• ISSN 1008-505X
  • CN 11-3996/S

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于主动冠层光谱仪的植被归一化指数估测莴苣(Lactuca sativa L.)生物量及氮素营养状况的准确性研究

纪荣婷 闵炬 王远 陆志新 路广 施卫明

引用本文:
Citation:

基于主动冠层光谱仪的植被归一化指数估测莴苣(Lactuca sativa L.)生物量及氮素营养状况的准确性研究

    作者简介: 纪荣婷 E–mail:jirongting@nies.org;
    通讯作者: 施卫明, E-mail:wmshi@issas.ac.cn
  • 基金项目: 国家重点研发计划项目(2017YFD0800404),江苏省农业科技自主创新资金项目(CX(18)1005);山东省重大科技创新工程项目(2019JZZY010701)。

Estimating accuracy of lettuce (Lactuca sativa L.) biomass and nitrogen status based on normalized difference vegetation index (NDVI) values obtained using an active canopy sensor

    Corresponding author: SHI Wei-ming, E-mail:wmshi@issas.ac.cn
  • 摘要:   【目的】  研究冠层光谱技术在蔬菜氮素营养诊断中应用的可行性和提高其准确性的方法,为推进蔬菜氮素营养管理与施肥推荐提供快速无损检测技术。  【方法】  以茎菜类蔬菜—莴苣 (Lactuca sativa L.) 为研究对象进行田间试验。设置5个化肥年施用梯度:0、108、162、216、270 kg/hm2,在莴苣幼苗期、莲座期、茎形成期和收获期,利用GreenSeeker冠层光谱仪获取冠层光谱特征值—植被归一化指数 (NDVI) 和比值植被指数 (RVI),并测定植株生物量和含氮量。计算了用生育期NDVI和RVI值预测蔬菜生物量和氮素营养的可行性与准确性。并验证了用移栽天数校正提高全生育期光谱值预测精度的可行性。  【结果】  NDVI和RVI与莴苣地上部生物量 (AGB)、根冠比 (RTS)、植株吸氮量 (PNU) 和植株氮浓度 (PNC) 等指标间均存在显著相关关系,尤其以NDVI相关性更高。相关性分析结果表明,NDVI与AGB和PNU呈正相关,相关系数分别介于0.779~0.945和0.819~0.938;与RTS和PNC呈负相关,相关性系数介于–0.367~–0.844和–0.328~–0.732。对比不同时期,莲座期和茎形成期的NDVI值对莴苣生物量和氮素营养指标预测的准确性较高,对AGB、RTS、PNU和PNC预测准确性分别为0.76~0.92、0.37~0.71、0.77~0.88和0.34~0.54。利用两年NDVI值建立各时期莴苣生物量和氮素营养状况统一预测方程,莲座期方程最为准确,对AGB、RTS、PNU、PNC预测准确性分别为73%、48%、52%、31%。综合全生育预测方程,冠层光谱仪测定的NDVI值对莴苣生物量和氮素营养预测指标的准确性较高,基于NDVI值的AGB、RTS、PNU和PNC预测方程准确度分别为54%、43%、57%和26%。引入移栽天数 (DAT) 对该预测方程进行校正后,AGB、PNU和PNC预测方程的准确度分别提高至62%、71%和34%。  【结论】  基于冠层光谱仪测定的各生育期的植被归一化指标 (NDVI) 可准确预测莴苣的生物量和氮素营养状况,尤以莲座期的预测结果最为准确。经移栽天数 (DAT) 校正后,基于全生育期的NDVI值建立的预测方程对AGB、PNU的预测准确度可分别提高到62%和71%,基本满足莴苣类低覆盖度蔬菜作物的氮素营养管理。
  • 图 1  小区布置图及莴苣冠层光谱测定示意图

    Figure 1.  Plot layout and canopy spectrum sensing diagram

    图 2  冠层植被指数NDVI(a) 和RVI(b) 与Year I和Year II莴苣地上部生物量 (AGB) 全生育期预测结果分析 (n=120)

