
基于无人机MiniSAR和多光谱遥感数据的冬小麦土壤墒情反演
张成才, 祝星星, 姜明梁, 王蕊, 侯佳彤
基于无人机MiniSAR和多光谱遥感数据的冬小麦土壤墒情反演
Soil Moisture Inversion of Winter Wheat Based on UAV Minisar and Multi-spectral Remote Sensing Data
为了去除农作物对雷达散射信号的影响,探究不同极化方式土壤后向散射系数与土壤墒情的响应关系,实现对冬小麦农田土壤墒情的精准监测,基于无人机MiniSAR多极化数据和多光谱数据,提出联合改进水云模型与BP神经网络反演土壤墒情的方法。首先利用植被覆盖度对水云模型进行改进,提取不同极化方式下的土壤后向散射系数,通过设置不同极化方式、极化差、极化比数据与归一化植被指数(NDVI)的多种组合模式,输入BP神经网络,构建冬小麦土壤墒情反演模型,并以河南省鹤壁市浚县中部的冬小麦种植区为试验区分析模型的预测效果。结果表明:相比于冬小麦土壤墒情线性回归模型,基于BP神经网络的土壤墒情反演模型精度更高,其中由改进水云模型计算得到的VV极化下的土壤后向散射系数、HH极化下的土壤后向散射系数以及两者的极化差、极化比组合输入BP神经网络得到的反演结果精度最高,R 2达到0.767,MAE为0.013 6 cm3/cm3,RMSE为0.017 6 cm3/cm3。表明联合改进水云模型与BP神经网络的冬小麦土壤墒情反演模型具有较高的反演精度,为准确监测冬小麦土壤墒情提供了一种新思路。
In order to reduce the influence of crops on radar scattering signals, explore the response relationship between radar backscattering coefficients and soil moisture with different polarization methods, and realize precise monitoring of soil moisture in winter wheat farmland, both mini synthetic aperture radar (MiniSAR) polarization data and multispectral data acquired via UAVs were used to estimate soil moisture. The impact of vegetation canopy on microwave signals was reduced by integrating vegetation cover fraction into Water Cloud Model (WCM). The adjusted WCM was subsequently employed to calculate the backscattering coefficients of soil under different polarizations. Subsequently, using the soil backscattering coefficient under different polarizations, polarization differences, polarization ratios, alongside Normalized Difference Vegetation Index (NDVI) values, as input variables, soil moisture is estimated using both a linear regression model and a BP neural network model. The results showed that compared with the linear regression model of soil moisture in winter wheat, the soil moisture inversion model based on BP neural network had higher accuracy, and the prediction effect of the model was analyzed by taking the winter wheat planting area in the central part of Jun County, Hebi City, Henan Province as the experimental area. By comparing all soil moisture regression models, it can be concluded that the model achieves its highest accuracy when using the backscattering coefficients derived from the improved WCM under VV and HH polarizations, along with their polarization difference and ratio, as inputs. This configuration yields a high coefficient of determination (R 2) of 0.767, indicating a strong correlation between predicted and actual soil moisture values. Additionally, the model demonstrates a mean absolute error of 0.013 6 cm³/cm³ and a root mean squared error of 0.017 6 cm³/cm³, highlighting its precision in estimating soil moisture content. This finding shows that the combination of the adjusted WCM and the BP neural network for winter wheat soil moisture inversion demonstrates high inversion accuracy, which can provide a novel approach for accurately monitoring winter wheat soil moisture.
土壤墒情 / 水云模型 / BP神经网络 / 后向散射系数 / MiniSAR数据 {{custom_keyword}} /
soil moisture / water-cloud model / back propagation neural network / backscattering coefficient / MiniSAR data {{custom_keyword}} /
表1 BP神经网络输入参数组合方案Tab.1 Input parameter combination scheme of BP neural network |
输入方案 | 输入参数 | 输出参数 |
---|---|---|
单个极化 | NDVI、 | 土壤体积含水率 |
NDVI、 | 土壤体积含水率 | |
NDVI、 | 土壤体积含水率 | |
不同极化方式组合 | NDVI、 | 土壤体积含水率 |
NDVI、 | 土壤体积含水率 | |
NDVI、 | 土壤体积含水率 | |
NDVI、 | 土壤体积含水率 | |
不同极化方式组合+极化差、极化比 | NDVI、 | 土壤体积含水率 |
NDVI、 | 土壤体积含水率 | |
NDVI、 | 土壤体积含水率 |
表2 基于BP神经网络的各组合反演精度评价Tab.2 Evaluation table of inversion accuracy of each combination based on BP neural network |
输入组合 | R 2 | 平均绝对误差MAE/(cm3·cm-3) | 均方根误差RMSE/(cm3·cm-3) |
---|---|---|---|
NDVI、 | 0.711 | 0.016 4 | 0.019 6 |
NDVI、 | 0.464 | 0.022 5 | 0.026 7 |
NDVI、 | 0.610 | 0.020 3 | 0.022 8 |
NDVI、 | 0.567 | 0.019 4 | 0.024 0 |
