植株密度是影响水稻生长发育和产量的重要因素。由于种植方式影响水稻田间分布特征，进而引起密度的差异，所以在水稻植株密度反演中，需要考虑种植方式的影响。合成孔径雷达(SAR)因其全天时、全天候、穿透性等优势，已被证明是水稻监测的重要手段之一。极化SAR不仅具有传统SAR的优势，还对目标结构、分布特征等非常敏感，在水稻植株密度反演中具有更大的应用潜力。基于RADARSAT-2全极化SAR数据，充分挖掘极化信息，并考虑插秧、撒播种植方式，利用弹性网络模型实现水稻植株密度反演。结果表明，极化SAR在水稻植株密度反演中具有较好的效果，插秧和撒播田植株密度的RMSE分别为25和39株/m2；而且与不考虑种植方式的反演结果相比，精度提高30%以上。
徐静璇
,
李坤
,
秦怡
,
张渤琳
,
张风丽
. 考虑种植方式的极化SAR水稻植株密度反演方法[J]. 中国科学院大学学报, 2025
, 42(2)
: 167
-175
.
DOI: 10.7523/j.ucas.2023.032
Plant density is an important factor affecting rice growth and yield. Because planting methods affect the distribution characteristics of rice in the field, and then cause the difference in density, it is necessary to consider the influence of planting methods in the inversion of rice plant density. Synthetic aperture radar (SAR) has been proved to be one of the important means of rice monitoring due to its advantages of all-day, all-weather and penetrability. Polarimetric SAR not only has the advantages of traditional SAR, but also is very sensitive to target structure and distribution characteristics. It has greater application potential in rice plant density inversion. Based on RADARSAT-2 full polarimetric SAR data, this study fully excavates the polarimetric information, considers the planting methods of transplanting and sowing, and uses the elastic network model to realize the inversion of rice plant density. The results show that polarimetric SAR has a good effect on the inversion of rice plant density. The RMSE of plant density in transplanting and sowing fields is 25 and 39 plants/m2 respectively. Compared with the inversion results without considering planting methods, the accuracy is improved by more than 30%.
[1] Karthikeyan L, Pan M, Wanders N, et al. Four decades of microwave satellite soil moisture observations: Part 1. A review of retrieval algorithms[J]. Advances in Water Resources, 2017, 109: 106-120. DOI: 10.1016/j.advwatres.2017.09.006.
[2] 化国强, 李晨, 杨沈斌, 等. 利用Radarsat-2数据基于比值检测的水稻制图[J]. 江苏农业学报, 2012, 28(6): 1451-1458.
[3] Bouvet A, Le Toan T. Use of ENVISAT/ASAR wide-swath data for timely rice fields mapping in the Mekong River Delta[J]. Remote Sensing of Environment, 2011, 115(4): 1090-1101. DOI: 10.1016/j.rse.2010.12.014.
[4] 赵路生. 基于ASAR数据的水稻制图及生物量参数反演[D]. 成都: 电子科技大学, 2013.
[5] 董彦芳, 庞勇, 孙国清, 等. ENVISAT ASAR数据用于水稻监测和参数反演[J]. 武汉大学学报(信息科学版), 2006, 31(2): 124-127. DOI: 10.3321/j.issn:1671-8860.2006.02.008.
[6] Yang Y, Lin Y. Crop detection and density estimation combing LiDAR points cloud with remote sensing image[C]//Proc SPIE 7494, MIPPR 2009: Multispectral Image Acquisition and Processing. October 30, 2009, Yichang, China. SPIE. 2009: 494-500. DOI: 10.1117/12.834012.
[7] 单捷, 孙玲, 王志明, 等. GF-1影像遥感监测指标与冬小麦长势参数的关系[J]. 江苏农业学报, 2019, 35(6): 1323-1333. DOI: 10.3969/j.issn.1000-4440.2019.06.008.
[8] 邹楠. 基于玉米冠层结构特点的遥感监测模型研究[D]. 新疆维吾尔自治区石河子: 石河子大学, 2017.
[9] 张远. 微波遥感水稻种植面积提取、生物量反演与稻田甲烷排放模拟 [D]. 杭州: 浙江大学, 2008.
[10] Inoue Y, Sakaiya E, Wang C Z. Capability of C-band backscattering coefficients from high-resolution satellite SAR sensors to assess biophysical variables in paddy rice[J]. Remote Sensing of Environment, 2014, 140: 257-266. DOI: 10.1016/j.rse.2013.09.001.