    Figure 2.  Relationships between active canopy sensor-based normalized difference vegetation index (NDVI) (a) and ratio vegetation index (RVI) (b) and the aboveground biomass (AGB) of lettuce at all stages in year I and year II (n=120)

    图 3  冠层植被指数NDVI(a) 和RVI(b) 与Year I和Year II莴苣根冠比 (RTS) 全生育期预测结果分析 (n=120)

    Figure 3.  Relationships between active canopy sensor-based normalized difference vegetation index (NDVI) (a) and ratio vegetation index (RVI) (b) and the root to shoot ratio (RTS) of lettuce at all stages in year I and year II (n=120)

    图 4  冠层植被指数NDVI(a) 和RVI(b) 与Year I和Year II莴苣植株吸氮量 (PNU) 全生育期预测结果分析 (n=120)

    Figure 4.  Relationships between active canopy sensor-based normalized difference vegetation index (NDVI) (a) and ratio vegetation index (RVI) (b) and the plant N uptake (PNU) of lettuce at all stages in year I and year II (n=120)

    图 5  冠层植被指数NDVI(a) 和RVI(b) 与Year I和Year II莴苣植株氮浓度 (PNC) 全生育期预测结果分析 (n=120)

    Figure 5.  Relationships between active canopy sensor-based normalized difference vegetation index (NDVI) (a) and ratio vegetation index (RVI) (b) and the plant N concentration (PNC) of lettuce at all stages in year I and year II (n=120)

    表 1  Year I和Year II莴苣移栽、收获日期及各生长发育时期具体采样日期 (mm-dd)

    Table 1.  Detailed transplanting, harvesting and sensing date in each stage of lettuces in Years I and II

    年份
    Year
    移栽日期 (month-day)
    Transplanting date
    收获日期 (month-day)
    Harvest date
    测定日期 (month-day)
    Sensing date (growth stage)
    Year I09-2812-2510-21 (幼苗期Seedling),11-05 (莲座期Rosette),
    11-16 (茎形成期Stem formation),12-03 (收获期Harvest)
    Year II10-2901-1511-24 (幼苗期Seedling),12-07 (莲座期Rosette),
    12-19 (茎形成期Tuber formation),01-07 (收获期Harvest)
    下载: 导出CSV

    表 2  Year I和Year II莴苣不同生长期冠层光谱仪测定的NDVI及RVI与植株生长和N利用指标间的相关系数值 (r)

    Table 2.  Correlation coefficients (r) of the sensed NDVI and RVI with plant growth and nitrogen use index of lettuce in Years I and II

    生长期
    Growth stage
    指标
    Index
    AGBRTSPNUPNCAGBRTSPNUPNC
    Year IYear II
    幼苗期
    Seedling stage
    NDVI0.803**–0.586*0.820**–0.557*0.944**–0.3670.925**–0.608*
    RVI0.820**–0.580*0.831**–0.599*0.953**–0.3610.933**–0.603*
    莲座期
    Rosette stage
    NDVI0.798**–0.703**0.819**–0.732**0.922**–0.582*0.908**–0.638*
    RVI0.862**–0.684**0.874**–0.677**0.958**–0.636*0.914**–0.689*
    茎形成期
    Stem formation stage
    NDVI0.945**–0.844**0.938**–0.720**0.895**–0.470*0.862**–0.586*
    RVI0.942**–0.852**0.938**–0.771**0.936**–0.478*0.889**–0.617*
    收获期
    Harvest stage
    NDVI0.879**–0.748**0.860**–0.617*0.779**–0.512*0.843**–0.328
    RVI0.880**–0.725**0.857**–0.616*0.866**–0.464*0.774**–0.495
    全生育期
    Pooled stages
    NDVI0.752**–0.594**0.840**–0.538**0.857**–0.674**0.918**–0.736**
    RVI0.729**–0.558**0.871**–0.544**0.891**–0.615**0.878**–0.706**
    注(Note):NDVI—植被归一化指数 Normalized difference vegetation index; RVI—比值植被指数 Rratio vegetation index; AGB—地上部生物量 Aboveground biomass; RTS—根冠比 Root to shoot ratio; PNU—植株吸氮量 Plant nitrogen uptake; PNC—植株氮浓度 Plant nitrogen concentration; *—P < 0.05;**—P < 0.01;各时期取样数为 n=15,全生育期为 n=60 Sampling number in each growing stage was n=15 and in pooled stages n=60.
    下载: 导出CSV

