应用主动传感器GreenSeeker估测大豆籽粒产量

引用本文

张宁, 齐波, 赵晋铭, 张小燕, 王素阁, 赵团结, 盖钧镒. 应用主动传感器GreenSeeker估测大豆籽粒产量. 作物学报, 40(4): 657-666
ZHANG Ning, QI Bo, ZHAO Jin-Ming, ZHANG Xiao-Yan, WANG Su-Ge, ZHAO Tuan-Jie, GAI Jun-Yi. Prediction for Soybean Grain Yield Using Active Sensor GreenSeeker. 作物学报, 40(4): 657-666 复制到剪切板

Permissions

作物学报编辑部

应用主动传感器GreenSeeker估测大豆籽粒产量

张宁^1,^*, 齐波^1,^*, 赵晋铭¹, 张小燕², 王素阁², 赵团结¹, 盖钧镒^1,^*

¹南京农业大学大豆研究所 / 国家大豆改良中心 / 农业部大豆生物学与遗传育种重点实验室(综合) / 作物遗传与种质创新国家重点实验室, 江苏南京 210095

²圣丰种业院士工作站, 山东济宁 272000

^*通讯作者(Corresponding author): 盖钧镒, E-mail:sri@njau.edu.cn

收稿日期:2014-01-12

基金:本研究由国家重点基础研究发展计划(973计划)项目(2009CB1184), 国家公益性行业(农业)科研专项经营项目(200803060, 201203026-4)和江苏省优势学科建设工程专项和国家重点实验室自主课题资助。

摘要

R²)可达到0.66; 用此模型对NJRIKY估产, 与实际产量的符合度可达0.59。所建模型具有一定的预测效果, 在规模化育种中可用于育种中期无重复试验的产量预测和初步选择。]]>

关键词: 大豆; 产量; 归一化植被指数; 主动传感器; 遥感估测

Prediction for Soybean Grain Yield Using Active Sensor GreenSeeker

ZHANG Ning^1,^**, QI Bo^1,^**, ZHAO Jin-Ming¹, ZHANG Xiao-Yan², WANG Su-Ge², ZHAO Tuan-Jie¹, GAI Jun-Yi^1,^*

¹Soybean Research Institute / National Center for Soybean Improvement / Key Laboratory for Biology and Genetic Improvement of Soybean (General), Ministry of Agriculture / National Key Laboratory of Crop Genetics and Germplasm Enhancement, Nanjing Agricultural University, Nanjing 210095, China

²Shofine Academician Workstation, Jining 272000, China

Fund:

Abstract

R²= 0.66. Using this model to predict the NJRIKY lines, the coincidence between the measured and predicted values was 0.59. This model can be used at the middle stage of breeding programs for yield prediction of the breeding lines without replicated yield test.]]>

Keyword: Soybean; Yield; NDVI; Active sensor; Remote sensing prediction

Show Figures

大豆在粮食作物构成和油料作物生产中占有重要地位^{[ 1, 2, 3]}。大豆产量是遗传、环境及栽培措施共同作用的结果, 产量性状的选择具有不确定性。大豆育种中选择产量性状是关键, 传统的直接测产方式耗费大量的人力、物力和财力, 规模化育种中, 快速预测田间成千上万个小区的产量, 可以帮助提高育种效率。

遥感技术提供了简单、快速、无损伤地监测作物生长和产量估测的有效途径^{[ 4]}。其中, 地面传感器可以获取作物特征波段光谱反射率并构建植被指数, 已经成功地反演作物生物量、氮含量和产量等^{[ 5, 6, 7, 8, 9, 10]}。吴琼等^{[ 10]}报道了高光谱遥感对大豆冠层生长监测和产量估测有相对可行性。地面传感器按照光源类型可分为被动和主动两种, 前者依赖于太阳辐射, 易受外界天气的影响, 其应用具有一定的时间局限性, 而后者不受太阳辐射变化的影响, 因其方便、快捷、成本低而在作物生长监测和产量估测上广泛应用, 其中, 主动传感器GreenSeeker是由美国俄克拉荷马大学与Ntech公司联合开发的一种便携式光谱仪, 通过红光(656 nm)和近红外(774 nm)的反射率获得归一化植被指数(normalized difference vegetation index, NDVI)和比值植被指数(ratio vegetation index, RVI)。实际应用中常常采用植被指数作为评价作物生长状况的标准, NDVI是应用最为广泛的植被光谱指数之一^{[ 11, 12]}, 它是利用绿色植物对红光的低反射率和对近红外光的高反射率的光谱特征值计算的植被指数, 能很好地反映绿色植被的生长情况、营养状况、潜在产量与病虫害影响。

近年来, 国内外关于应用主动传感器GreenSeeker对作物生长指标和产量潜力的研究很多, 已经在小麦、玉米等作物产量估测方面取得一定进展^{[ 13, 14, 15, 16, 17]}。Erdle等^{[ 18]}研究发现主动传感器在监测作物生长参数上具有灵活、不受时间和天气条件限制的优点。Raun等^{[ 19]}利用主动传感器估测大面积种植的小麦产量, 结合2年的田间数据建立了精确度较高的冬小麦产量估测模型。Thomason等^{[ 20]}利用GreenSeeker光谱仪确定了影响玉米产量的敏感生育期, 并以此指导玉米实际生产。Inman等^{[ 21]}研究发现玉米八叶期冠层NDVI与产量呈极显著正相关, 因此可在该时期利用NDVI值估测其产量。另外, Phillips等^{[ 22]}研究表明NDVI比RVI能更好地进行小麦产量的估测。目前, 关于以主动传感器GreenSeeker估测大豆产量的研究鲜有报道。

本研究以2011—2012年黄淮海地区大豆育种品系和重组自交家系群体(NJRIKY)共1272份材料为研究对象, 采用GreenSeeker测定生育关键时期冠层NDVI, 探讨大豆产量与NDVI之间的关系, 解析大豆冠层NDVI变化的规律, 选择产量估测的最佳生育期, 确定基于冠层NDVI的产量最佳估测模型, 期望能够应用于大豆育种中期无重复试验时的产量选择, 提高效率, 缩短周期。

1 材料与方法

1.1 试验材料

选取圣丰种业黄淮海地区大豆育种品系和重组自交家系群体(NJRIKY)两类试验材料, 供试材料具有大幅度遗传差异, 以保证所建模型具有较广的适用范围。所选黄淮海地区大豆育种品系材料包括: 试验I, 2011年黄淮海地区大豆育种品系材料240份, 生育期变化在93.0~117.0 d。试验II, 2012年品系比较试验, 黄淮海地区3组育种品系材料, 其中, 1组共42份材料, 生育期变化在95.7~113.7 d; 2组共100份材料, 生育期变异范围是94.0~109.0 d; 3组共63份材料, 生育期变化在96.0~111.0 d。试验III, 2012年品系鉴定试验, 黄淮海地区两组育种材料, 其中1组共339份材料, 生育期变化在94.0~117.0 d; 2组共42份材料, 生育期变化在96.7~115.7 d。试验IV, 2012年山东省区域试验, 代表黄淮海地区大豆高产育种优系, 共有29份材料, 生育期变化在100.0~ 114.0 d。

另外, 试验V, 2012年重组自交家系群体(NJRIKY)生育期的变化在92.0~114.0 d, 该群体的母本科丰1号是我国北方大豆品种, 植株较矮小, 产量、生物量较低; 父本南农1138-2是南方生产推广品种, 植株相对较高, 产量、生物量等也较高, 与母本在地理来源、农艺性状上有较大差异。二者杂交衍生的重组自交家系群体(NJRIKY), 家系间产量、生物量、叶部性状等均有较宽的变异。

1.2 试验设计

2011—2012年在山东圣丰院士工作站进行大豆田间试验。

试验I为2011年的品系鉴定试验, 分4组, 每组60份材料。均采用随机区组设计, 重复3次。每小区4行, 行距40 cm, 小区面积1.6 m×5.0 m, 以中间2行的中间3 m计产, 实收面积2.4 m²。

试验II为2012年的品系比较试验(1组、2组和3组), 均采用重复内分组设计, 重复3次。每小区5行, 行距50 cm, 小区面积2.5 m × 5.0 m, 以中间3行的中间3 m计产, 实收面积4.5 m²。

