Soltani, A. and Meinke, H. and de Voil, P. (2004) Assessing linear interpolation to generate daily radiation and temperature data for use in crop simulations. European Journal of Agronomy, 21 (2). pp. 133-148.
Full text not currently attached. Access may be available via the Publisher's website or OpenAccess link.
Article Link(s): http://dx.doi.org/10.1016/S1161-0301(03)00044-3
Publisher URL: http://www.elsevier.com
In this study we compare simulated, water non-limited yields obtained from the use of actual daily radiation and temperature data against yields simulated with weather data estimated using linear interpolation (LI) from monthly means. Using LI-generated radiation and temperature data (instead of stochastically generated or actual data) greatly reduces the effort required for the simulation task. LI data do not have the day-to-day variations that occur naturally. This paper determines under what condition crop model yields are insensitive to this day-to-day variation. Four locations across Iran with different climates were selected for the study. For each location a complete set of 20-year daily radiation and temperature data were available. Another 20-year sets of daily data were produced by LI from monthly means values. For the assessment we used the wheat, maize and soybean models of Decision Support System for Agrotechnology Transfer (DSSAT) at several sowing dates. Generally, simulated crop yields using actual and LI weather data showed similar responses to sowing date. Across crop models and locations, mean, variance and distribution differences between yields simulated with actual and LI data were significant in 23, 24 and 22% of cases, respectively. In many cases, the lack of day-to-day variation in LI data and hence its inability to reproduce extreme temperatures (cold or hot events) caused these significant differences, especially at unseasonally early or late sowing dates. In other cases, over-prediction of biomass production with LI data resulted in an over-prediction of yield. However, this growth over-prediction is only significant where growing season air temperatures are optimal or supra optimal for growth and have a high day-to-day variation (standard deviation greater than 3.5 °C). We concluded that for conventional sowing dates and in situations where unseasonally high or low temperatures are either unlikely or of little importance, the LI method could be used to generate daily radiation and temperature data as input for crop simulations.
|Additional Information:||© Elsevier B.V.|
|Keywords:||Weather generation; temperature; solar radiation; model; Decision Support System for Agrotechnology Transfer (DSSAT).|
|Subjects:||Agriculture > Agriculture (General) > Agricultural meteorology. Crops and climate|
Science > Statistics > Simulation modelling
|Deposited On:||28 Jun 2004|
|Last Modified:||17 Nov 2010 02:39|
Repository Staff Only: item control page