引用本文:梁 静,丁建丽,王敬哲,王 飞.基于反射光谱与Landsat 8 OLI多光谱数据的艾比湖湿地土壤盐分估算[J].土壤学报,2019,56(2):320-330.
LIANG Jing,DING Jianli,WANG Jingzhe,WANG Fei.Estimation and Mapping of Soil Salinity in the Ebinur Lake Wetland Based on Vis-NIR Reflectance and Landsat 8 OLI Data[J].Acta Pedologica Sinica,2019,56(2):320-330
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 120次   下载 250 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于反射光谱与Landsat 8 OLI多光谱数据的艾比湖湿地土壤盐分估算
梁静,丁建丽,王敬哲
新疆大学资源与环境科学学院,新疆大学资源与环境科学学院,新疆大学资源与环境科学学院
摘要:
土壤盐渍化是农业生产中最关键的生态问题之一,是一种降低土壤质量,严重影响作物产出的缓慢土壤退化过程。因此,土壤盐分的监测及评估对干旱区的土地管理与恢复至关重要。选取艾比湖湿地为研究区,基于2016年干湿两季 (5月/9月) 的Landsat8 OLI遥感影像,147个土壤实测样点的电导率 (Electrical Conductivity, EC) 数据及其对应的室内反射光谱数据,通过光谱重采样技术,建立该研究区土壤EC的偏最小二乘 (partial least-squares regression, PLSR) 定量估算模型,并尝试对干湿两季的土壤盐渍化状况进行对比分析。结果表明:(1) 艾比湖湿地土壤盐渍化较为严重,湿季土壤EC(23.90 mS•cm-1)高于干季(1.62 mS•cm-1);(2)基于重采样处理后的光谱数据及13个光谱指数所建立的PLSR模型具有较好的精度(R2 = 0.91,RMSE = 6.48 mS•cm-1,RPD = 2.45),说明该模型可以有效地对区域尺度的土壤EC进行定量估算。(3) 从2016年5月至9月,艾比湖湿地的中度和重度盐渍土面积增加,轻度盐渍土和盐土面积减少。本研究将两种不同分辨率的数据进行联合建模,即提升了传统光学遥感影像模型的精度,又将高光谱数据扩展至像元尺度上,为土壤盐分信息的遥感提取提供了科学参考。
关键词:  盐渍化  艾比湖湿地  Landsat  光谱指数  偏最小二乘
DOI:10.11766/trxb201805070182
分类号:
基金项目:国家自然科学基金(41771470, U1603241, 41661046)
Estimation and Mapping of Soil Salinity in the Ebinur Lake Wetland Based on Vis-NIR Reflectance and Landsat 8 OLI Data
Liang Jing,Ding Jianli and Wang jingzhe
Xinjiang University, College of Resource and Environment Sciences,Xinjiang University, College of Resource and Environment Sciences,College of Resource and Environment Sciences
Abstract:
【Objective】Soil salinization, one of the most critical ecological problems in agriculture, is a progressive soil degradation process that reduces soil quality and hence crop yield and agricultural production. Therefore, it is necessary to monitor soil salinity for prevention and mitigation of land degradation in the arid regions. Producers and decision-makers also require updated accurate soil salinity maps of agronomically and environmentally relevant regions. 【Method】A total of 147 soil samples were collected from the Ebinur Lake wetland, Xinjiang Uyghur Autonomous Region of China, during the rainy season (May) and dry season (September) in 2017 for analysis of electrical conductivity (EC) when prepared into suspensions, 1:5 in soil and distilled water ratio, and for acquisition of Vis-NIR (visible-near infrared) reflectance spectra in the laboratory. Spectra were resampled in line with Landsat8 OLI sensor’s resolution, i.e., band 1 (Coastal) 433~453 nm, band 2 (Blue) 450~515 nm, band 3 (Green) 525~600 nm, band 4 (Red) 630~680 nm, band 5 (Red) 845~885 nm, band 6 (SWIR 1) 1 560~1 660 nm, and band 6 (SWIR 7) 2 100~2 300 nm. Furthermore, NDSI (Normalized Difference Vegetation Index), SI (Salinity Index), SI1 (Salinity Index 1), SI2 (Salinity Index 2), SI3 (Salinity Index 3), S1 (Salinity Index, S1)、S2 (Salinity Index), S3 (Salinity Index), S5 (Salinity Index), S6 (Salinity Index), Int1 (Intensity Index 1), Int2 (Intensity index 2) and COSRI (Combined Spectral Response Index) were also calculated in this study. A quantitative estimating model was constructed based on partial least squares regression (PLSR), and evaluated in light of its root mean square error (RMSE), determination coefficient (R2) and ratio of performance to deviation (RPD). 【Result】Results show that the surface soil of the Ebinur lake wetland was strongly salt-affected, with soil salinity during the rainy season (23.90 mS•cm-1) being much higher than that during the dry season (11.62 mS•cm-1). 2) The PLSR model based on resampled spectral data and 13 spectral indexes performed quite ideally in predicting soil EC in the study area, with quite high accuracy (R2 = 0.91, RMSE = 6.48 mS•cm-1, and RPD = 2.45), which indicates that the model constructed in the study could be used to predict quantitatively EC in the Ebinur Lake wetland; and 3), the areas of slightly saline soil and saline soil decreased, while those of moderately and heavily salinized soils increased during the study period (from May to September). 【Conclusion】In the present study, the model, established by combining two types of remote sensing data different in resolution, has obviously improved the traditional optical remote sensing (Landsat8 OLI) model in precison, as well as elevated the Vis-NIR spectral data to the pixel scale, thus providing certain scientific reference for remote sensing extraction of soil salinity information. The performance of Vis-NIR-based prediction of soil salinity might be affected by adsorption capacity of soluble salts in these electromagnetic ranges being lower than that of: water, soil iron, organic matter, certain types of clay minerals, and some other soil components. To further improve the prediction accuracy, further efforts should be done to define the most dominated factor affecting spectral reflectance of soils different in salinity degree.
Key words:  Soil salinization  Ebinur Lake wetland  Landsat  Spectral Index  PLSR