试验III为2012年的品系鉴定试验(1组和2组), 均采用重复内分组设计, 重复3次。每小区4行, 行距50 cm, 小区面积2 m × 5 m, 以中间2行的中间 3 m计产, 实收面积3 m²。

试验IV为2012年的山东省大豆区域试验, 采用随机区组设计, 重复3次。每小区6行, 行距50 cm, 小区面积3 m × 5 m, 以中间4行计产, 实收面积10 m²。

试验V为2012年的NJRIKY群体试验, 采用格子设计, 重复3次。每小区5行, 行距50 cm, 小区面积为2.5 m × 2.0 m, 以中间3行计产, 实收面积3 m²。

1.3 数据获取

利用GreenSeeker分别在大豆苗期、开花期、结荚期和鼓粒期测定冠层NDVI。选择晴朗无风、风速较小或微风的天气, 在叶片露水消退后, 将主动传感器探头始终平行于大豆冠层, 垂直高度约60 cm, 以小区的中间行为采样点, 沿行向保持匀速行进测定大豆冠层NDVI值和RVI值。其中, 选择2012年重组自交家系试验作为验证数据。

式中, R_NIR为近红外光特征波长处(774 nm)的植被反射率; R_RED表示红光特征波长处(656 nm)的植被反射率。大豆成熟后, 收获晒干至恒重时, 称取大豆产量。

1.4 数据处理

利用Microsoft Excel 2010整理数据, 并基于R

平台(R Core Team, 2013)进行方差分析和回归分析。分析结果表明, 基于大豆冠层NDVI和RVI所建估产模型中, NDVI的效果均优于RVI, 因此本研究主要探讨利用冠层NDVI估测大豆产量。对一组数据建立回归模型时, 模型的适合程度用决定系数表示(估计值和实验值间相关系数平方值); 用一组数据的模型去预测另一组数据的结果时将估计值和实验值间相关系数平方值称为符合度。

2 结果与分析

2.1 大豆各组试验材料产量的变异

由表1可以看出, 2011年大豆产量比2012年略低, 且2011年大豆产量变幅较大。其中, 试验I-2组变幅为750.0~2666.7 kg hm^-2, 变异系数( CV)最大(0.20), 说明试验I不同大豆材料间产量变异幅度均较大。试验II为2012年品系比较试验, 试验材料均是代表黄淮海育种改良要求的不同熟期组的大豆材料, 其中II-1组大豆产量变幅为2568.9~4456.3 kg hm^-2, 变异系数( CV)为0.10, II-2组大豆产量变幅为2745.2~4020.7 kg hm^-2, 变异系数( CV)为0.07, II-3组大豆产量变幅为2542.2~4367.4 kg hm^-2, 变异系数( CV)为0.11; 试验III与试验I为不同年份的品系鉴定试验, 与2011年品系鉴定试验相比, 2012年品系鉴定试验的大豆产量水平相对较高。试验IV为2012年山东省区域试验, 参试材料均是选拔出的山东省高产高抗品系, 其产量变幅为2685.3~ 4396.0 kg hm^-2, 大豆产量均值较大(3789.1 kg hm^-2), 变异系数( CV)较小(0.12), 说明参试材料集中代表了山东省大豆高产育种的成果。试验V为2012年重组

表1 大豆产量的次数分布和变异 Table 1 Frequency distribution and variation of yield in soybean

表2 大豆冠层NDVI的分布和变异 Table 2 Frequency distribution and variation of soybean canopy NDVI

生育期 Growth stage	组限 Class limit																					变异范围 Range	均值 Mean	变异系数 CV
	<0.49	0.49-	0.51-	0.53-	0.55-	0.57-	0.59-	0.61-	0.63-	0.65-	0.67-	0.69-	0.71-	0.73-	0.75-	0.77-	0.79-	0.81-	0.83-	>0.85	∑ f
	<0.49	0.51	0.53	0.55	0.57	0.59	0.61	0.63	0.65	0.67	0.69	0.71	0.73	0.75	0.77	0.79	0.81	0.83	0.85	>0.85	∑ f
I-1
开花期FS																	6	41	13		60	0.80-0.84	0.82	0.01
结荚期PS												1			2	6	30	19	2		60	0.69-0.84	0.80	0.03
鼓粒期SFS	16	1	4	4	4	2	3	4	3	3	5	4	4	2	1						60	0.13-0.76	0.55	0.29
I-2
开花期FS															1	4	10	38	7		60	0.76-0.84	0.81	0.02
结荚期PS														2	1	29	20	7	1		60	0.74-0.83	0.79	0.02
鼓粒期SFS	24	3	4	3	1	1	4	2	0	2	2	0	2	6	3	2					59	0.16-0.78	0.53	0.28
I-3
开花期FS																1	4	33	22		60	0.78-0.84	0.83	0.01
结荚期PS												1			3	31	24	1			60	0.71-0.81	0.79	0.02
鼓粒期SFS	3		1			3	4	1	3	4	14	6	9	8	1						57	0.21-0.75	0.66	0.14
I-4
开花期FS																		3	42	15	60	0.82-0.86	0.84	0.01
结荚期PS															8	38	14				60	0.75-0.81	0.78	0.02
鼓粒期SS	4	2		4	1	3	3	3	4	2	4	4	2	12	7	3	2				60	0.42-0.80	0.67	0.15
II-1
苗期 SS						1		1		3	2	4	7	10	10	3	1				42	0.58-0.79	0.73	0.06
开花期FS																2	13	27			42	0.79-0.83	0.81	0.01
结荚期PS																8	30	4			42	0.77-0.82	0.80	0.01
鼓粒期SFS	2	2			1		1	1	1	5	6	3	12	5	2		1				42	0.42-0.81	0.68	0.13
II-2
苗期 SS					1		2	2	6	8	14	15	25	15	8	4					100	0.57-0.79	0.70	0.06
开花期FS													1	1	3	49	46				100	0.73-0.81	0.79	0.01
结荚期PS																15	71	14			100	0.77-0.82	0.80	0.01
鼓粒期SFS					1	1			1	3	7	13	17	17	15	21	3				99	0.56-0.81	0.73	0.06
II-3
苗期 SS											1	1	8	24	18	11					63	0.69-0.79	0.75	0.02
开花期FS																2	48	13			63	0.79-0.82	0.80	0.01
结荚期PS															2	30	29	1			62	0.76-0.81	0.79	0.01
鼓粒期SFS	1					1			1	1	1	3	3	10	24	18					63	0.40-0.78	0.74	0.08
III-1
苗期 SS	0	1	2	4	4	10	13	24	39	55	54	52	44	19	16	2					339	0.51-0.78	0.67	0.07
开花期FS														2	12	87	212	26			339	0.73-0.82	0.79	0.02
结荚期PS																52	240	47			339	0.77-0.82	0.80	0.01
鼓粒期SFS										1	2	2	7	24	48	116	121	18			339	0.67-0.82	0.78	0.03
III-2
开花期FS												1		1	5	10	23	2			42	0.69-0.81	0.78	0.03
结荚期PS														1	2	21	18				42	0.75-0.81	0.79	0.01
鼓粒期SFS														1	16	20	5				42	0.74-0.81	0.78	0.02
IV
苗期 SS									1			4	3	8	11	2					29	0.64-0.78	0.74	0.04
开花期FS																	5	23	1		29	0.80-0.83	0.82	0.01
结荚期PS															2	25	2				29	0.75-0.80	0.78	0.01
鼓粒期SFS											1	5	5	14	1	3					29	0.67-0.78	0.73	0.04
V
苗期 SS	389	14	10	2	1	1															417	0.27-0.58	0.41	0.14
开花期FS													2	1	6	79	288	41			417	0.73-0.82	0.80	0.01
结荚期PS											1		1	2	8	48	178	162	17		417	0.69-0.84	0.81	0.02
鼓粒期SFS						1	2	3	1	7	7	9	8	25	70	179	101	4			417	0.58-0.81	0.77	0.05

FS: flowering stage; PS: podding stage; SFS: seed-filling stage; SS: seeding stage.

