Simulation of grass sward dry matter yield in Slovenia using the LINGRA-N model

Calibration and validation of the LINGRA-N model were performed using herbage dry matter (DM) yield data from field studies conducted at three locations in Slovenia. Calibration was done by minimising root mean square error (RMSE) and validation by using RMSE and Willmott’s index of agreement (d_w). Calibration of LINGRA-N was not successful for the experiment conducted on permanent grassland in Ljubljana in the period 1974-1993 (RMSE% = 14%, d_w = 0.37). Better results were obtained for grass monocultures in Jablje (J) and RakiÄan (R) in the period 1998– 2013, with the best fit for cocksfoot (Dactylis glomerata L.; RMSE% = 12%, d_w = 0.84). Fifty-year simulations were performed for cocksfoot (J-DG) and timothy grass (Phleum pratense L.) in Jablje (J-PP) and perennial ryegrass (Lolium perenne L.) in Jablje (J-LP) and RakiÄan. Outliers with very low simulated herbage DM yield were detected only in the second half of the study period and were associated with drought and/or high maximum air temperatures. A time series analysis of annual potential yield values showed a statistically significant (P=0.05) negative trend for J-LP (–24 kg DM ha^–1year^–1) and J-PP (–29 kg DM ha^–1year^–1). A change in the variability of the reduction factor for crop growth due to drought was already noticeable.