冬小麦干旱指标及干旱预测模型研究

Drought indices and prediction models for winter wheat

干旱是河北省冬小麦生长期内主要的气象灾害之一。准确监测、预测干旱发生程度,可以为防灾、减灾、救灾提供科学的决策依据。本研究以位于河北省南部冬小麦区的南宫县为例,选取1991～2007年冬小麦全生育期农业气象观测数据及常规气象资料,基于Jensen模型得到冬小麦返青～拔节、拔节～抽穗、抽穗～乳熟、乳熟～成熟4个生育阶段的水分敏感系数;在减产百分率标准的基础上,确定了冬小麦返青后4个生育阶段以相对蒸散表示的轻旱、中旱、重旱、严重干旱4个等级冬小麦干旱指标值;并应用回归分析方法,建立了4个生育阶段的干旱预测模型。结果表明:考虑冬小麦不同发育阶段对水分的敏感程度,确定的冬小麦干旱指标值比较客观地反映了干旱程度。建立的干旱预测模型均通过了0.05的显著性检验。模型的拟合正确率70.8%,预测正确率75.0%,平均正确率71.4%;经简化干旱等级,即轻旱为1个等级,中旱、重旱、严重干旱为1个等级,则模型的拟合正确率达81.3%,预测正确率达75.0%,平均正确率达80.4%,模型预测结果可信。

Drought is one of the meteorological hazards that severely affects winter wheat production in Hebei Province.Accurate monitoring and prediction of drought occurrence provides the scientific basis for hazard control decision-making.This paper ana-lyzed drought conditions in the winter wheat production belt of Nangong County,South Hebei Province.The study used observed agro-meteorological data and regular meteorological data for 1991～2007 to establish a drought index and prediction model for winter wheat.The water sensitive coefficients of winter wheat during re-greening to jointing,jointing to heading,heading to milky maturity and milky maturity to grain maturity stages were calculated using the Jensan model.Then percent yield reduction and relative evapo-transpiration at each growth stage were used as index value to determine light drought,moderate drought,heavy drought and severe drought.A simulation test for 12 years（1991～2005） with 2006 and 2007 as evaluation periods was conducted.The results showed that the established index values for drought degree in different growth periods objectively reflected the active drought degree in the region.The criteria for drought degree took into account of the sensitivity of winter wheat at different growth stages.Regression models were used to predict drought in the four growth stages of winter wheat.The established drought prediction model results were significant at P=0.05.The correct rate of model simulating was 70.8%,correct rate of prediction was 75.0%,and the average correct rate was 71.4%.Assuming that drought was classified as level-one drought（light drought） and level-two drought（medium,heavy and severe droughts）,then the model simulating correct rate was 81.3%,correct rate of prediction was 75.0%,and average correct rate was 80.4%.In summary,model calculations and predictions were in good agreement with observed data.Thus the prediction model had practical application in early warning and control of the impact of drought in winter wheat production.