表2 大豆冠层NDVI的分布和变异 Table 2 Frequency distribution and variation of soybean canopy NDVI

自交家系群体试验, 大豆产量变幅为1730.0~ 4613.3 kg hm^-2, 变异系数( CV)为0.14, 说明家系间的材料有比较宽的变异。

通过对2011年和2012年的5个试验的产量进行方差分析, 11组不同试验大豆材料间的产量均存在极显著差异( P< 0.01), 说明不同试验内的材料间均存在显著差异, 适合进行估产建模的研究。

2.2 不同生育期大豆冠层NDVI的变化

冠层NDVI与大豆植株生长状况密切相关, 由表2可以看出, 不同大豆材料苗期NDVI偏低, 与开花期和结荚期相比, 苗期冠层NDVI变异幅度较大。其中, 试验V为NJRIKY群体试验, 不同材料间生长差异变化较大, 冠层NDVI变异幅度最大(0.27~ 0.58), 变异系数( CV)为0.14; 试验II-3组冠层NDVI变幅最小(0.69~0.79), 变异系数( CV)为0.02。开花期大豆生物量显著增加, 冠层NDVI明显增大, 不同试验的冠层NDVI平均值约为0.8, 最大的NDVI可达0.84, 但该时期冠层NDVI变异幅度较小。其中, 试验III-2组冠层NDVI变幅(0.69~0.81)最大, 变异系数( CV)相对最大, 但仅有0.03。从开花期到结荚期植株群体结构变化不大, 与开花期相比, 大豆结荚期冠层NDVI变异不大, 不同试验的冠层NDVI也维持在较高水平, 结荚期NDVI变异幅度也相对较小。随着生育期的推移, 大豆鼓粒期冠层NDVI变化明显。其中, 试验I-1组冠层NDVI变幅最大(0.13~0.76), 变异系数( CV)高达0.29; 试验III-2组变幅最小(0.74~0.81), 变异系数最小( CV=0.02), 说明从鼓粒期开始, 植株生长变化差异显著。不同大豆试验材料冠层NDVI随生育期的推进均呈“低-高-低”趋势变化。苗期NDVI整体偏低, 开花期与结荚期冠层NDVI维持在较高水平, 鼓粒期冠层NDVI开始下降。

2.3 基于单一生育期冠层NDVI的大豆产量估测

将单一生育期冠层NDVI作为自变量, 以大豆产量为因变量, 通过回归分析, 筛选出基于不同生育期NDVI的大豆产量估测的最佳模型(表3)。由表3可以看出, 基于不同生育期NDVI所建立的回归模型均达到显著( P<0.05)或极显著水平( P<0.01), 说明总体中存在着线性或非线性的回归关系, 其中所建模型以线性关系为主。进一步比较4个大豆试验在不同生育期所建估产模型的决定系数( R²)和标准误( SE)发现, 利用冠层NDVI所建估产模型的决定系数( R²)均随生育期的推进而不断增大, 其中在鼓粒期所建模型的决定系数( R²)最大, 明显地高于苗期、开花期和结荚期, 且该时期模型的标准误( SE)最小, 明显地低于其他生育期, 表明基于鼓粒期冠层NDVI建立的模型决定度较好, 说明大豆鼓粒期是利用主动传感器进行产量估测的最佳生育期。

综合比较不同生育期大豆冠层NDVI所建产量模型, 发现试验II-1组基于鼓粒期冠层NDVI所建模型( y= 1174.7+3395.1 x)决定系数最好, 所建模型可以估测产量的变异范围是2568.9~4456.3 kg hm^-2, 生育期变异范围是95.7~113.7 d, 该模型的预测精度较高( R²= 0.68), 且该模型的标准误较小( SE= 203.38)。

2.4 基于多生育期冠层NDVI的大豆产量估测

进一步分析多个生育期大豆冠层NDVI与产量的多元回归关系, 然后筛选出大豆产量估测最佳回归模型(表4和表5)。由表4可以看出, 基于结荚期和鼓粒期两个时期冠层NDVI建立的大豆估产模型均达到极显著水平( P< 0.01), 说明总体中存在着线性或非线性的回归关系; 同时对所建模型进行参数的显著性检验, 4个试验10组材料的决定系数为0.45~ 0.69, 其中试验II-1组基于结荚期和鼓粒期冠层NDVI所建模型( y= 6534.3-7484.4 x₂+4319.5 x₃)的决定系数( R²= 0.69)最大, 且标准误( SE= 199.71)较小, 模型的拟合效果相对较好。

根据大豆结荚期和鼓粒期2个生育期冠层NDVI构建估产模型的决定系数有所提高, 但提高幅度不大。因此, 进一步用开花期、结荚期和鼓粒期3个生育期的冠层NDVI来构建大豆估产模型。由表5可以看出, 基于开花期、结荚期和鼓粒期3个生育期冠层NDVI建立的大豆估产模型均达到极显著水平( P< 0.01), 同时对所建模型进行参数的显著性检验, 4个试验10组材料的决定系数为0.58~0.86, 比结荚期和鼓粒期2个时期的预测效果又有提高。其中, 试验III-1组所建模型 y= 2484.5-45760.4 x₁+25691.7 x₂+ 21285.4 x₃效果最好( R²= 0.86, SE= 110.89), 可以估测的产量变异范围是2261.1~4097.9 kg hm^-2, 生育期变异范围是94.0~117.0 d。

为进一步探索估产模型的普适性, 将试验材料汇总为育种品系和NJRIKY群体两大类, 探索产量变异幅度更广的大豆估产模型(表6)。由表6可以看出, 基于结荚期和鼓粒期2个生育期冠层NDVI所建估产模型最高的决定系数仅为0.57。但基于开花期、结荚期和鼓粒期3个生育期冠层NDVI建立的大豆估产模型均达到极显著水平( P< 0.01), 并且所建模型的回归参数均达到极显著水平( P< 0.01)。其

表3 大豆不同生育期冠层NDVI和产量的回归方程 Table 3 Regression equations of yield to NDVI at different growth stages in soybean