    表 3  不同时期莴苣冠层NDVI和RVI值与其生长和氮素营养指标间3 种回归方程分析对比结果

    Table 3.  The correlation coefficients (R2) between canopy indexes (NDVI and RVI) and plant growth and nitrogen status indexes of lettuce using three different equations at different stages

    年份
    Year
    时期
    Growth stage
    指标
    Index
    AGBRTSPNUPNC
    LEPLEPLEPLEP
    Year I幼苗期
    Seedling stage
    NDVI0.65**0.68**0.66**0.34*0.35*0.35*0.67**0.69**0.68**0.31*0.300.26
    RVI0.67**0.70**0.68**0.34*0.34*0.34*0.69**0.71**0.70**0.36*0.35*0.31*
    莲座期
    Rosette stage
    NDVI0.64**0.76**0.75**0.49*0.51*0.54**0.67**0.77**0.76**0.54**0.54**0.54**
    RVI0.74**0.71**0.75**0.47*0.47*0.50*0.76**0.74**0.77**0.46**0.47**0.52**
    茎形成期
    Stem formation stage
    NDVI0.89**0.89**0.89**0.71**0.69**0.68**0.88**0.88**0.88**0.52**0.50**0.49**
    RVI0.89**0.85**0.88**0.73**0.73**0.71**0.87**0.85**0.87**0.60**0.58**0.53**
    收获期
    Harvest stage
    NDVI0.77**0.77**0.77**0.56**0.58**0.58**0.74**0.73**0.74**0.38*0.38*0.38*
    RVI0.77**0.76**0.77**0.53**0.52**0.57**0.74**0.71**0.73**0.37*0.37*0.38*
    Year II幼苗期
    Seedling stage
    NDVI0.89**0.91**0.92**0.110.110.010.86**0.88**0.88**0.37*0.37*0.36*
    RVI0.91**0.90**0.91**0.120.120.110.87**0.87**0.88**0.36*0.37*0.37*
    莲座期
    Rosette stage
    NDVI0.85**0.92**0.90**0.62**0.66**0.71**0.82**0.83**0.84**0.41*0.40*0.33*
    RVI0.92**0.91**0.93**0.53**0.58**0.64**0.83**0.78**0.82**0.47**0.46**0.42**
    茎形成期
    Stem formation stage
    NDVI0.80**0.87**0.84**0.36*0.37*0.37*0.74**0.79**0.770.34*0.34*0.29
    RVI0.88**0.87**0.88**0.32*0.34*0.36*0.79**0.77**0.79**0.38*0.38*0.35*
    收获期
    Harvest stage
    NDVI0.61**0.60**0.61**0.43*0.44*0.44*0.65**0.59**0.66**0.110.110.10
    RVI0.73**0.77**0.74**0.40*0.42*0.45*0.60**0.55**0.64**0.240.230.23
    注(Note):NDVI—植被归一化指数 Normalized difference vegetation index; RVI—比值植被指数 Rratio vegetation index; AGB—地上部生物量 Aboveground biomass; RTS—根冠比 Root to shoot ratio; PNU—植株吸氮量 Plant nitrogen uptake; PNC—植株氮浓度 Plant nitrogen concentration; L—线性方程 Linear equation:$y = a \times {x_{NDVI}} + b$; E—指数方程 Exponential equation: $y = a \times {e^{b \times {x_{NDVI}}}}$;P—幂函数 Power function equation:$y = a \times {x_{NDVI}}^b$,a、b 为方程的回归参数 a and b are the regression parameters of equations.
    下载: 导出CSV

