A more immediate prerequisite to the successful identification of profitable tactical management decisions based on seasonal forecasts is that crop simulation models must first be able to effectively reproduce the characteristics of local cropping systems (Phillips et al. 1998). This presents difficult challenges for a range of models available for use in West Africa, because of such peculiarities as surface crusting, low planting densities and heterogenous canopies, etc. Critical advances were however made, e.g. in the adaptation of PP response in the CERES-Sorghum and -Millet modules of the DSSAT-Century cropping systems model. In its original form, a linear photothermal response resulted in underestimates of crop cycle duration for late-maturing landraces (Fig. 19.2a). A threshold-hyperbolic function was shown to simulate cycle duration (Fig. 19.2b) for both PP-sensitive and PP-insensitive material (Folliard et al. 2004). This work allowed to revisit the popular, but incorrect photothermal approach widely used in modeling crop development (Robertson 1973). Other ongoing work seeks to improve models poor ability to simulate yield components because of flaws in the timing of stem elongation (occurring after panicle initiation in CERES) and the inadequate partitioning of assimilates resulting in inflated harvest indices for Sudano-Sahelian landraces.
Was this article helpful?