试验组 Exp. group	生育期 Growth stage	回归方程 Regression equation	模型检验 Model test	模型参数检验 Parameter test		R²	SE
试验组 Exp. group	生育期 Growth stage	回归方程 Regression equation	模型检验 Model test	x	截距 Intercept	R²	SE
I-1	开花期FS	y= -15514.1+21465.1 x	P<0.01	21465.1^**	-15514.1^**	0.31	287.46
	结荚期PS	y= -7123.3+11544.3 x	P<0.01	11544.3^**	-7123.3^**	0.47	252.35
	鼓粒期SFS	y= 1168.9+1746.4 x	P<0.01	1746.4^**	1168.9^**	0.62	217.59
I-2	开花期FS	y= -11679.4+16906.0 x	P<0.01	16906.0^**	-11679.4^**	0.36	344.38
	结荚期PS	y= -12079.4+17945.1 x	P<0.01	17945.1^**	-12079.4^**	0.45	321.58
	鼓粒期SFS	y= 970.7+2114.3 x	P<0.01	2114.3^**	970.7^**	0.55	279.49
I-3	开花期FS	y= -11258.2+16274.5 x	P<0.01	16274.5^**	-11258.2^**	0.27	320.77
	结荚期PS	y= 5.1e^7.7 ^x	P<0.01	7.7^**	5.1^**	0.42	284.95
	鼓粒期SFS	y= -498.9+4001.2 x	P<0.01	4001.2^**	-498.9^**	0.52	266.72
I-4	开花期FS	y= 21236.6+27600.3 x	P<0.01	27600.3^**	21236.6^**	0.32	323.90
	结荚期PS	y= -15619.6+22573.8 x	P<0.01	22573.8^**	-15619.6^**	0.48	282.84
	鼓粒期SFS	y= -64.8+3171.9 x	P<0.01	3171.9^**	-64.8	0.64	237.91
II-1	苗期 SS	y= 1229.3+3082.9 x	P<0.05	3082.9	1229.3^**	0.14	332.46
	开花期FS	y= -12401.8+19527.6 x	P<0.01	19527.6^**	-12401.8^**	0.29	302.85
	结荚期PS	y= -13300.2+20971.9 x	P<0.01	20971.9^**	-13300.2^**	0.49	257.19
	鼓粒期SFS	y= 1174.7+3395.1 x	P<0.01	3395.1^**	1174.7^**	0.68	203.38
II-2	苗期 SS	y= 1625.1+2624.7 x	P<0.01	2624.7^**	1625.1^**	0.20	221.99
	开花期FS	y= -5837.5+11836.4 x	P<0.01	11836.4^**	-5837.5^**	0.30	206.72
	结荚期PS	y= -10289.1+17214.3 x	P<0.01	17214.3^**	-10289.1^**	0.47	179.67
	鼓粒期SFS	y= 47.0+4675.1 x	P<0.01	4675.1^**	47.0	0.66	144.91
II-3	苗期 SS	y= -1060.3+6061.7 x	P<0.01	6061.7^**	-1060.3	0.11	349.93
	开花期FS	y= -21653.0+31288.3 x	P<0.01	31288.3^**	-21653.0^**	0.33	304.83
	结荚期PS	y= -15152.8+23621.6 x	P<0.01	23621.6^**	-15152.8^**	0.43	282.55
	鼓粒期SFS	y= 689.6e^2.2 ^x	P<0.01	689.6^**	2.2^**	0.53	258.97
III-1	苗期 SS	y= 1771.9+2243.0 x	P<0.01	2243.0^**	1771.9^**	0.15	270.33
	开花期FS	y= -6593.8+12427.0 x	P<0.01	12427.0^**	-6593.8^**	0.27	249.50
	结荚期PS	y= -12463.6+19686.4 x	P<0.01	19686.4^**	-12463.6^**	0.41	223.72
	鼓粒期SFS	y= -3854.5+9142.2 x	P<0.01	9142.2^**	-3854.5^**	0.57	192.62
III-2	开花期FS	y= 8197.0+14189.5 x	P<0.01	14189.5^**	8197.0^**	0.38	308.00
	结荚期PS	y= -14524.3+22228.5 x	P<0.01	22228.5^**	-14524.3^**	0.45	287.23
	鼓粒期SFS	y= -13594.5+21378.9 x	P<0.01	21378.9^**	-13594.5^**	0.65	229.00
IV	苗期 SS	y= -867.6+6304.7 x	P<0.05	6304.7^**	-867.6	0.17	406.15
	开花期FS	y= -17864.5+26514.3 x	P<0.01	26514.3^**	-17864.5^**	0.32	368.48
	结荚期PS	y= -30275.2+43609.8x	P<0.01	43587.4^**	-30275.2^**	0.46	319.54
	鼓粒期SFS	y= -6019.4+13395.3 x	P<0.01	13395.3^**	-6019.4^**	0.62	273.49
V	苗期 SS	y= 1866.3+3408.8 x	P<0.01	3408.8^**	1866.3^**	0.18	421.68
	开花期FS	y= -12942.6+20323.5 x	P<0.01	20323.5^**	-12942.6^**	0.28	393.43
	结荚期PS	y= -10891.5+17560.2 x	P<0.01	17560.2^**	-10891.5^**	0.44	347.07
	鼓粒期SFS	y= 38832.0-108362.4 x+80560.9 x²	P<0.01	-108362.4^ 80560.9^	38832.0^**	0.58	299.00

FS: flowering stage; PS: podding stage; SFS: seed-filling stage; SS: seeding stage.

表3 大豆不同生育期冠层NDVI和产量的回归方程 Table 3 Regression equations of yield to NDVI at different growth stages in soybean

表4 大豆结荚期和鼓粒期冠层NDVI与产量的多元回归方程 Table 4 Multiple regression equations of yield to NDVI at podding and seed-filling stages in soybean

表5 大豆多生育期冠层NDVI与产量的多元回归方程 Table 5 Multiple regression equations of yield to NDVI at three growth stages in soybean

中, 对于育种品系材料, 2011—2012年两年的材料基于开花期、结荚期和鼓粒期3个生育期冠层NDVI所建的模型效果( y= )较好(图1), 该模型决定系数( R²= 0.66)较大, 所建模型可以估测产量的变异范围是750.0~4456.3 kg hm^-2, 生育期变异范围是93.0~117.0 d; 对于NJRIKY群体, 基于开花期、结荚期和鼓粒期3个生育期冠层NDVI所建的模型为 y= , 该模型决定系数( R²= 0.78)较大, 所建模型可以估测产量的变异范围是1730.0~4613.3 kg hm^-2, 生育期变异范围是92.0~114.0 d。

2.5 模型的相互验证

用2年育种品系与2012年NJRIKY群体材料相互验证3个生育期所建模型, 发现2年的大豆育种品系材料所建估产模型( y= )对NJRIKY群体材料的产量效果预测较好(图2), 产量预测值与实测值之间的符合度可达到0.59, 且标准误较小( SE= 438.13), 反之, 预测效果较差。这说明两年育种品系材料所建模型具有一定的预测性和适用性, 而NJRIKY群体材料所建模型由于其产量变异幅度小于两年育种品系材料, 具有特异性, 预测效果不佳。

综合模型比较和验证分析结果, 发现2011— 2012年两年的育种品系材料基于开花期、结荚期和鼓粒期3个生育期冠层NDVI所建的模型( y = )

表6 大豆多生育期冠层NDVI与产量的多元回归方程 Table 6 Multiple regression equations of yield to NDVI at multiple growth stages in soybean

表7 不同试验模型检验 Table 7 Model validation in different experiments

	Figure Option View Download New Window
	图1 多生育期大豆冠层NDVI与产量回归模型的投影图 x₁: 开花期冠层NDVI; x₂: 结荚期冠层NDVI; x₃: 鼓粒期冠层NDVI。Fig. 1 Projection graph of regression model between canopy NDVI combined different growth stages and yield in soybean x₁: NDVI at flowering stage; x₂: NDVI at podding stage; x₃: NDVI at seed-filling stage.

效果最好, 该模型决定系数( R²= 0.66)较大, 所建模型可以估测产量的变异范围最广(750.0~4456.3 kg hm^-2), 生育期变异范围最广(93.0~117.0 d), 模型验证结果也较好。因此, 确定该模型为采用主动传感

	Figure Option View Download New Window
	图2 大豆预测产量和实际产量的散点图Fig. 2 Scatter plot between estimated and measured yield of soybean

器GreenSeeker估测大豆产量的最佳模型, 可望用作大规模育种试验中品系产量预测的模型, 在预测的基础上收获前对品系做出预选。

3 讨论

已有研究表明冠层植被指数的变化与作物生长状况和发育时期关系紧密^{[ 11, 12, 23]}。本试验针对2011

和2012年的5个试验, 选择大豆关键生育期(苗期、开花期、结荚期和鼓粒期), 测定冠层NDVI, 并分析了大豆生育进程中冠层NDVI值的变化趋势。王磊等^{[ 24]}利用GreenSeeker获取的小麦冠层NDVI均随生育进程表现为先升高后下降的变化趋势。本研究结果表明, 大豆冠层NDVI随生育期的推进均呈现“低—高—低”变化趋势, 与前人研究具有相似的结果。

已有研究表明, 利用冠层植被指数可以间接估算作物的产量^{[ 25, 26, 27, 28, 29, 30]}。本研究将2011—2012年所有试验材料归为两大类, 分别为育种材料和遗传材料。这样以育种材料为建模数据, 增大了样本容量以及样本数据所涵盖的变异范围, 使其所建模型更具普适性和代表性。2011—2012年育种品系材料基于3个生育期(开花期、结荚期和鼓粒期)冠层NDVI所建的多元回归模型表现较好, 其模型的决定系数( R²)可达到0.66。用NJRIKY群体检验育种品系材料模型的效果, 大豆产量预测值与实测值之间的符合度可达到0.59, 说明所建模型具有一定的预测效果。

本文主旨是探寻一种可在大豆育种程序中期预测产量的间接工具, 以降低育种成本, 缩短育种时间。面对大豆育种程序中期数以千计的育种品系, 所建预测模型的决定系数如果达到0.6以上, 就可以对大豆产量进行定性分析。尽管本研究所建模型的决定系数不高( R²= 0.66), 对大豆产量性状不能实现定量化预测, 但可以对大豆产量进行定性分析, 遴选出一定比例的高产品系, 这对于大豆规模化育种来说尤为重要。大规模育种试验中期参试材料成千上万, 不设重复, 这时利用冠层植被指数间接估算家系产量, 可以帮助做初选, 田间淘汰一部分, 减少收获工作量, 提高工作效率。育种试验后期通常依靠严格的产量比较试验做决选, 但利用冠层植被指数, 可以帮助做好收获前的辅助选择和备收工作。