    表 4  不同时期莴苣冠层植被指数 (NDVI和RVI) 与植物生长和氮素营养指标间预测方程及决定系数

    Table 4.  The prediction equation and coefficients of determination between canopy indexes (NDVI and RVI) and biomass and nitrogen status indexes (AGB, RTS, PNU and PNC) of lettuce using three different equations at different stages

    年份
    Year
    生长期
    Growth stage
    指标
    Index
    AGBRTSPNUPNC
    预测方程
    Equation
    R2RMSE预测方程
    Equation
    R2RMSE预测方程
    Equation
    R2RMSE预测方程
    Equation
    R2RMSE
    Year I幼苗期
    Seedling stage
    NDVIy = 41.6e4.20x0.6848.8y = 0.096x–0.430.350.021y = 0.19e3.42x0.690.14y = –2.14 x + 4.270.310.27
    RVIy = 33.8e0.79 x0.7047.5y = 0.23x–0.520.340.021y = 0.16e0.65x0.710.13y = –0.44 x + 4.470.360.26
    莲座期
    Rosette stage
    NDVIy = 3.84e7.82 x0.76235y = 0.057 x–1.280.540.029y = 0.026e6.10x0.770.40y = 2.78 x–0.360.540.22
    RVIy = 23.9 x2.080.75240y = 0.33 x–0.760.500.030y = 0.099 x1.680.770.40y = 4.48 x–2.040.520.22
    茎形成期
    Tuber formation
    stage
    NDVIy = 2489x5.870.8956.6y = –0.60 x–0.590.710.016y = 7.54 x5.910.880.18y = –6.68 x + 8.500.520.27
    RVIy = 107 x – 2950.8957.9y = –0.017 x + 0.260.730.015y = 0.32 x–0.900.870.18y = –0.20 x + 4.930.600.25
    收获期
    Harvest stage
    NDVIy = 4832 x7.480.7771.7y = 0.028 x–5.630.580.016y = 17.74 x–11.780.740.21y = 2.77 x–0.540.380.058
    RVIy = 175 x – 6640.7771.9y = 2.71 x–1.520.570.016y = -0.017 x + 0.260.740.21y = 4.28 x–0.140.380.058
    Year II幼苗期
    Seedling stage
    NDVIy = 5655 x2.630.9219.7y = –0.19 x + 0.210.110.028y = 0.030e9.89x0.880.057y = –4.51 x + 4.630.370.30
    RVIy = 10.1 x5.160.9121.0y = –0.061 x + 0.260.120.028y = 0.033 x4.650.880.057y = –6.55e-0.38x0.370.30
    莲座期
    Rosette stage
    NDVIy = 41.2e4.73x0.9247.8y = 0.063 x–0.750.710.019y = 4.38 x1.560.830.20y = –1.83 x + 4.140.410.24
    RVIy = 58.2 x1.820.9346.8y = 0.27 x–0.840.640.021y = 0.65 x–0.500.830.20y = –0.32 x + 4.180.470.24
    茎形成期
    Tuber formation
    stage
    NDVIy = 78.8e3.66x0.8757.1y = 0.23e–1.37x0.370.022y = 0.32e3.39x0.790.25y = –1.63 x + 3.870.340.21
    RVIy = 109 x1.360.8854.4y = 0.21 x–0.550.360.022y = 0.71 x–0.420.790.25y = –0.24 x + 3.790.380.21
    收获期
    Harvest stage
    NDVIy = 1765 x1.090.61198y = 0.092 x–0.370.440.013y = 3.90 x0.830.660.37y = –0.64 x + 3.160.110.25
    RVIy = 417e0.21x0.77150y = 0.16 x–0.280.450.013y = 1.15 x0.570.640.38y = –0.098 x + 3.170.240.23
    Year I &
    Year II
    幼苗期
    Seedling stage
    NDVIy = 766x–65.80.7643.9y = –0.14 x + 0.200.240.025y = 3.12x1.520.850.11y = 2.95 x–0.150.290.30
    RVIy = 169 x–1620.7445.4y = 0.23e–0.20x0.240.025y = 0.59 x–0.610.840.12y = 4.12 x–0.230.290.30
    莲座期
    Rosette stage
    NDVIy = 21.9e5.38x0.73195y = 0.075 x-0.640.480.027y = 0.27e2.77x0.520.46y = –1.15 x + 3.950.310.26
    RVIy = 54.5 x1.630.75188y = 0.22 x–0.530.450.028y = 0.52e0.23x0.550.44y = 3.87 x–0.120.310.26
    茎形成期
    Tuber formation
    stage
    NDVIy = 791 x0.850.55119y = 0.11 x–0.260.140.026y = 2.25 x0.580.330.43y = –0.67 x + 3.490.130.30
    RVIy = 329e0.077x0.49126y = –0.003 x + 0.140.120.026y = 1.24e0.052x0.260.45y = –0.040 x + 3.310.150.30
    收获期
    Harvest stage
    NDVIy = 923 x0.290.07238y = 0.11 x–0.200.120.019y = 2.38 x0.230.080.51y = 2.87 x–0.0580.010.19
    RVIy = 709 x0.0810.02244y = 0.14 x–0.100.100.019y = 1.93 x0.0640.020.52y = 0.0036 x + 2.860.010.19
    注(Note):NDVI—植被归一化指数 Normalized difference vegetation index; RVI—比值植被指数 Rratio vegetation index; AGB—地上部生物量 Aboveground biomass; RTS—根冠比 Root to shoot ratio; PNU—植株吸氮量 Plant nitrogen uptake; PNC—植株氮浓度 Plant nitrogen concentration.
    下载: 导出CSV

