Rice production in Asia was forecasted to increase by about 25% from 1999 to 2020 using the IMPACT model of the International Food Policy Research Institute (IFPRI). Based on this forecast, different scenarios were explored in a study to determine production requirements in irrigated rice for the given time period based on assumptions on harvest area and production growth on irrigated and non-irrigated rice land (Witt et al., 2002). On average, required rice production on irrigated land in Asia would need to increase at an annual rate of about 1.25% from 406 Mio. t in 1999 to about 527 Mio. t by the year 2020. This would assume a slight increase in irrigated rice land of 0.15% yr^-1, so that yields would have to increase by 1.1% yr^-1 from about 5.3 t ha^-1 in 1999 to 6.7 t ha^-1 in 2020.

Using a large data set from on-farm experiments with irrigated rice in six Asian countries conducted between 1997 and 1999, fertilizer use, recovery efficiencies and indigenous nutrient supplies for N, P and K measured in farmers' fields were used as initial input parameters for 1999 to simulate fertilizer requirements until 2020 using a modification of the QUEFTS model (Janssen et al., 1990). Two scenarios were evaluated with i) no changes in nutrient use efficiencies and a baseline growth in total fertilizer use of 1.1% yr^-1 comparable to the required growth rate in yield (scenario 1), and ii) annual increases in fertilizer rates of 0.55% for N, 0% for P and 4% for K, assuming efficient fertilizer use and balanced nutrition through sitespecific nutrient management (SSNM) by 2020 (scenario 2). Model simulations and results from on-farm evaluation of SSNM suggest that future yield and production requirements are likely to be met, if farmers had access to improved nutrient management strategies while gradually improving the general crop management. Germplasm was assumed to improve at rates comparable to those of last 30 years. With scenario 1, only 90% of the targeted yield of 6.7 t ha^-1 could be achieved in 2020 so that production would fall short by 56 Mio. t because of inefficient use of fertilizer N and unbalanced fertilization (N:P₂O₅:K₂O ratio of 6.3 : 1 : 0.75). In scenario 2, yield and production targets were met through proper N management and a N:P₂O₅:K₂O ratio of 3.1 : 1 : 1.6 that would be required by the year 2020. Total fertilizer consumption in 2020 was similar for both scenarios but fertilizer K would have to increase at the expense of fertilizer N and P in scenario 2 (see Fig. 1). Data from on-farm testing of SSNM in 1997-2000 supported the simulation results showing increases in yield (+7%), profit (+55 US$ ha^-1 crop^-1), N recovery from applied fertilizer (+24%) and agronomic N use efficiencies (+29%) compared with the farmers' fertilizer practice (Dobermann et al., 2004). Responses to additional fertilizer K were evaluated where required and yield increases ranged from 0.2-0.4 t ha^-1.

References
Dobermann, A., Witt, C. and D. Dawe (eds.). 2004. Increasing the productivity of intensive rice systems through site-specific nutrient management. Enfield, NH (USA) and Los Baños (Philippines): Science Publishers, Inc., and International Rice Research Institute (IRRI). p 1-420.

Janssen, B.H., Guiking, F.C.T., van der Eijk, D., Smaling, E.M.A., Wolf, J. and H. van Reuler. 1990. A system for quantitative evaluation of the fertility of tropical soils (QUEFTS). Geoderma. 46:299-318.

Witt, C., Dobermann, A. and D. Dawe. 2002. Implications of site-specific nutrient management in irrigated rice on future fertilizer use in selected Asian countries. Proceedings of the IFA Regional Conference for Asia and the Pacific, 18-20 November 2002, Singapore [online]. Available at http://www.fertilizer.org (last update 2002; accessed 20 Oct. 2006). Paris: International Fertilizer Association (IFA).