摘要: |
耕地集约化利用是国际社会普遍关注的热点问题,而目前衡量耕地集约化利用程度的指标繁多且没有统一的定义。其中,种植频率是对种植模式的硬性划分(如单季或双季),对耕地利用程度缺乏更精细的评价。复种指数则是以农业统计数据为基准计算得到,缺乏详细的空间分布信息,忽略了行政区内作物分布的空间异质性。故而现有指标不能满足耕地集约化利用的精细评价的要求。【目的】文章引入“种植强度指数”的概念,对现有评价指标进行改进,利用基于遥感数据的种植强度指数实现耕地集约化利用程度的精细化表达。【方法】该文以湖北省为研究区,融合Landsat 8遥感数据和MODIS时间序列植被指数数据,构建了人工神经网络模型估算湖北省耕地种植强度。【结果】利用BP神经网络提取的研究区耕地种植强度与验证样区耕地种植强度间决定系数达到0.923,证明了本文研究方法的可靠性。【结论】该文方法得到的高时空融合的种植强度数据集,可为智慧农业提供技术方法和基础数据,对于耕地集约化利用的研究具有重要意义。 |
关键词: 种植强度 BP神经网络 多源遥感数据 |
DOI: |
分类号:TP79 |
基金项目:国家自然科学基金项目(面上项目,重点项目,重大项目) |
|
Estimating Cropping Intensity of croplands using multi-source remote sensing data and ANN |
xumeng1,2, taojianbin1,2
|
1.The College of Urban &2.Environmental Sciences, Central China Normal University
|
Abstract: |
The intensive use of cultivated land is a hot issue of general concern in the international community, but there are many types of indicators for measuring the degree of intensive use of cultivated land and there is no unified definition. Cropping Frequency is a hard division of the planting pattern (such as single or double season), which will lose some information. The multiple cropping index is calculated based on agricultural statistics, but the statistical data lacks detailed spatial distribution information and ignores the spatial heterogeneity of crop distribution in the administrative region. Therefore, the existing indicators can not meet the requirements of the fine evaluation of the intensive use of cultivated land. [Purpose]This paper introduces the concept of “Cropping Intensity” and defines the Cropping Intensity Index to express the intensive use of cultivated land. [Method]The Landsat 8 remote sensing data and MODIS time series vegetation index data were combined to construct an artificial neural network (ANN) model to estimate the Cropping Intensity of cultivated land in Hubei Province. [Result]The estimation accuracy based on sample area verification reaches 92.3%, which proves the reliability of this method. [Conclusion]The high-temporal fusion Cropping Intensity data set obtained by this method can provide technical methods and basic data for smart agriculture, which is of great significance for the study of intensive use of cultivated land. |
Key words: Cropping Intensity, Back Propagation Neural Network, Multi-source Remote Sensing Data |