首页|季节性淹水湿地表层土壤有机碳含量遥感预测及空间分布特征

季节性淹水湿地表层土壤有机碳含量遥感预测及空间分布特征

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
为探讨鄱阳湖季节性淹水湿地土壤有机碳的空间分布特征及遥感方法在土壤有机碳估算中的适用性,依托江西鄱阳湖国家级自然保护区,选择蚌湖、常湖池和泗洲头湿地为研究区域,基于野外实测土壤有机碳含量数据和同期的Landsat8 OLT遥感影像,采用遥感图像处理和GIS技术提取影像中遥感特征因子,构建遥感参数与土壤有机碳的一元线性、一元曲线和多元逐步线性回归模型,通过对比分析选择最优遥感估算模型,预测鄱阳湖季节性淹水湿地表层(0～20 cm)土壤有机碳含量.结果表明,提取了影像中33个遥感特征因子,包括7个波段的反射率值(b1～b7)、4个植被指数(NDVI、SR、SAVI、EVI)、第一主成分特征(PCA1)、单波段纹理特征的均值(MEAN)、熵(ENT)和相关性(COR),其中纹理特征是研究区土壤有机碳含量预测的重要遥感因子,其与土壤有机碳含量构建的多元逐步线性回归模型拟合效果最优,模型决定系数R2=0.772,平均相对误差45.53%,均方根误差2.417.遥感反演发现,研究区预测表层土壤有机碳含量主要集中在0～20 g/kg,土壤有机碳含量平均值约为10.75 g/kg.
Remote sensing prediction and spatial distribution characteristics of content of organic carbon in surface soil of seasonal flooded wetlands in Poyang Lake
Banghu Lake,Changhu Lake,and Sizhoutou wetland on the Poyang Lake National Na-ture Reserve in Jiangxi Province were used to study the spatial distribution characteristics of the content of organic carbon in surface soil in seasonal flooded wetlands of Poyang Lake and the applicability of remote sensing methods for estimating the content of organic carbon in surface soil.Remote sensing image process-ing and GIS technology were used to extract feature factors of remote sensing from the images based on the data about the content of organic carbon in soil measured in the field and Landsat8 OLT remote sensing im-ages from the same period.The regression models of univariate linear,univariate curve,and multiple step-wise linear for parameters of remote sensing and the content of organic carbon in soil were constructed.The optimal estimation models of remote sensing were selected by comparing and analyzing to predict the con-tent of organic carbon in the surface layer(0-20 cm)of seasonal flooded wetlands in Poyang Lake.The re-sults showed that 33 feature factors of remote sensing including reflectance values(b1-b7)in 7 bands,4 vegetation indices(NDVI,SR,SAVI,EVI),first principal component feature(PCA1),the mean(MEAN),entropy(ENT),and correlation(COR)of single band texture features were extracted from the images.Texture features were important factors of remote sensing for predicting the content of organic carbon in the areas studied,and their fitting effect with the multiple stepwise linear regression model Y=42.708-2.817Xb3MEAN-4.887Xb5COR+0.667Xb7MEAN(b3MEAN,b5COR and b7MEAN representing the mean value,correlation and mean value of texture features in bands 3,5 and 7,respectively)constructed for the content of organic carbon in soil was the best.The determination coefficient of model,R2 was 0.772,with an aver-age relative error(MRE)of 45.53%and a root mean square error(RMSE)of 2.417.The results of re-mote sensing inversion showed that the predicted content of organic carbon in surface soil in the areas stud-ied was mainly concentrated at 0-20 g/kg,with an average content of organic carbon in soil about 10.75 g/kg.It is indicated that it is feasible to use remote sensing to predict the content of organic carbon in soil of wetlands.

Poyang Lake wetlandsoil organic carbonremote sensing predictionfeature factors of remote sensingcarbon cyclecarbon storage in soilcarbon fixation capacity of soil

邹霞、钱海燕、周杨明、黄灵光、杨梅花

展开 >

东华理工大学地球科学学院,南昌 330013

江西师范大学地理与环境学院,南昌 330200

江西省自然资源政策调查评估中心,南昌 330046

豫章师范学院生态与环境学院,南昌 330103

展开 >

鄱阳湖湿地 土壤有机碳 遥感预测 遥感特征因子 碳循环 土壤碳储量 土壤固碳能力

江西省自然科学基金江西省自然科学基金国家自然科学基金东华理工大学博士科研启动基金

20212BAB20300220212BAB20502241561105DHBK201909838

2024

10.13300/j.cnki.hnlkxb.2024.03.011

华中农业大学学报
华中农业大学

华中农业大学学报

CSTPCD北大核心
影响因子：1.09
ISSN：1000-2421
年,卷(期)：2024.43(3)
  • 36