    表 5  引入移栽天数 (DAT) 与冠层植被指数 (NDVI和RVI) 估测莴苣地上部生物量 (AGB)、根冠比 (RTS)、植株吸氮量 (PNU) 和植株氮浓度 (PNC) 的预测方程

    Table 5.  Equations for estimating aboveground biomass (AGB), root to shoot ratio (RTS), plant nitrogen uptake (PNU) and plant nitrogen concentration (PNC) using sensor-based measured normalized difference vegetation index (NDVI) and ratio vegetation index (RVI) and days after transplanting (DAT)

    预测指标 Predicted index预测方程 Predicted equation
    NDVI-AGBAGB = 830 × NDVI + 7.81 × DAT-303
    RVI-AGBAGB = 50.0 × RVI + 9.91 × DAT − 179
    NDVI-RTSRTS = –0.0094 × NDVI − 1.38 × 10–4 × DAT + 0.18
    RVI-RTSRTS = –0.0051 × RVI − 4.23 × 10–4 × DAT + 0.17
    NDVI-PNUPNU = 1.87 × NDVI-0.023 × DAT – 0.65
    RVI-PNUPNU = 0.11 × RVI − 0.028 × DAT + 0.37
    NDVI-PNCPNC = –0.30 × NDVI − 0.011 × DAT + 3.96
    RVI-PNCPNC = –0.0066 × RVI − 0.013 × DAT + 3.91
    下载: 导出CSV

    表 6  引入移栽天数 (DAT) 校正后莴苣生物量和氮素营养状况预测方程参数变化

    Table 6.  Change of the regression parameters in the model for lettuce biomass and N status predictions before and after modified with transplanting days (DAT)