NDVI、 | 0.714 | 0.017 5 | 0.019 5 |
NDVI、 | 0.603 | 0.017 6 | 0.023 0 |
NDVI、 | 0.719 | 0.016 7 | 0.019 3 |
NDVI、 | 0.736 | 0.013 1 | 0.018 7 |
NDVI、 | 0.754 | 0.015 2 | 0.018 1 |
NDVI、 | 0.767 | 0.013 6 | 0.017 6 |
1 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
2 |
方西瑶, 蒋玲梅, 崔慧珍. 基于Sentinel-1雷达数据的青藏高原地区土壤水分反演研究[J]. 遥感技术与应用, 2022,37(6):1 447-1 459.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
3 |
李 艳, 张成才, 恒卫东. 基于深度学习的多源遥感反演麦田土壤墒情研究[J]. 节水灌溉, 2023(2):57-64.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
4 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
5 |
孔金玲, 李菁菁, 甄珮珮, 等. 微波与光学遥感协同反演旱区地表土壤水分研究[J]. 地球信息科学学报, 2016,18(6):857-863.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
6 |
曾旭婧, 邢艳秋, 单 炜, 等. 基于Sentinel-1A与Landsat 8数据的北黑高速沿线地表土壤水分遥感反演方法研究[J]. 中国生态农业学报, 2017,25(1):118-126.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
7 |
赵建辉, 张 蓓, 李 宁, 等. 基于Sentinel-1/2遥感数据的冬小麦覆盖地表土壤水分协同反演[J]. 电子与信息学报, 2021,43(3):692-699.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
8 |
阙艳红, 吴 苏, 姜明梁, 等. 融合多源遥感数据的夏玉米土壤水分反演方法对比研究[J]. 节水灌溉, 2024(3):91-98.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
9 |
李 艳, 张成才, 恒卫冬, 等. 基于多源遥感数据反演土壤墒情方法研究[J]. 节水灌溉, 2020(8): 76-81.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
10 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
11 |
徐洪刚, 陈 震, 程 千, 等. 无人机热红外反演土壤含水率的方法[J]. 排灌机械工程学报, 2022,40(11):1 181-1 188.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
12 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
13 |
冯珊珊, 梁雪映, 樊风雷, 等. 基于无人机多光谱数据的农田土壤水分遥感监测[J]. 华南师范大学学报(自然科学版), 2020,52(6):74-81.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
14 |
郭 文, 马梦梦, 孙培彦. 基于Sentinel-1A数据和BP神经网络的裸露地表土壤含水量反演研究[J]. 中国农村水利水电, 2023(1):89-94.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
15 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
16 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
17 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
18 |
王树果, 马春锋, 赵泽斌, 等. 基于Sentinel-1及Landsat 8数据的黑河中游农田土壤水分估算[J]. 遥感技术与应用, 2020,35(1):13-22+47.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
19 |
李 艳, 张成才, 罗蔚然. 利用改进的水云模型反演夏玉米拔节期土壤墒情方法研究[J]. 水利水电技术, 2019,50(3):212-218.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
20 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
21 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
22 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
23 |
张成才, 王 蕊, 侯佳彤, 等. 基于特征变量筛选的无人机多光谱遥感土壤含水量反演[J]. 中国农村水利水电, 2024(5):147-154.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
24 |
聂素云, 杨 彬, 夏 微, 等. 冬小麦多时期冠层含水量遗传优化遥感反演[J]. 华东师范大学学报(自然科学版), 2023(3): 71-81.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
25 |
尹承深, 刘全明, 王春娟, 等. 基于地面光谱测量和主动微波遥感的反演土壤水分研究[J]. 西南农业学报, 2022,35(11):2 595-2 602.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
26 |
李 虎, 钟 韵, 冯雅婷, 等. 无人机遥感的多植被指数土壤水分反演模型[J]. 光谱学与光谱分析, 2024,44(1):207-214.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
27 |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
28 |
李伯祥, 陈晓勇, 徐雯婷. 基于水云模型的Sentinel-1A双极化反演植被覆盖区土壤水分[J]. 水土保持研究, 2019,26(5):39-44.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
29 |
鲍艳松, 刘良云, 王纪华, 等. 利用ASAR图像监测土壤含水量和小麦覆盖度[J]. 遥感学报, 2006,10(2):263-271.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
30 |
吴善玉, 鲍艳松, 李叶飞, 等. 基于神经网络算法的Sentinel-1和Sentinel-2遥感数据联合反演土壤湿度研究[J]. 大气科学学报, 2021,44(4):636-644.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
31 |
谢凯鑫, 张婷婷, 邵 芸, 等. 基于Radarsat-2全极化数据的高原牧草覆盖地表土壤水分反演[J]. 遥感技术与应用, 2016,31(1):134-142.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
32 |
孙时雨, 宋承运, 周 露. 基于优化BP神经网络ESA CCI土壤水分重建方法研究[J]. 无线电工程, 2023,53(11):2 507-2 514.
{{custom_citation.content}}
{{custom_citation.annotation}}
|
{{custom_ref.label}} |
{{custom_citation.content}}
{{custom_citation.annotation}}
|
/
〈 |
|
〉 |