Representing Crop Growth

We conducted a pilot study for central Iowa where CERES-Maize was used to simulate crop yields when supplied with 10-year data sets representing (1) observed weather conditions, (2) a climate produced by a regional climate model driven by the NNR, (3) a contemporary climate produced by a global model downscaled by the regional model, and (4) a future scenario climate (2040-2049) produced by a global model downscaled by the regional model. Results of each of these were compared to observed yields in this region. The validation data on corn yields for Ames, IA were taken from annual yields for the north-central reporting district of Iowa, and therefore represent a regional average rather than results from a single locale.

Results in Table 15.1 and shown in the lower panel of Figure 15.2 suggest that the crop model has higher interannual variability (standard deviation) than the observed values, but produces a mean yield close to observed yield. The climates produced by the regional model produced mean yields well below observed yields and standard deviations well above observed levels. When results of a global model for the contemporary climate are used to drive the regional model, the mean yield is modestly reduced, but the standard deviation is substantially reduced. The combination CERES/RegCM2 produces a doubling of maize yields for Ames under the future scenario climate and reduced variance. The reason for the large increase in yield is not completely clear, but may be a combination of more soil moisture and higher mean temperatures, but with little or no increase in the daily maximum temperature during critical parts of the growing season in the scenario climate.

From this pilot study, we conclude that the crop model is very sensitive to biases introduced by a regional climate model. The crop model seems to react more strongly than actual crops to interannual variability in precipitation. This sensitivity is exacerbated by inability of the climate models to simulate the proper distribution of rainfall intensities. The consequences of higher daytime maximum temperatures are amplified by crops (Mearns et al., 1984), so any termperature biases in the climate models will substantially impact yields. By these means, the crop model has exposed weaknesses in the climate model for simulating agricultural impacts of climate change.

Crops are not passive acceptors of externally imposed climate, but rather are being recognized for their role as a coupling mechanism between the soil and atmosphere. Sparks et al. (2002) suggest that extensive plantings of corn and soybeans in the Midwest are contributing to higher dew-point temperatures, which exacerbate heat waves (Kunkel et al., 1996). Kalnay and Cai (2003) compare a reanalysis of upper air data with surface observations to conclude that about half of the observed decrease in diurnal temperature range over the last 50 years is due to changes in land use, primarily agriculture. These and other reports suggest a need for more direct coupling of interactive ecosystem models, such as crop models for the central United States, that incorporate biophysical processes into weather and climate models for more accurate simulation of crop impacts on local weather. This, in turn, should lead to more accurate simulation of crop physiology and yield.

Was this article helpful?

0 0

Post a comment