    预测方程
    Predicted equation
    校正前 Original parameter校正后 Modified parameter校正效果 Change rate (%)
    R2RMSER2RMSER2RMSE
    NDVI-AGB0.542280.6220514.81–10.09
    RVI-AGB0.442510.5721929.55–12.75
    NDVI-RTS0.430.0250.370.026–13.954.00
    RVI-RTS0.360.0270.280.028–22.223.70
    NDVI-PNU0.570.520.710.4224.56–19.23
    RVI-PNU0.470.580.660.4640.43–20.69
    NDVI-PNC0.260.320.340.3030.77–6.25
    RVI-PNC0.170.340.330.3194.12–8.82
    注(Note):校正效果 Change rate (%) = (校正后值 Modified parameter-校正前值 Original parameter) /校正前值 Original parameter × 100.
    下载: 导出CSV
  • [1] Michel P. Viruses and virus diseases of vegetables in the Mediterranean Basin[J]. In Advances in Virus Research, 2012.
    [2] Greenwood D J, Cleaver T J, Turner M K, et al. Comparison of the effects of nitrogen fertilizer on the yield, nitrogen content and quality of 21 different vegetable and agricultural crops[J]. Journal of Agricultural Science, 1980, 95: 471–485. doi:  10.1017/S0021859600039514
    [3] 汤丽玲, 张晓晟, 陈清, 等. 蔬菜氮素营养与品质[J]. 北方园艺, 2002, (3): 6–7. Tang L L, Zhang X S, Chen Q, et al. Nitrogen nutrition and quality of vegetable[J]. Northern Horticulture, 2002, (3): 6–7. doi:  10.3969/j.issn.1001-0009.2002.03.002
    [4] Liu C W, Sung Y, Chen B C, et al. Effects of nitrogen fertilizers on the growth and nitrate content of lettuce (Lactuca sativa L.)[J]. International Journal of Environmental Research and Public Health, 2014, 11(4): 4427–4440. doi:  10.3390/ijerph110404427
    [5] Min J, Zhao X, Shi W M, et al. Nitrogen balance and loss in a greenhouse vegetable system in southeastern China[J]. Pedosphere, 2011, 21: 464–472. doi:  10.1016/S1002-0160(11)60148-3
    [6] Li B, Bi Z C, Xiong Z Q. Dynamic responses of nitrous oxide emission and nitrogen use efficiency to nitrogen and biochar amendment in an intensified vegetable field in southeastern China[J]. GCB Bioenergy, 2017, 9: 400–413. doi:  10.1111/gcbb.12356
    [7] Burns I G. Assessing N fertiliser requirements and the reliability of different recommendation systems[J]. Acta Horticulturae, 2006: 35–48.
    [8] 万春雁, 糜林, 李金凤, 等. 氮素营养诊断技术在我国园艺作物上的应用现状[J]. 江苏农业科学, 2011, 39(6): 322–324. Wang C Y, Mi L, Li J F, et al. Application status of nitrogen nutrition diagnosis technology of horticultural crops in China[J]. Jiangsu Agricultural Sciences, 2011, 39(6): 322–324. doi:  10.3969/j.issn.1002-1302.2011.06.129
    [9] Samborski S M, Tremblay N, Fallon E. Strategies to make use of plant sensors-based diagnostic information for nitrogen recommendations[J]. Agronomy Journal, 2009, 101: 800–816. doi:  10.2134/agronj2008.0162Rx
    [10] Raun W R, Solie J B, Johnson G V, et al. In-season prediction of potential grain yield in winter wheat using canopy reflectance[J]. Agronomy Journal, 2001, 93: 131–138. doi:  10.2134/agronj2001.931131x
    [11] Campbell J B. Introduction to remote sensing[M]. New York: The Guilford Press, 2002.
    [12] Raun W R, Solie J B, Stone M L, et al. Optical sensor-based algorithm for crop nitrogen fertilization[J]. Communications in Soil Science and Plant Analysis, 2005, 36: 2759–2781. doi:  10.1080/00103620500303988
