Using Weather Data to Estimate Forage Production of Nilegrass

Abstract

In the previous investigation [Lee et al. (2004)], the aboveground fresh-based forage yield at harvest was found best-fitted to a four-variable linear regression model established from seasonal accumulated values of weather data. In this study, whether forage production during the growing periods may also be estimated by a multivariate regression model from weather data was further investigated. Both fresh and dry weights of the aboveground plant parts and plant height of nilegrass (Acroceras macrum Stapf) were measured from field experiments conducted at Taiwan Livestock Research Institute Experimental Farm (Hsinhua) during eight growing seasons from 2002 to 2004. The meteorological data were collected from on-site weather station to calculate the cumulative values along plant development during growing seasons, including the accumulated values of daily mean air temperature (DMAT) daily precipitation (DP), daily evaporation (DE), daily irradiance (IR), and daily sunshine hours (DSH). These climatic variables were used to correlate with plant height and forage production. Results showed that changes of forage production and plant height were shown linearly corresponding to the cumulative values of DMAT, DP, DE, IR, and DSH. By multiple linear regression analyses, estimations of fresh based and dry based forage production may be improved by two-variable linear regression models containing DP and IR, with R^2=0.667 (P＜0.0001) for fresh-based and R^2=0.767 (P＜0.0001) for dry-based forage production, respectively. Plant height may be best estimated by a two-variable model (R^2=0.808, P＜0.0001) containing DP and DE. When validating the models with another set of data from different growing seasons, models gave slight over-predicted values of fresh-based and dry-based forage production and plant height. Results also indicated that forage fresh weight was linearly correlated with plant height (R^2=0.702, P＜O.0001) and forage dry weight (R^2=0.874, P＜0.0001). Accordingly, unit forage production of nilegrass can be estimated by plant height at the time of sampling.