototype for simulating regional yields of rainfed lowland rice.

Rice productivity in rainfed paddy fields varies with seasonal changes of water availability in which the conditions of flooding are affected by the water balance. Hydrometeorological measurements were performed in a rainfed paddy field in Northeast Thailand from July 2004 to December 2006 to analyze the water balance. As a result of our measurements, climatologically conditions were classified as semi-humid with an annual precipitation of 1,100 mm/year and annual potential evaporation of 1,660 mm/year in both the year. The surface layer of the paddy soil was clayey and the hydraulic conductivity was very low, so groundwater levels remained below the soil surface even under flooded conditions during the rainy season. Seasonal changes in the amount of soil water were very small, comprising only less than 16% of the total precipitation during the rainy season. Consequently, an effective precipitation of less than 180 mm was enough to establish standing water in the rainfed paddy field.