    [13] Yao Y K, Miao Y X, Huang S Y, et al. Active canopy sensor-based precision N management strategy for rice[J]. Agronomy for Sustainable Development, 2012, 32: 925–933. doi:  10.1007/s13593-012-0094-9
    [14] Xia T T, Miao Y X, Wu D L, et al. Active optical sensing of spring maize for in-season diagnosis of nitrogen status based on nitrogen nutrition index[J]. Remote Sensing, 2016, 8: 605. doi:  10.3390/rs8070605
    [15] Li F, Miao Y X, Chen X P, et al. Estimating winter wheat biomass and nitrogen status using an active crop sensor[J]. Intelligent Automation and Soft Computing, 2010, 16: 1221–1230.
    [16] Xia T T, Miao Y X, Wu D L, et al. Active optical sensing of spring maize for in-season diagnosis of nitrogen status based on nitrogen nutrition index[J]. Remote Sensing, 2016, 8: 605. doi:  10.3390/rs8070605
    [17] Dunn B, Shrestha A, Goad C. Use of nondestructive sensors to quantify ornamental kale nitrogen status[J]. Journal of Plant Nutrition, 2015, 39: 1123–1130.
    [18] Sanderson K R, Fillmore S A E. Slow-release nitrogen fertilizer in carrot production on Prince Edward Island[J]. Canada Journal of Plant Science, 2012, 92: 1223–1228. doi:  10.4141/cjps2011-201
    [19] Ji R T, Min J, Wang Y, et al. In-season yield prediction of cabbage with a hand-held active canopy sensor[J]. Sensors, 2017, 17: 2287. doi:  10.3390/s17102287
    [20] Bremner J M. Determination of nitrogen in soil by the Kjeldahl method[J]. Journal of Agricultural Science, 1960, 55: 11–33. doi:  10.1017/S0021859600021572
    [21] Muñoz-Huerta R F, Guevara-Gonzalez R G, Contreras-Medina L M, et al. A review of methods for sensing the nitrogen status in plants: Advantages, disadvantages and recent advances[J]. Sensors, 2013, 13(8): 10823–10843. doi:  10.3390/s130810823
    [22] Cao Q, Miao Y X, Feng G H, et al. Active canopy sensing of winter wheat nitrogen status: An evaluation of two sensor systems[J]. Computers and Electronics in Agriculture, 2015, 112: 54–67. doi:  10.1016/j.compag.2014.08.012
    [23] Padilla F M, Peña-Fleitas M T, Gallardo M, et al. Derivation of sufficiency values of a chlorophyll meter to estimate cucumber nitrogen status and yield[J]. Computers and Electronics in Agriculture, 2017, 141: 54–64. doi:  10.1016/j.compag.2017.07.005
    [24] 于静, 李斐, 秦永林, 等. 应用主动作物冠层传感器对马铃薯氮素营养诊断[J]. 光谱学与光谱分析, 2013, 33(11): 3092–3097. Yu J, Li F, Qin Y L, et al. Active crop canopy sensor-based nitrogen diagnosis for potato[J]. Spectroscopy and Spectral Analysis, 2013, 33(11): 3092–3097. doi:  10.3964/j.issn.1000-0593(2013)11-3092-06
    [25] 张绪成, 郭天文, 谭雪莲, 等. 氮素水平对小麦根-冠生长及水分利用效率的影响[J]. 西北农业学报, 2008, 17(3): 97–102. Zhang X C, Guo T W, Tan X L, et al. The effects of nitrogen level on the root and shoot growth and their relationship with the water use efficiency in wheat plants[J]. Acta Agriculturae Boreali-Occidentalis Sinica, 2008, 17(3): 97–102. doi:  10.3969/j.issn.1004-1389.2008.03.023
    [26] Cao Q, Miao Y X, Feng G H, et al. Active canopy sensing of winter wheat nitrogen status: An evaluation of two sensor systems[J]. Computers and Electronics in Agriculture, 2015, 112: 54–67. doi:  10.1016/j.compag.2014.08.012