鉴于冠层NDVI的估产模型与作物品种、株型、密度、肥水等条件相关, 受环境影响变化明显^{[ 24]}, 基于主动传感器获取冠层NDVI对于大豆产量估测具有重要的指导意义, 但本研究采用的主动传感器由于受光谱波段的限制, 对大豆产量估测的精度有一定的影响。为了进一步提高估产模型的预测性和准确性, 我们将针对大量的田间育种品系, 利用高光谱技术提取大豆产量性状的核心波段和敏感光谱参数, 建立预测性和稳定性更好的估产模型, 为开发适合大豆估产的便携式主动传感器提供技术支持, 从而提高产量估测精度及育种选择效率。

4 结论

利用主动传感器GreenSeeker获取单一生育期冠层NDVI以估测大豆产量, 鼓粒期的估产效果明显优于苗期、开花期和结荚期。基于2011—2012年育种品系材料综合开花期、结荚期和鼓粒期3个生育期的冠层NDVI所建的多元回归模型 y= ( R²= 0.66, SE= 416.77), 可估测的产量和生育期变异范围广(750.0~4456.3 kg hm^-2, 93.0~117.0 d)。用NJRIKY群体检验所获模型, 预测值与实测值之间的符合度可达到0.59, 且标准误较小( SE= 438.13 kg hm^-2)。所建模型具有一定的预测效果, 在规模化育种中可用于育种中期无重复试验的产量预测和初步选择。

The authors have declared that no competing interests exist.

作者已声明无竞争性利益关系。The authors have declared that no competing interests exist.

参考文献

View Option

[1]	Wilcox J R. World distribution and trade of soybean. In: Soybeans: Improvement, Production and Uses, 3rd edn. Madison: American Society of Agronomy Press, 2004. pp1-13 [本文引用:1]
[2]	Clake E J, Wiseman J. Developments in plant breeding for improved nutritional quality of soybean: I. Protein and amin acids content. Agric Sci, 2000, 134: 111-124 [本文引用:1] [JCR: 0.796]
[3]	Frideman M, Brand on D L. Nutritional and health benefits of soybean proteins. Agric Food Chem, 2001, 49: 1069-1086 [本文引用:1] [JCR: 2.906]
[4]	Tennakoon S B, Murty V V N, Eiumnoh A. Estimation of cropped area and grain yield of rice using remote sensing data. Int J Remote Sens, 1992, 13: 426-439 [本文引用:1] [JCR: 1.138]
[5]	Trishchenko A P. Effects of spectral response function on surface reflectance and NDVI measured with moderate resolution satellite sensors: extension to AVHRR NOAA-17, 18 and METOP-A. Remote Sens Environ, 2009, 113: 335-341 [本文引用:1] [JCR: 5.103]
[6]	Hansen P M, Schjoerring J K. Reflectance measurement of canopy biomass and nitrogen status in wheat crops using normalized difference vegetation indices and partial least squares regression. Remote Sens Environ, 2003, 86: 542-553 [本文引用:1] [JCR: 5.103]
[7]	Aparicio N, Villegas D, Casadesus J, Araus J L, Royo C. Spectral vegetation indices as nondestructive tools for determining durum wheat yield. Agron J, 2000, 92: 83-91 [本文引用:1] [JCR: 1.518]
[8]	Soenen S A, Peddle D R, Hall R J, Coburn C A, Hall F G. Estimating aboveground forest biomass from canopy reflectance model inversion in mountainous terrain. Remote Sens Environ, 2010, 114: 1325-1337 [本文引用:1] [JCR: 5.103]
[9]	Gianquinto G, Orsini F, Fecondini M, Mezzetti M, Sanbo P, Bona S. A methodological approach for defining spectral indices for assessing tomato nitrogen status and yield. Eur J Agron, 2011, 35: 135-143 [本文引用:1] [JCR: 2.8]
[10]	吴琼, 齐波, 赵团结, 姚鑫锋, 朱艳, 盖钧镒. 高光谱遥感估测大豆冠层生长和籽粒产量的探讨. 作物学报, 2013, 39: 309-318 Wu Q, Qi B, Zhao T J, Yao X F, Zhu Y, Gai J Y. A tentative study on utilization of canopy hyperspectral reflectance to estimate canopy growth and seed yield in soybean. Acta Agron Sin, 2013, 39: 309-318 (in Chinese with English abstract) [本文引用:2] [CJCR: 1.667]
[11]	Mkhabela M S, Bullock P, Raj S, Wang S, Yang Y. Crop yield forecasting on the Canadian Prairies using MODIS NDVI data. Agric For Meteorol, 2011, 115: 385-393 [本文引用:2]
[12]	Hasegawa K, Matsuyama H, Tsuzuki H, Sweda T. Improving the estimation of leaf area index by using remotely sensed NDVI with BRDF signatures. Remote Sens Environ, 2010, 114: 514-519 [本文引用:2] [JCR: 5.103]
[13]	William R R, Gordon V J. Improving nitrogen use efficiency forcereal production. Agron J, 1999, 91: 357-363 [本文引用:1] [JCR: 1.518]
[14]	吴军华, 岳善超, 侯鹏, 孟庆峰, 崔振领, 李雯, 陈新平. 基于主动遥感的冬小麦群体动态监测. 光谱学与光谱分析, 2011, 31: 535-538 Wu J H, Yue S C, Hou P, Meng Q F, Cui Z L, Li W, Chen X P. Monitoring winter wheat population dynamics using an activecrop sensor. Spectrosc Spectr Anal, 2011, 31: 535-538 (in Chinese with English abstract) [本文引用:1]
[15]	谭昌伟, 王纪华, 朱新开, 王妍, 王君婵, 童璐, 郭文善. 基于Land sat^TM影像的冬小麦拔节期主要长势参数遥感监测. 中国农业科学, 2011, 44: 1358-1366 Tan C W, Wang J H, Zhu X K, Wang Y, Wang J C, Tong L, Guo W S. Monitoring main growth status parameters at jointing stage in winter wheat based on Land sat^TM images. Sci Agric Sin, 2011, 44: 1358-1366 (in Chinese with English abstract) [本文引用:1] [CJCR: 1.889]
[16]	冯美臣, 肖璐洁, 杨武德, 丁光伟. 基于遥感数据和气象数据的水旱地冬小麦产量估测. 农业工程学报, 2010, 26(11): 183-188 Feng M C, Xiao L J, Yang W D, Ding G W. Predicting grain yield of irrigation-land and dry-land winter wheat based on remotesensing data and meteorological data. Trans CSAE, 2010, 26(11): 183-188 (in Chinese with English abstract) [本文引用:1] [CJCR: 1.299]
[17]	Reyniers M, Vrindts E, Baerdemaeker J D. Comparison of an aerialbased system and an on the ground continuous measuring device to predict yield of winter wheat. Eur J Agron, 2006, 24: 87-94 [本文引用:1] [JCR: 2.8]
[18]	Erdle K, Mistele B, Schmidhalter U. Comparison of active and passive spectral sensors in discriminating biomass parameters and nitrogen status in wheat cultivars. Field Crops Res, 2011, 124: 74-84 [本文引用:1] [JCR: 2.474]
[19]	Raun W R, Solie J B, Stone M L, Martin K L, Freeman K W, Mullen R W, Zhang H, Schepers J S, Johnson G V. Optical sensor-based algorithm for crop nitrogen fertilization. Commun Soil Sci Plant Anal, 2005, 36, 2759-2781 [本文引用:1] [JCR: 0.42]
[20]	Thomason W E, Phillips S B, Raymond F D. Defining useful limits for spectral reflectance measures in corn. J Plant Nutr, 2007, 30: 1263-1277 [本文引用:1] [JCR: 0.526]
[21]	Inman D, Khosla R, Reich R M, Westfall D G. Active remote sensing and grain yield in irrigated maize. Precision Agric, 2007, 8: 241-252 [本文引用:1] [JCR: 1.728]
[22]	Phillips S B, Keahey D A, Warren J G, Mullins G L. Estimating winter wheat tiller density using spectral reflectance sensors for early-spring, Variable-Rate Nitrogen Applications. Agron J, 2004, 96: 591-600 [本文引用:1] [JCR: 1.518]
[23]	Aboelghar M, Arafat S, Saleh A, Naeemb S, Shirbeny M, Belal A. Retrieving leaf area index from SPOT4 satellite data. Egypt J Remote Sens Space Sci, 2010, 13: 121-127 [本文引用:1]
[24]	王磊, 白由路, 卢艳丽, 王贺, 杨俐苹. 基于GreenSeeker的冬小麦NDVI分析与产量估算. 作物学报, 2012, 38: 747-753 Wang L, Bai Y L, Lu Y L, Wang H, Yang L P. NDVI analysis and yield estimation in winter wheat based on GreenSeeker. Acta Agron Sin, 2012, 38: 747-753 (in Chinese with English abstract) [本文引用:2] [CJCR: 1.667]
[25]	Ferrio J P, Villegas D, Zarco J, Aparicio N, Araus J L, Royo C. Assessment of durum wheat yield using visible and near-infrared reflectance spectra of canopies. Field Crops Res, 2005, 94: 126-148 [本文引用:1] [JCR: 2.474]
[26]	Pinter P J, Jackson R D, Idso S B, Reginato R J. Multidate spectral reflectance as predictors of yield in water stressed wheat and barley. Int J Remote Sens, 1981, 2: 43-48 [本文引用:1] [JCR: 1.138]
[27]	Mahey R K, Singh R, Sidhu S S, Narang R S. The use of remote sensing to assess the effects of water stress on wheat. Exp Agric, 1991, 27: 423-429 [本文引用:1] [JCR: 1.621]
[28]	Lukina E V, Freeman K W, Wynn K J, Thomason W E, Mullen R W, Stone M L, Solie J B, Klatt A R, Johnson G V, Elliott R L, Raun W R. Nitrogen fertilization optimization algorithm based on in-season estimates of yield and plant nitrogen uptake. J Plant Nutr, 2001, 24: 885-898 [本文引用:1] [JCR: 0.526]
[29]	Raun W R, Johnson G V, Stone M L, Solie J B, Lukina E V, Thomason W E, Schepers J S. In-season prediction of potential grain yield in winter wheat using canopy reflectance. Agron J, 2001, 93: 131-138 [本文引用:1] [JCR: 1.518]
[30]	Hansen P M, Jorgensen J R, Thomsen A. Predicting grain yield and protein content in winter wheat and spring barley using repeated canopy reflectance measurements and partial least squares regression. J Agric Sci, 2002, 139: 307-318 [本文引用:1] [JCR: 2.041]

