Calibration and Performance Evaluation of the STICS Crop Model using Destructive and Micrometeorological Flux Datasets for Corn in Eastern Canada

The Use of a crop growth model in a new agro-environmental context requires calibration and performance evaluation of the model. In particular, cultivars representative of a given ecoregion need to be defined before the model could be applied. In this study a generic corn cultivar was calibrated and evaluated in the STICS crop model (Brisson et al., 1998) for the Mixedwood Plains ecozone of Eastern Canada using destructive and micrometeorological datasets. In this ecozone, extending from Southwestern Quebec to Southern Ontario, corn is the largest grain crop (AAFC, 2010) and the available Crop Heat Units (CHU; Brown and Boostma, 1993) for plant growth varies between 2500 CHU to 3500 CHU. First, a cultivar was calibrated using a destructive dataset of Leaf Area Index (LAI) and shoot dry biomass (7 measurements spread through the growing season) and final yield from Ottawa (ON, Canada) in 1998. Seven parameters, concerning LAI dynamic, phenology (duration of vegetative and reproductive stages) and yield (maximum grain number) have been calibrated. Performance of this new cultivar, named CanMaïsNE, has been evaluated by comparing LAI, biomass and yield predictions on 12 datasets collected from 1994 to 2008 in the Ottawa area, representative of the 2500-2900 CHU range (Jego et al., submitted). CanMaïsNE gave good predictions of LAI, biomass and yield. The Root Mean Square Error (RMSE) was 28.1 % for LAI, 17.5 % for biomass, and 10.1 % for yield predictions. This work is likely the first calibration and performance evaluation of the STICS crop model for corn in North America. Moreover, these new grain corn cultivars, adapted to shorter growing season, open new opportunities for using STICS in northern countries. However it remains some discrepancy between predictions and measurements. In order to refine the calibration of the new cultivar daily and ten-day total biomass datasets were calculated using C02 eddy flux and chamber soil respiration measurements made in the Ottawa area during several growing seasons (1996, 1998, 2000, 2002 and 2006) following the approach presented in Pattey et al. (2001). The daily changes in the accumulation of biomass during the growing season enable to assess the temporal resolution of the response of STICS crop model predictions to weather variations. A comparison of the predicted and measured total and above-ground biomass showed that the model tended to overestimate root biomass. The average predicted Root to Shoot ratio (R/S) before calibration was about 0.35 while the measurements and the bibliographic references gave average R/S close to 0.17 (Amos et al., 2006). After calibration of the specific root length parameter from 11000 cm g^-1 to 22000 cm g^-1 the predicted average R/S was close to 0.17. Finally, although biomass was well predicted compared to destructive measurements, the comparison with the daily and ten-day biomass measurements calculated from fluxes data showed some discrepancy. Indeed, in 1996, and 2006, the predicted biomass was underestimated at the beginning of the growing season (June) while it was overestimated at the end (from late August). Two ways could be explore to improve the prediction: 1) calibration of the radiation use efficiency parameters which can be adjusted separately during juvenile, vegetative and reproductive stages and/or 2) calibration of the temperature threshold parameters which determines optimal temperature for crop growth. This work shows that micrometeorological C0₂ fluxes measurements can be very useful for crop model calibration and performance evaluation. However, some discrepancies between predictions and measurements still remain. Some of them can be due to wrong estimations of crop water requirement and inaccurate water stress intensity, in particular in Eastern Canada, where only rainfed corn is cultivated. Daily evapotranspiration calculated from fluxes measurements can be an efficient way to assess the performance and temporal resolution of predicted water stress and water use efficiency during growing season (see Pattey et al., this meeting), and then increase the model performance.

References Agriculture and Agri-Food Canada, 2010. Canada: grains and oilseeds outlook. Publ. 1496-967X. AAFC, Ottawa, ON. Amos, B., Walters, D.T., 2006. Maize root biomass and net rhizodeposited carbon: an analysis of the literature. Soil Sci. Soc. Am. J., 70: 1489-1503. Brisson, N., Mary, B., Ripoche, D., Jeuffroy, M.H., Ruget, F., Nicoullaud, B., Gate, P., Devienne-Barret, F., Antonioletti, R., Durr, C., Richard, G., Beaudoin, N., Recou,s S., Tayot, X., Plenet, D., Cellier, P., Machet, J.M.,. Meynard, J.M, Delécolle, R., 1998. STICS: A generic model for the simulation of crops and their water and nitrogen balances. I. Theory and parameterization applied to wheat and corn. Agronomie, 18: 311-346. Brown, D.M., Bootsma, A., 1993. Crop Heat Units for corn and other warm season crops in Ontario. Publ. 111/31 OMAFRA, Ottawa, ON. Jégo, G., Pattey, E., Bourgeois, Tremblay, N., Drury, C.F., 2010. Calibration and performance evaluation of a corn cultivars adapted to Eastern Canada in the STICS crop model. Sustain. Agron.: submitted. Pattey, E., Strachan, I.B., Boisvert, J.B., Desjardins, R.L. and McLaughlin, N.B., 2001. Detecting effects of nitrogen rate and weather on corn growth using micrometeorological and hyperspectral reflectance measurements. Agric. For. Meteorol., 108: 85-99.