摘要
针对植被覆盖度动态变化对地区生态的深刻影响,急需深入分析其变化规律的问题,基于湖南省2000—2020年MODIS NDVI遥感数据,结合同期气象数据,通过Hurst指数、Sen+Mann-Kendall趋势分析、Pearson相关分析等方法研究湖南省植被覆盖度时空变化,并提出一种注意力机制的CNN-GRU植被覆盖度预测模型。2000—2020年湖南省的植被覆盖度总体上显示出波动性的增长趋势,21 a间植被覆盖度增长率为0.0048/a;Hurst指数小于0.50大于0.35的区域占湖南省面积的56.66%;基于注意力机制的CNN-GRU植被覆盖度预测模型RMSE的值为0.0710,MAE的值为0.0595。结果表明:湖南省大部分地区具有相对稳定且较高的植被覆盖度且未来植被覆盖度将呈现微弱下降趋势,本文提出的基于注意力机制的CNN-GRU植被覆盖度预测模型在评价指标上优于GRU模型以及CNN-GRU模型。
Aiming at the profound impact of the dynamic change of vegetation cover on regional ecology and the urgent need to analyze its change rule in depth,this paper is based on the MODIS NDVI remote sensing data of Hunan Province from 2000 to 2020,combined with the meteorological data of the same period,and studied the temporal and spatial change of vegetation cover in Hunan Province through the methods of Hurst index,Sen+Mann-Kendall trend analysis,Pearson correlation analysis,etc.and proposed an attention mechanism CNN-GRU prediction model.The vegetation cover in Hunan Province from 2000 to 2020 showed a fluctuating growth trend,with a growth rate of 0.0048/a in 21 a.The Hurst index was less than 0.50 and greater than 0.35,and the percentage of lakes with a Hurst index less than 0.50 and greater than 0.35 was 0.0048/a.The vegetation cover in Hunan Province was 0.0048/a in 21 a,and the percentage of lakes with a Hurst index less than 0.50 and greater than 0.35 was 0.0048/a.The area with Hurst index less than 0.50 and greater than 0.35 accounted for 56.66%of the area of Hunan Province;the value of RMSE of the CNN-GRU vegetation cover prediction model based on the attention mechanism was 0.0710,and the value of MAE was 0.0595.The results show that most of the areas in Hunan Province have a relatively stable and high vegetation cover,and the vegetation cover will show a slightly decreasing trend in the future,and the CNN-GRU vegetation cover prediction model based on the attention mechanism proposed in this paper will be able to predict the vegetation coverage in most areas of Hunan Province.The CNN-GRU vegetation cover prediction model based on the attention mechanism proposed in this paper is better than the GRU model and the CNN-GRU model in terms of evaluation indexes.
作者
饶雨晨
孙承志
RAO Yuchen;SUN Chengzhi(School of Remote Sensing&Geomatics Engineering,Nanjing University of Information Science&Technology,Nanjing 210044,China)
出处
《测绘科学》
CSCD
北大核心
2024年第7期153-163,共11页
Science of Surveying and Mapping
基金
高分专项航空观测系统“高分航空载荷自然资源调查应用示范”项目(04-H30G01-9001-20/22)。
作者简介
饶雨晨(1998-),女,湖南怀化人,硕士研究生,主要研究方向为摄影测量与遥感。E-mail:2045719036@qq.com;通信作者:孙承志,教授,E-mail:003453@nuist.edu.cn。
