摘要
以广东省清远市清新区内某针阔混交林为研究区,基于Sentinel-1和Sentinel-2遥感数据,结合地面实测数据,构建了蓄积量估测模型。研究结果表明:使用Sentinel-1数据+Sentinel-2数据为数据源,以距离相关系数法筛选的特征,采用随机森林模型构建的蓄积量估测模型精度最高,相对均方根误差RRMSE为0.2217,决策系数R^(2)为0.61,模型拟合效果良好;预测的蓄积量集中分布在300~400 m^(3)/hm^(2),与实际值进行配对t检验,结果表明预测值和实际值之间不存在显著差异(t=0.001,P=0.999>0.01),进一步证明了模型的可靠性。
This study takes a coniferous broad leaved mixed forest in XX city,Guangdong province as the research area based on Sentinel-1 and Sentinel-2 remote sensing data.Combined with ground measurement data,a volume estimation model is constructed.The research results indicate that using Sentinel-1 data+Sentinel-2 data as the data source,using the distance correlation coefficient method to screen features,and using the random forest model to construct the volume estimation model has the highest accuracy,with a relative root mean square error(RRMSE)of 0.2217 and a decision coefficient R^(2)0.61.The model fitting effect is good;The predicted volume is concentrated between 300 m^(3)/hm^(2)and 400 m^(3)/hm^(2).Paired t-tests were conducted with actual values,and the results showed that there was no significant difference between the predicted and actual values(t=0.001,P=0.999>0.01),further proving the reliability of the model.
作者
林瑞明
Lin Ruiming(Qingxin District Forestry Bureau,Qingyuan 511800,Guangdong,China)
出处
《绿色科技》
2024年第9期151-154,162,共5页
Journal of Green Science and Technology
