基于投影寻踪的农业基本旱情评估

针对干旱指标的不相容性以及现有方法对评估等级的分辨率较粗略等问题，提出了农业基本旱情评估的投影寻踪方法（Projection Pursuit，PP）。根据农业旱情等级划分标准，构建了能够反映农业基本旱情系统变异信息的投影指标函数，引入自由搜索（Free Search，FS）算法优化投影指标函数寻求最佳投影方向，利用最佳投影方向信息研究了各种干旱指标对农业基本旱情评估的影响程度。考虑到各种旱情监测信息的时空代表性以及不同指标对旱情评估的影响程度，在甘南县域尺度下，选用降水量距平百分比和连续无雨日数两种指标计算了投影值，建立了农业基本旱情评估的投影寻踪模型。实例应用表明，该模型评估结果符合当地干旱实情，且对旱情等级划分清晰，为农业基本旱情评估提供了一种新方法。

Assessment of agricultural basic drought based on projection pursuit

The method of projection pursuit (PP) was presented to assess the agricultural basic drought for solving the problems such as the incompatibility of drought indexes and the rough grade levels assessed by other methods. Based on the grade division standard of agricultural drought, the projection function which could reflect the variability of the agricultural basic drought system was constructed, and free search (FS) was introduced to optimize the function and seek the optimum projection vector, while the data of the vector were used to research the influential level of each drought index. In Gannan County, considered the space-time representativeness of observation data and influential degree of drought indexes on drought assessment, the percentage of precipitation anomalies and the number of days which had no available rainfall were selected to establish assessment model of agricultural basic drought based on projection pursuit. The achieved results suggested that the conclusion assessed by the model was accordant with the factual drought, and the projection pursuit model could identify the grade of agricultural drought clearly. Projection pursuit model is an effective method for assessing the agricultural basic drought.