Prediction of the natural environmental high temperature influences on mid-season rice seed setting rate in the middle-lower yangtze river valley

Abstract

The impact of high temperature on rice yield has been evaluated by using simulation models or conducting experiments with controlled high temperature and sowing times. In this paper, observed daily maximum temperature data at 6 representative stations in the Middle-Lower Yangtze River Valley (MLYRV) from 1984 to 2013 was analyzed to examine the daily relationship between rice seed setting rate (RSSR) and the maximum temperature by using data from the mid-season late-maturing indica rice variety regional experiments conducted in field conditions at the same representative stations from 2004 to 2011 (totally using 615 samples and 69 rice varieties). The results indicated that RSSR appears to be sensitive to high temperature from 36 days before full heading to 4 days before full heading (with the significance of the negative correlation between RSSR and maximum temperature in this period above 99 % confidence level) and the most sensitive at about 14 days before full heading (near the meiosis phase), indicating that for the mid-season late-maturing indica rice variety in the MLYRV, more attention should be paid to the high temperature damage at the meiosis stage. According to the extracted high temperature sensitive period, statistical forecast models were established to predict the regional rice high temperature damage in the MLYRV by using atmospheric circulation indices in preceding 12 months with a correlation coefficient between the predicted and observed heat stress index of 0.95 and a normalized root mean square error (NRMSE) of 28.4 %. In addition, a high temperature-induced rice sterility simulation model was also used to quantitatively forecast the meiosis phase high temperature influence on rice at the site scale. The NRMSE of the simulated and forecasted relative seed setting rate was 4.74 % and 2.84 %, respectively. In conclusion, the presented prediction models were useful to improve the rice high temperature damage forecast and were expected to be helpful to rice high temperature disaster prevention and reduction in the MLYRV.