| 引用本文: | 吴迪,杨鹏,周黎勇,等.基于Sentinel-2破碎化地块灌区作物种植结构的提取[J].灌溉排水学报,0,():-. |
| WU Di,YANG Peng,ZHOU Li-yong,et al.基于Sentinel-2破碎化地块灌区作物种植结构的提取[J].灌溉排水学报,0,():-. |
|
| 摘要: |
| 【目的】探究基于Sentinel-2遥感影像的决策树分类模型提取破碎化地块灌区作物种植结构的适用性。【方法】选取新疆阿拉沟灌区为研究区，以2021年覆盖作物全生育期的Sentinel-2遥感影像为数据源，结合田间调查和Google高清影像目视解译采样，基于主要作物物候信息、NDVI时序特征等分析确定作物识别的关键期阈值，构建决策树模型进行灌区主要作物分类，并对分类结果精度验证。【结果】基于Sentinel-2提取的灌区种植结构分布图地块纹理清晰，能够满足灌区用水管理需要；构建的决策树分类模型可在灌区尺度实现作物分类，方法简便易行，总体精度达到81.56%，Kappa系数为0.7166。【结论】采用Sentinel-2遥感影像和决策树分类方法识别破碎化地块灌区复杂作物分类是可行的，可为灌区输配水决策和农业用水精细化管理提供基础信息。 |
| 关键词: Sentinel-2；灌区作物分类；NDVI时间序列；决策树；破碎化地块 |
| DOI： |
| 分类号:TP79 |
| 基金项目:新疆水利科技项目(XSKJ-2022-12) |
|
| Extraction of Crop Planting Structure in Irrigated Area of Fragmentation Field Based on Sentinel-2 |
|
WU Di1, YANG Peng2, ZHOU Li-yong3, LI Fang-song3, LI Ling-feng3, ZHANG Xu-dong2
|
|
1.China Irrigation and Drainage Development Center,Beijing ,china;2.College of Water Conservancy,Shenyang Agricultural University;3.Xinjiang Institute of Water Resources and Hydropower Research
|
| Abstract: |
| 【Objective】To explore the applicability of decision tree classification model based on Sentinel-2 remote sensing image for crop structure extraction in irrigated area of fragmented land parcel.【Method】The Alagou irrigation area in Xinjiang was selected as the research area, and the Sentinel-2 remote sensing images of the whole growth period of major crops in 2021 were used as data sources, combined with field investigation and visual interpretation sampling of Google HD images. Based on the phenological information and NDVI time series characteristics of major crops, the threshold values of the critical period for crop recognition were determined. The classification model of decision tree was constructed and the accuracy of classification results were verified.【Result】The plots of planting structure distribution map extracted from Sentinel-2 remote sensing images had clear texture and natural block, which could meet the needs of water management in irrigation areas. The established decision tree classification model can realize crop classification at the irrigated area scale. The method is simple and feasible, with an overall accuracy of 81.56% and a Kappa coefficient of 0.7166.【Conclusion】It is feasible to use Sentinel-2 remote sensing data and decision tree classification method to identify complex crop classification in irrigated areas of fragmentation field, which can provide basic information for decision-making of water supply and distribution and refined management of agricultural water in irrigated areas. |
| Key words: Sentinel-2; crops classification; NDVI time series; decision tree |