2004

0.0

... 大豆在粮食作物构成和油料作物生产中占有重要地位^[1,2,3] ...

2000

0.796

0.0

... 大豆在粮食作物构成和油料作物生产中占有重要地位^[1,2,3] ...

2001

2.906

0.0

... 大豆在粮食作物构成和油料作物生产中占有重要地位^[1,2,3] ...

1992

1.138

0.0

... 遥感技术提供了简单、快速、无损伤地监测作物生长和产量估测的有效途径^[4] ...

2009

5.103

0.0

... 其中, 地面传感器可以获取作物特征波段光谱反射率并构建植被指数, 已经成功地反演作物生物量、氮含量和产量等^{[5,6,7,8,9,10]} ...

2003

5.103

0.0

... 其中, 地面传感器可以获取作物特征波段光谱反射率并构建植被指数, 已经成功地反演作物生物量、氮含量和产量等^{[5,6,7,8,9,10]} ...

2000

1.518

0.0

... 其中, 地面传感器可以获取作物特征波段光谱反射率并构建植被指数, 已经成功地反演作物生物量、氮含量和产量等^{[5,6,7,8,9,10]} ...

2010

5.103

0.0

... 其中, 地面传感器可以获取作物特征波段光谱反射率并构建植被指数, 已经成功地反演作物生物量、氮含量和产量等^{[5,6,7,8,9,10]} ...

2011

2.8

0.0

. 2011, 35:135-143 DOI:10.1016/j.eja.2011.05.005

A methodological approach for defining spectral indices for assessing tomato nitrogen status and yield

Abstract

Efficient use of fertilizers is a main goal that research in agriculture shall address. Dynamic fertilization management is a way of bringing nutrients to the plant when they are crucial for its development. However, destructive measurements of crop nitrogen status are still too costly and time consuming to justify their use, and the implementation of methodologies based upon non-destructive, quick and easy to use tools for plant nutritional status monitoring appears as an appealing opportunity. In this paper vegetation indices based on crop reflectance through the use of a multispectral radiometer (MSR-87, Cropscan Inc.) are compared. Results from three years of experimentation are provided in order to evaluate (a) relations existing between several vegetation indices and both leaf chlorophyll and nitrogen content; (b) the ability of the tool to respond to different nutrient supply and/or (c) to predict yield. A methodological approach for the definition of the most reliable index is proposed and the adopted procedure allows to define the NIR/R₅₆₀, even with some degrees of incertitude, as the most suitable index for the evaluation of tomato crop nutritional status.

Highlights

? In this paper vegetation indices based on crop reflectance are compared. ? Results from three years of experimentation are provided. ? The paper assess relations with chlorophyll and N content, N supply and yield. ? A methodological approach for the definition of the most reliable index is proposed. ? NIR/R₅₆₀ appears as a reliable tool for the evaluation of tomato crop N status.

... 其中, 地面传感器可以获取作物特征波段光谱反射率并构建植被指数, 已经成功地反演作物生物量、氮含量和产量等^{[5,6,7,8,9,10]} ...

2013

0.0

1.667

. 2013, 39:309-318 DOI:10.3724/SP.J.1006.2013.00309

A tentative study on utilization of canopy hyperspectral reflectance to estimate canopy growth and seed yield in soybean

(高光谱遥感估测大豆冠层生长和籽粒产量的探讨) . (作物学报) , 2013 , 39 : 309 - 318 Wu Q , Qi B , Zhao T J , Yao X F , Zhu Y , Gai J Y . A tentative study on utilization of canopy hyperspectral reflectance to estimate canopy growth and seed yield in soybean

Wu Q

(吴琼) , (齐波) , (赵团结) , (姚鑫锋) , (朱艳) , (盖钧镒) . (高光谱遥感估测大豆冠层生长和籽粒产量的探讨) . (作物学报) , 2013 , 39 : 309 - 318 Wu Q

1 Soybean Research Institute / National Center for Soybean Improvement / Key Laboratory for Biology and Genetic Improvement of Soybean (General), Minister of Agriculture / National Key Laboratory of Crop Genetics and Germplasm Enhancement, Nanjing Agricultural University, Nanjing 210095, China; 2 National Engineering and Technology Center for Information Agriculture / Jiangsu Key Laboratory for Information Agriculture, Nanjing Agricultural University, Nanjing 210095, China

Modern plant breeding needs to monitor the growth and evaluate the yield potential for an accurate selection in a great number of breeding lines. The hyperspectral reflectance technology has been demonstrated to be potential in meeting this kind of requirement with a simple, fast and nondestructive technology. Thirty soybean cultivars from Middle and Lower Yangtze Valleys with close growing days to maturity were chosen and tested in a randomized blocks design experiment during the two consecutive years. The measurement of above-ground dry biomass (ADM) and leaf area index (LAI) was synchronized with the information collection of the canopy hyperspectral reflectance by using a portable spectroradiometer (FieldSpec Pro FR2500, Analytical Spectral Devices, Inc., Boulder, CO, USA) at three different growth stages (R2, R4, and R5) in soybean. Significant differences in ADM, LAI and plot yield among the tested cultivars were detected, which allowed a further regression analysis of the traits on the hyperspectral reflectances. There existed significant correlations between hyperspectral reflectance in the visible and infrared region and LAI, ADM, and yield, respectively. In particular, the highest correlations were observed at R4 and R5 stages. Based on a large number of spectral parameters in the literature, we selected the regression models with the best accuracy for ADM, LAI, and yield prediction. Among them, the regression model of LAI at R5 on P_Area560 and that of ADM at R4 on V_Area1450 were the best ones with their determination coefficients of 0.582 and 0.692, respectively. There was no single spectral index found for yield prediction. But the multiple regression of yield on NPH1280 at R2, V_Area1190 at R4 and NPH560 at R5 was found to provide a best yield prediction with R²=0.68. The obtained results suggested that hyperspectral remote sensing for monitoring growth status and estimating yields in soybean is feasible and potential, providing that a more accurate and stable regression model is searched based on an enlarged testing program under multiple environments. It might be especially useful and valuable for early generation yield prediction in a large-scale breeding program.

