Johnson Dennis A, Alldredge J Richard, Hamm Philip B
Plant Pathologist, Department of Plant Pathology, Washington State University, P.O. Box 646430, Pullman 99164-6430.
Associate Professor, Program in Statistics, Washington State University.
Plant Dis. 1998 Jun;82(6):642-645. doi: 10.1094/PDIS.1998.82.6.642.
A regional potato late blight forecasting system for irrigated potatoes in the semiarid environment of the Columbia Basin was expanded by developing specific forecasting models for four vicinities throughout the Basin. Relationships between weather and outbreaks of late blight at the locations over a 27-year period were examined using logistic regression analysis. The response variable was a year either with or without a late blight outbreak. An indicator variable representing the occurrence of an outbreak during the preceding year (Yp) and number of days of rain during April and May (Ram) correctly classified the disease status (presence or absence of late blight) of 89, 82, 78, and 78% of the years at Prosser, Washington, Hermiston, Oregon, and Hanford and Othello, Washington, respectively. The percentage of years with disease outbreaks correctly classified was 93, 85, 79, and 79% at the four respective locations. All years with late blight outbreaks and 96% of the total years were correctly classified using data from at least one of the four locations. These predictors are particularly important early in the growing season and can be used to make area forecasts. A second set of predictors, Yp and number of days of rain in July and August (Rja), for Hermiston and Hanford, and a third set, Yp, Ram, and Rja, for Prosser and Othello were found effective for making additional late blight forecasts later in the growing season.
通过为哥伦比亚盆地四个不同区域开发特定的预测模型,扩展了适用于该盆地半干旱环境中灌溉马铃薯的区域马铃薯晚疫病预测系统。利用逻辑回归分析,研究了27年间这些地点的天气与晚疫病爆发之间的关系。响应变量为有或没有晚疫病爆发的年份。一个代表上一年爆发情况的指标变量(Yp)以及4月和5月的降雨天数(Ram),分别正确分类了华盛顿州普罗瑟、俄勒冈州赫米斯顿以及华盛顿州汉福德和奥赛罗89%、82%、78%和78%年份的病害状况(晚疫病存在与否)。在四个相应地点,正确分类有病害爆发年份的百分比分别为93%、85%、79%和79%。利用来自四个地点中至少一个地点的数据,所有有晚疫病爆发的年份以及96%的总年份都被正确分类。这些预测因子在生长季节早期尤为重要,可用于进行区域预测。还发现,对于赫米斯顿和汉福德,第二组预测因子Yp以及7月和8月的降雨天数(Rja),对于普罗瑟和奥赛罗,第三组预测因子Yp、Ram和Rja,在生长季节后期进行额外的晚疫病预测时是有效的。