| 摘要: |
| 冬小麦条锈病是影响冬小麦生产的主要病害之一，陕西、甘肃、宁夏三省（区）是条锈病主要菌源越冬地，也是向我国华北等麦区传播的重要桥梁，对该区域进行总体的监测和预警对于预防条锈病流行意义重大。由于研究区域幅员辽阔，气象和地形条件差异较大，条锈病发病规律和发病特点差异明显，因此无法将整个研究区域作为一个整体来研究。考虑地理位置因素对研究区进行了一级分区，考虑地形因素对研究区进一步进行了二级分区。进行一级分区时以县级行政区域为单元，通过筛选后选定旬均温度、年最低温度、旬降雨量和越冬菌源地距离等8个因子，结合历年发病等级统计数据，采用除趋势对应分析（DCA）和聚类分析相结合的方法进行分析，初步将研究区域划分为5个一级区，一级区内环境因子间具有较高的同质性。地形因素的叠加增加了冬小麦条锈病分布格局的复杂性，地理位置相近的不同地形区条锈病发病差异可能很大。故在一级区的基础上通过数字高程模型（DEM）设定海拔阈值的方法得到12个二级区，二级分区内条锈病发病规律相同、发病等级相近。可以作为下一步冬小麦条锈病预测研究的基本单元。 |
| 关键词: 冬小麦 条锈病 分区 预测 |
| DOI：10.7621/cjarrp.1005-9121.20160431 |
| 分类号: |
| 基金项目: |
|
| THE RESEACH ON THE PARTITION OF WINTER STRIPE RUST BASED ON CLUSTERING ANALYSIS |
|
Gao Jianmeng1, Wang Limin2, Gao Lingwang1
|
|
1.China Agricultural University，Beijing 100193，China;2.Institute Of Agricultural Resources And Regional PlanningChinese Academy Of Agricultural Sciences，Beijing 100081，China
|
| Abstract: |
| Winter wheat stripe rust is one of the key diseases that influences the yield of winter wheat in China.The study area consists of two provinces of Shaanxi Gansu and one district of Ningxia.The main overwintering region of wheat stripe rust in China is located in the south west of the study area and it is also the bridge area of disease spreading outwards to the wheat-planting area in North China plain and other wheat-planting field nearby as well.As a result，it is of great significance to monitor and forecast the dynamic of wheat stripe rust in the study area.As the study area is quite big，the difference of environment factors and disease occurrence characteristic among different parts are so significant that we cannot regard the whole area as one entire part.So we separate the area into several parts by two steps of partition，among different parts of the study area，the location and the distance from overwintering region are taken into consideration in the first step.During the second partition procedure，the terrain factor is mainly considered.In order to get the parts where the environment factors that influence or even determine the occurrence level of wheat stripe rust are similar，county is defined as the basic partition unit in the first partition step.Referring to the life cycle of the wheat stripe rust，8 parameters including the mean air temperature in a XUN period（about 10days）the mean precipitation in a XUN period the minimum temperature in a year and the distance from overwintering area factors are selected for partition.During the first partition process，the DCA and the clustering analysis are combined together and the 8 parameters above are set as the input information of both analyses for partition.The actual occurrence level of wheat stripe rust in the study area in history is also taken into account for validation then.At last the whole study area is divided into 5 parts after the first partition step.The environment in the same part is similar and changes heavily in different parts.The terrain factor makes the disease distribution more sophisticated，the disease occurrence level in regions nearby may differ a lot among different terrain regions.On the basis of the first partition step，elevation threshold in the Digital Elevation Model（DEM）image is set to differentiate different terrainAfter the second partition step process，12 parts are got and the disease occurrence characteristics are the same and the disease level is much more similar than the first partition step.The results imply that partition based on the weather and elevation factors is feasible and the partition result is potential to be the basic unit in the disease forecast process. |
| Key words: winter wheat stripe rust partition forecast |