现代作物育种需要监测大量育种材料的生长并估测产量潜势, 高光谱遥感技术为此提供了简单、快捷、非损伤性测定的可能途径。选取30份大豆育成品种进行连续2年的产量比较试验, 在盛花期(R2)、盛荚期(R4)和鼓粒始期(R5)测定地上部生物量(ADM)和叶面积指数(LAI), 并利用ASD高光谱地物仪同步收集大豆冠层反射光谱信息。供试品种间ADM、LAI和产量差异显著或极显著。不同生育期可见光和近红外区域的光谱反射率与大豆ADM、LAI及产量均有显著相关, 尤其在R4和R5期相关性最高。在构建大量光谱参数的基础上, 遴选出对ADM、LAI及产量预测精度较好的回归模型。其中, R5期的P_Area560光谱参数与LAI和R4期的V_Area1450光谱参数与ADM构建的两个生长性状的监测模型效果最好, 决定系数(R²)分别为0.582和0.692。未发现单一生育期光谱参数对大豆估产的有效模型, 但综合R2期NPH1280、R4期V_Area1190以及R5期NPH560构建的产量估测模型, 决定系数(R²)达到0.68, 效果较好。本研究

... 其中, 地面传感器可以获取作物特征波段光谱反射率并构建植被指数, 已经成功地反演作物生物量、氮含量和产量等^{[5,6,7,8,9,10]} ...

... 吴琼等^[10]报道了高光谱遥感对大豆冠层生长监测和产量估测有相对可行性 ...

2011

0.0

... 实际应用中常常采用植被指数作为评价作物生长状况的标准, NDVI是应用最为广泛的植被光谱指数之一^[11,12], 它是利用绿色植物对红光的低反射率和对近红外光的高反射率的光谱特征值计算的植被指数, 能很好地反映绿色植被的生长情况、营养状况、潜在产量与病虫害影响 ...

... 3 讨论已有研究表明冠层植被指数的变化与作物生长状况和发育时期关系紧密^[11,12,23] ...

2010

5.103

0.0

... 3 讨论已有研究表明冠层植被指数的变化与作物生长状况和发育时期关系紧密^[11,12,23] ...

1999

1.518

0.0

... 近年来, 国内外关于应用主动传感器GreenSeeker对作物生长指标和产量潜力的研究很多, 已经在小麦、玉米等作物产量估测方面取得一定进展^{[13,14,15,16,17]} ...

2011

0.0

2011

0.0

1.889

2010

0.0

1.299

. 2010, 26(11):183-188

Predicting grain yield of irrigation-land and dry-land winter wheat based on remotesensing data and meteorological data

(基于遥感数据和气象数据的水旱地冬小麦产量估测) . (农业工程学报) , 2010 , 26 ( 11 ): 183 - 188 Feng M C , Xiao L J , Yang W D , Ding G W . Predicting grain yield of irrigation-land and dry-land winter wheat based on remotesensing data and meteorological data

Feng M C

(冯美臣) , (肖璐洁) , (杨武德) , (丁光伟) . (基于遥感数据和气象数据的水旱地冬小麦产量估测) . (农业工程学报) , 2010 , 26 ( 11 ): 183 - 188 Feng M C

The relationship between NDVI and grain yield was studied using remote sensing data at the different stages of winter wheat in Yuncheng region, and spectral yield model, meteorological yield model and spectrometeorological yield model were built. The results showed that the correlation between NDVI value of winter wheat in irrigation and dry-land on approximately May 8th and yield in Yuncheng region was the highest and extremely significant, so this period was the optimum period to establish remote sensing model of estimating yields in Yuncheng region. The spectral, meteorology and spectrometeorological yield models passed F test, and there were extremely significant level. Compared with other models, RRMSE and RE of spectrometeorological yield model apparently declined and the declining range was large, revealing better anticipating effect of spectrometeorological yield model compared to the model of spectrum. Remote sensing estimating value of average yield per unit was slightly larger than statistical value, while yield-estimating precision in dry-land was 80.91% and yield-estimating precision in irrigation-land was 87.72%. Estimating values of total yield were slightly higher than statistical values, where yield-estimating precision in dry-land was 99.20% and 80.54% in irrigation-land.

研究利用遥感数据进行了运城地区冬小麦不同生育时期归一化差值植被指数和产量关系的研究，利用气象数据和光谱数据构建了冬小麦光谱产量模型、气象产量模型以及光谱气象产量模型。结果表明：运城地区水旱地冬小麦均以5月8日左右的NDVI值与产量相关性最好，且达极显著水平，因此该时期为建立冬小麦遥感估产模型的最佳时相。通过对冬小麦光谱产量模型、气象产量模型以及光谱气象产量模型预测效果进行的F检验，表明各模型均达到极显著水平。与其他两种模型相比，光谱气象产量模型的决定系数（R2）有明显的提高，并且相对均方根误差（RRMSE）和相对误差（RE）降低，且降低幅度较大。说明光谱气象产量模型比气象产量模型和光谱产量模型有较好的预测效果。平均单产的遥感估产值略高于统计数据值，旱地估产精度为80.91%，水地估产精度为87.72%；总产量的估算值略高于统计值，旱地精度为99.20%，水地精度为80.54%。因此，利用遥感和气象数据建立模型进行单产和总产估测是可行的，且精度更高。

2006

2.8

0.0

. 2006, 24:87-94 DOI:10.1016/j.eja.2005.05.002

Comparison of an aerialbased system and an on the ground continuous measuring device to predict yield of winter wheat

Abstract

The objective of the research was to compare an aerial image with optical data of an on the ground platform device measuring continuously while driving through the field. For this latter, a Cropscan, Inc. multi-spectral radiometer was used to automatically take a measurement every 1–1.5 m. Field images of the normalized difference vegetation index (NDVI) of both devices were compared in their capability estimating yield variability in a wheat field. An experimental field with three seeding densities and five nitrogen application rates was investigated. Every treatment had four repetitions, resulting in 60 plots of 12 m × 16 m. At harvest, yield variables of the plots (grain and straw yield (t/ha), dry matter content in grain and straw (t/ha), nitrogen in grain and straw (kg[N]/ha) and protein content (%) in grain) were collected to relate to the images of both systems. The NDVI of both measurement systems was well related to applied nitrogen in the field: correlation coefficients were 0.78 and 0.85 for the aerial-based and the ground-based system, respectively. There was a high similarity (correlation coefficient of 0.94) between NDVI measurements of both systems. NDVI was also well related to yield variables at harvest. The NDVI of the ground system was better related to yield variables at harvest compared to NDVI of the aerial system. Best correlation coefficient found for both systems was with nitrogen in grain: 0.84 and 0.91 for the aerial-based and the ground-based system, respectively. NDVI images were more related to yield quality. Grain yield already reached its optimum in the field, while grain quality still increased with increasing nitrogen application rate. For the ground system, the average estimation errors, when using linear regression analysis, were smaller than 10% for all yield variables except for estimated straw yield and estimated nitrogen content in straw. For the aerial system, also estimated total nitrogen content exceeded the 10% error percentage. Besides the higher accuracy in the estimation of yields variables, the ground system had the advantage of being cheaper and that the data were immediately available.

2011

2.474

0.0

... Erdle等^[18]研究发现主动传感器在监测作物生长参数上具有灵活、不受时间和天气条件限制的优点 ...

2005

0.42

0.0

... Raun等^[19]利用主动传感器估测大面积种植的小麦产量, 结合2年的田间数据建立了精确度较高的冬小麦产量估测模型 ...

2007

0.526

0.0

... Thomason等^[20]利用GreenSeeker光谱仪确定了影响玉米产量的敏感生育期, 并以此指导玉米实际生产 ...

