基于无人机高光谱数据的甘蔗糖分估算模型研究
Research on Sugarcane Sugar Estimation Model based on UAV Hyperspectral Data
陈燕丽 1黄璐 1杨邵锷 2孙明 1丁美花 1黄立宁 3马瑞升 1梁驰 3杨鑫 1陈诚1
作者信息
- 1. 广西壮族自治区气象科学研究所,广西南宁 530022
- 2. 广西壮族自治区农业科学院农业科技信息研究所,广西南宁 530007
- 3. 广西糖业集团良圻制糖有限公司,广西南宁 530300
- 折叠
摘要
为探究无人机高光谱遥感直接估算甘蔗糖分的能力,本研究以广西甘蔗主栽品种桂糖 42 号(GT42)和桂柳 05136(GL05136)为研究对象,布设了多种氮肥水平试验,利用无人机航拍获取新植蔗茎伸长期和工艺成熟期冠层高光谱数据,结合大田测定的糖分数据构建了甘蔗糖分估算模型.结果表明:①从茎伸长期至工艺成熟期,GT42 和GL05136 不同氮肥处理的糖分均呈上升变化,而同一品种不同处理的糖分差异变小.②不同品种甘蔗在不同生育期的糖分敏感高光谱波段有差异,红边波段和蓝红光波段分别对茎伸长期和工艺成熟期糖分变化较敏感,491、549、622、718 和 834 nm可以分别作为红光、绿光、蓝光、红边和近红外波段用于糖分植被指数建模的特征波段.③高光谱植被指数对GT42 茎伸长期糖分估算精度明显高于 GL05136,但工艺成熟期相反.PSRI(Plant senescence reflectance index)和MCARI(Modified chlorophyll absorption in reflectance index)分别是GT42 茎伸长期(R2=0.987)和工艺成熟期(R2=0.587)的较优选择,对于GL05136 茎伸长期(R2=0.850)和工艺成熟期(R2=0.231)的糖分拟合,优选指数是VARI(Visible light atmospherically resistant vegetation index).
Abstract
In order to explore the ability of UAV hyperspectral remote sensing to directly estimate sugarcane sugar content,two main sugarcane varieties in Guangxi,Guitang 42(GT42)and Guiliu 05136(GL05136),were used in field experiment with many nitrogen fertilizer levels.Then,the canopy hyperspectral reflectance of newly planted cane at elongation stage and process maturity stage was obtained by UAV.Combined with sugar content data measured in the field,sugarcane sugar content estimation models were constructed.The results showed that:①From stem elongation stage to process maturity stage,the sugar content of GT42 and GL05136 under different nitrogen fertilizer treatments increased,while the difference in sugar content between different treatments of the same variety decreased.②The sugar-sensitive hyperspectral bands were different for different varieties of sugarcane at different growth stages.The red-edge band and the blue-red band were more sensitive to the changes of sucrose content at the elongation stage and the technical maturity stage,respectively.491,549,622,718,and 834 nm can be used as the characteristic bands of red,green,blue,red and near-infrared spectra respectively to model the sugar vegetation index.③At stem elongation stage,the estimation accuracy of all hyperspectral vegetation indices for GT42 was significantly higher than that of GL05136,but the opposite was true at process maturity stage.PSRI and MCARI were the optimal sugar fitting index of GT42 at stem elongation stage(R2=0.987)and process maturity stage(R2=0.587).For GL05136 at stem elongation stage(R2=0.850)and process maturity stage(R2=0.231),the preferred index was VARI.
关键词
无人机遥感/高光谱/甘蔗糖分/植被指数/模拟模型Key words
UAV remote sensing/Hyperspectral/Sugar content of sugarcane/Vegetation index/Estimating models引用本文复制引用
基金项目
国家自然科学基金青年科学基金(42105175)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据