[11] Jia M Q, Tong L, Zhang Y Z, et al. Rice biomass estimation using radar backscattering data at S-band[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2014, 7(2): 469-479. DOI: 10.1109/JSTARS.2013.2282641.
[12] 贾明权. 水稻微波散射特性研究及参数反演[D]. 成都: 电子科技大学, 2013.
[13] Zhang Y, Liu X H, Su S L, et al. Retrieving canopy height and density of paddy rice from Radarsat-2 images with a canopy scattering model[J]. International Journal of Applied Earth Observation and Geoinformation, 2014, 28: 170-180. DOI: 10.1016/j.jag.2013.12.005.
[14] 景卓鑫, 张远, 王珂靖, 等. 基于Radarsat-2雷达数据的水稻参数反演[J]. 江苏农业科学, 2013, 41(11): 61-65. DOI: 10.15889/j.issn.1002-1302.2013.11.066.
[15] 邵芸, 郭华东, 范湘涛, 等. 水稻时域散射特征分析及其应用研究[J]. 遥感学报, 2001, 5(5): 340-345. DOI: 10.3321/j.issn: 1007-4619.2001.05.004.
[16] Gu S H, Kelly B, Xiu D C. Empirical asset pricing via machine learning[J]. The Review of Financial Studies, 2020, 33(5): 2223-2273. DOI: 10.1093/rfs/hhaa009.
[17] Jiang F W, Tang G H, Zhou G F. Firm Characteristics and Chinese Stocks[J]. Journal of Management Science and Engineering, 2018, 3(4): 259-283. DOI: 10.3724/SP.J.1383.304014.
[18] 李斌, 邵新月, 李玥阳. 机器学习驱动的基本面量化投资研究[J]. 中国工业经济, 2019(8): 61-79. DOI: 10.19581/j.cnki.ciejournal.2019.08.004.
[19] Zou H, Hastie T. Regularization and variable selection via the elastic net[J]. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 2005, 67(2): 301-320. DOI: 10.1111/j.1467-9868.2005.00503.x.
[20] 周亚. 基于XGBoost与弹性网络回归的集成模型对上证指数日极差的预测分析[J]. 中国物价, 2021(5): 68-71.
[21] 陈洪芳, 孙梦阳, 高毅, 等. 利用弹性网络算法求解大型三坐标测量机几何误差的方法[J]. 中国机械工程, 2022, 33(9): 1077-1083. DOI: 10.3969/j.issn.1004-132X.2022.09.009.
[22] 卢颖. 广义线性模型基于Elastic Net的变量选择方法研究[D]. 北京: 北京交通大学, 2011.
[23] 李光早, 王士同. 基于稀疏表示和弹性网络的人脸识别[J]. 计算机应用, 2017, 37(3): 901-905. DOI: 10.11772/j.issn.1001-9081.2017.03.901.
[24] 黄登香. Elastic Net方法在几类模型变量选择中的应用[D]. 南宁: 广西大学, 2014.
[25] 姚明煌. 随机森林及其在遥感图像分类中的应用[D]. 福建泉州: 华侨大学, 2014.
[26] 孙永平, 王立峰, 张震伟, 等. 基于随机森林回归的火电机组供电煤耗遗传优化模型[J]. 信息通信技术与政策, 2021, 47(3): 76-82. DOI: 10.12267/j.issn.2096-5931.2021.03.013.
[27] Breiman L. Random forests [J]. Machine Learning, 2001, 45(1): 5-32. DOI: 10.1023/A:1010933404324.
[28] 王晓燕, 李美洲. 浅谈等级相关系数与斯皮尔曼等级相关系数[J]. 广东轻工职业技术学院学报, 2006, 5(4): 26-27. DOI: 10.3969/j.issn.1672-1950.2006.04.008.
[29] Hoerl A E, Kennard R W. Ridge regression: biased estimation for nonorthogonal problems[J]. Technometrics, 1970, 12(1): 55-67. DOI: 10.1080/00401706.1970.10488634.
[30] Tibshirani R. Regression shrinkage and selection via the lasso[J]. Journal of the Royal Statistical Society Series B: Statistical Methodology, 1996, 58(1): 267-288. DOI: 10.1111/j.2517-6161.1996.tb02080.x.