2007

1.728

0.0

. 2007, 8:241-252 DOI:10.1007/s11119-007-9043-z

Active remote sensing and grain yield in irrigated maize

1.Department of Soil and Crop Sciences, Colorado State University, Fort Collins, CO USA<br/>2.Department of Forest, Rangeland, and Watershed Stewardship, Colorado State University, Fort Collins, CO USA<br/>

Advances in agricultural technology have led to the development of active remote sensing equipment that can potentially optimize N fertilizer inputs. The objective of this study was to evaluate a hand-held active remote sensing instrument to estimate yield potential in irrigated maize. This study was done over two consecutive years on two irrigated maize fields in eastern Colorado. At the six- to eight-leaf crop growth stage, the GreenSeeker™ active remote sensing unit was used to measure red and NIR reflectance of the crop canopy. Soil samples were taken before side-dressing from the plots at the time of sensing to determine nitrate concentration. Normalized difference vegetation index (NDVI) was calculated from the reflectance data and then divided by the number of days from planting to sensing, where growing degrees were greater than zero. An NDVI-ratio was calculated as the ratio of the reflectance of an area of interest to that of an N-rich portion of the field. Regression analysis was used to model grain yield. Grain yields ranged from 5 to 24 Mg ha⁻¹. The coefficient of determination ranged from 0.10 to 0.76. The data for both fields in year 1 were modeled and cross-validated using data from both fields for year 2. The coefficient of determination of the best fitting model for year 1 was 0.54. The NDVI-ratio had a significant relationship with observed grain yield (r ² = 0.65). This study shows that the GreenSeeker™ active sensor has the potential to estimate grain yield in irrigated maize; however, improvements need to be made.

... Inman等^[21]研究发现玉米八叶期冠层NDVI与产量呈极显著正相关, 因此可在该时期利用NDVI值估测其产量 ...

2004

1.518

0.0

... 另外, Phillips等^[22]研究表明NDVI比RVI能更好地进行小麦产量的估测 ...

2010

0.0

. 2010, 13:121-127 DOI:10.1016/j.ejrs.2010.06.001

Retrieving leaf area index from SPOT4 satellite data

Abstract

A research project was conducted as collaboration between the National Authority for Remote Sensing and Space Sciences (NARSS) in Egypt and the Institute of Remote Sensing Applications (IRSA), Chinese Academy of Sciences. The objective of this study is to generate normalized difference vegetation index (NDVI)–leaf area index (LAI) statistical inversion models for three rice varieties planted in Egypt (Giza-178, Sakha-102, and Sakha-104) using the data of two rice growing seasons. Field observations were carried out to collect LAI field measurements during 2008 and 2009 rice seasons. The SPOT4 satellite data acquired in rice season of 2008 and 2009 conjunction with field observations dates were used to calculate the vegetation indices values. Statistical analyses were performed to confirm the assumptions of inversion modeling for plant variables and to get reliable models that fit the inversion relationship between LAI and NDVI. The inversion process resulted in three NDVI–LAI models adequate to predict LAI with 95% confidence for the three different rice varieties. The accuracy of the generated models ranged between 50% in the case of Sakha-104 and 82% in the case of Giza-178. LAI maps were produced from NDVI imageries based on the generated models.

... 3 讨论已有研究表明冠层植被指数的变化与作物生长状况和发育时期关系紧密^[11,12,23] ...

2012

0.0

1.667

. 2012, 38:747-753 DOI:10.3724/SP.J.1006.2012.00747

NDVI analysis and yield estimation in winter wheat based on GreenSeeker

(基于GreenSeeker的冬小麦NDVI分析与产量估算) . (作物学报) , 2012 , 38 : 747 - 753 Wang L , Bai Y L , Lu Y L , Wang H , Yang L P . NDVI analysis and yield estimation in winter wheat based on GreenSeeker

Wang L

(王磊) , (白由路) , (卢艳丽) , (王贺) , (杨俐苹) . (基于GreenSeeker的冬小麦NDVI分析与产量估算) . (作物学报) , 2012 , 38 : 747 - 753 Wang L

Key Laboratory of Plant Nutrition and Fertilizer, Ministry of Agriculture / Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing 100081, China

The field experiment was carried out in two winter wheat growing seasons during 2007–2009. Winter wheat canopy reflectances in the 656 nm and 770 nm wavelengths were obtained by GreenSeeker at over-wintering stage, reviving stage, jointing stage, early-filling stage, and last-filling stage, respectively, in order to compute NDVI [(NIR₇₇₀-R₆₅₆)/(NIR₇₇₀+R₆₅₆)]. Yields were collected at harvest stage and compared among different nitrogen application rates. Canopy NDVI changes were analyzed with nitrogen increasing and growth, respectively. Furthermore, correlation analysis was done between yield and canopy NDVI. Yield estimation models were established for winter wheat based on canopy NDVI. Results showed that, winter wheat yield changed in parabola shape with N increasing. Wheat canopy NDVI value presented almost changeless before reviving stage, but great increase from reviving stage to early-filling stage with N increasing. In the whole growth stage, canopy NDVI value presented a trend of “low–high–low”. The correlation between canopy NDVI and yield was positive in the whole growth stage with different N rates and gradually became higher with growth, and up to the highest at the late-filling stage. The yield estimation models based on canopy NDVI at early-filling stage (P=0.005) and late-filling stage (P<0.001) were greatly significant. Regression validated between predicted and measured values was significant at early-filling stage (P=0.0129) and greatly significant at late-filling stage (P=0.0002). So, it is feasible that using canopy NDVI at early-filling stage and late-filling stage estimates winter wheat yield, especially at late-filling stage.

以2007—2009年连续2个冬小麦生长季的田间试验数据为基础，利用GreenSeeker获取冠层归一化差值植被指数(NDVI)，分别对不同氮营养条件下冬小麦的产量变化、冠层NDVI值随施氮量和生育期的动态变化，以及NDVI与产量的相关性定量分析，建立基于NDVI的冬小麦产量估算模型。结果表明，冬小麦的产量变化随施氮量的增加呈抛物线趋势变化；冠层NDVI在返青期前随施氮量增加基本不变，返青期至灌浆初期随施氮量增加呈显著增加趋势；整个生育期冠层NDVI呈现“低–高–低”变化趋势。冬小麦整个生育期不同施氮水平下的NDVI值与产量的相关性均为正相关关系，且相关性随生育期逐渐增强，在灌浆末期达到最大。利用NDVI建立的冬小麦产量估算模型，以灌浆初期(P=0.005)和灌浆末期(P<0.001)的模型达到极显著水平。经验证，灌浆初期的冬小麦产量预测值与实测值的回归关系达到了显著水平(P=0.0129)，灌浆末期则达到极显著水平(P=0.0002)。因此，利用灌浆初期和灌浆末期的NDVI值可以预测冬小麦产量，尤以灌浆末期预测效果更佳。

... 王磊等^[24]利用GreenSeeker获取的小麦冠层NDVI均随生育进程表现为先升高后下降的变化趋势 ...

... 鉴于冠层NDVI的估产模型与作物品种、株型、密度、肥水等条件相关, 受环境影响变化明显^[24], 基于主动传感器获取冠层NDVI对于大豆产量估测具有重要的指导意义, 但本研究采用的主动传感器由于受光谱波段的限制, 对大豆产量估测的精度有一定的影响 ...

2005

2.474

0.0

... 已有研究表明, 利用冠层植被指数可以间接估算作物的产量^{[25,26,27,28,29,30]} ...

1981

1.138

0.0

... 已有研究表明, 利用冠层植被指数可以间接估算作物的产量^{[25,26,27,28,29,30]} ...

1991

1.621

0.0

... 已有研究表明, 利用冠层植被指数可以间接估算作物的产量^{[25,26,27,28,29,30]} ...

2001

0.526

0.0

... 已有研究表明, 利用冠层植被指数可以间接估算作物的产量^{[25,26,27,28,29,30]} ...

2001

1.518

0.0

... 已有研究表明, 利用冠层植被指数可以间接估算作物的产量^{[25,26,27,28,29,30]} ...

2002

2.041

0.0

... 已有研究表明, 利用冠层植被指数可以间接估算作物的产量^{[25,26,27,28,29,30]} ...