    [27] Qi J, Chehbouni A, Huete A R, et al. A modified soil adjusted vegetation index[J]. Remote Sensing of Environment, 1994, 48(2): 119–126. doi:  10.1016/0034-4257(94)90134-1
    [28] Gu Y X, Wylie B K, Howard D M, et al. NDVI saturation adjustment: A new approach for improving cropland performance estimates in the Greater Platte River Basin, USA[J]. Ecological Indicators, 2013, 30: 1–6. doi:  10.1016/j.ecolind.2013.01.041
    [29] Li F, Miao Y X, Hennig S D, et al. Evaluating hyperspectral vegetation indices for estimating nitrogen concentration of winter wheat at different growth stages[J]. Precision Agriculture, 2010, 11: 335–357. doi:  10.1007/s11119-010-9165-6
    [30] Hatfield J L, Prueger J H. Value of using different vegetative indices to quantify agricultural crop characteristics at different growth stages under varying management practices[J]. Remote Sensing, 2010, 2(2): 562–578. doi:  10.3390/rs2020562
    [31] Ali A M, Thind H S, Sharma S, et al. Prediction of dry direct-seeded rice yields using chlorophyll meter, leaf color chart and Greenseeker optical sensor in northwestern India[J]. Field Crops Research, 2014, 161: 11–15. doi:  10.1016/j.fcr.2014.03.001
    [32] Liu X, Ferguson R B, Zheng H., et al Using an active-optical sensor to develop an optimal NDVI dynamic model for high-yield rice production (Yangtze, China)[J]. Sensors, 2017, 17(4): 672. doi:  10.3390/s17040672
    [33] Xue L H, Yang L Z. Recommendations for nitrogen fertiliser topdressing rates in rice using canopy reflectance spectra[J]. Biosystems Engineering, 2008, 100(4): 524–534. doi:  10.1016/j.biosystemseng.2008.05.005
    [34] Wang Y, Wang D J, Shi P H, et al. Estimating rice chlorophyll content and leaf nitrogen concentration with a digital still color camera under natural light[J]. Plant Methods, 2013, 10: 36.
  • [1] 张银杰王磊白由路卢艳丽张静静李格 . 玉米不同层位叶片生理生化指标与SPAD值的关系. 植物营养与肥料学报, doi: 10.11674/zwyf.20182
    [2] 白丽敏李粉玲常庆瑞曾凤曹吉芦光旭 . 结合SPA和PLS法提高冬小麦冠层全氮高光谱估算的精确度. 植物营养与肥料学报, doi: 10.11674/zwyf.17478
    [3] 李青军张炎哈丽哈什•依巴提冯固 . 漫灌和滴灌棉花土壤有效磷丰缺指标与临界值研究. 植物营养与肥料学报, doi: 10.11674/zwyf.17419
    [4] 毛伟李文西高晖陈欣姜义杭天文龚鑫鑫陈明张月平 . 扬州市耕地土壤pH值30年演变及其驱动因子. 植物营养与肥料学报, doi: 10.11674/zwyf.16442
    [5] 杨虎戈长水应武杨京平李金文何俊俊 . 遮荫对水稻冠层叶片SPAD值及光合、 形态特性参数的影响. 植物营养与肥料学报, doi: 10.11674/zwyf.2014.0308
    [6] 张德闪王宇蕴汤利郑毅左建花 . 小麦蚕豆间作对红壤有效磷的影响及其与根际pH值的关系. 植物营养与肥料学报, doi: 10.11674/zwyf.2013.0115
    [7] 沈月依艳丽* . 不同因素交互作用对棕壤硝态氮累积及pH值的影响. 植物营养与肥料学报, doi: 10.11674/zwyf.2013.0517
    [8] 刘小勇董铁王发林张坤张辉元马明 . 甘肃省元帅系苹果叶营养元素含量标准值研究. 植物营养与肥料学报, doi: 10.11674/zwyf.2013.0130
    [9] 许学慧姜冠杰付庆灵刘永红胡红青 . 活化磷矿粉对重金属污染土壤上莴苣生长与品质的影响. 植物营养与肥料学报, doi: 10.11674/zwyf.2013.0212
    [10] 赵士诚何萍仇少君徐新朋串丽敏邢素丽 . 相对SPAD值用于不同品种夏玉米氮肥管理的研究. 植物营养与肥料学报, doi: 10.11674/zwyf.2011.1059
    [11] 张俊华张佳宝钦绳武 . 不同施肥长期定位试验地夏玉米冠层光谱特征研究. 植物营养与肥料学报, doi: 10.11674/zwyf.2010.0414
    [12] 蔡红光米国华陈范骏张秀芝高强 . 玉米叶片SPAD值、全氮及硝态氮含量的品种间变异. 植物营养与肥料学报, doi: 10.11674/zwyf.2010.0413
    [13] 胡昊白由路*杨俐苹卢艳丽王磊王贺孔庆波 . 不同氮营养冬小麦冠层光谱红边特征分析. 植物营养与肥料学报, doi: 10.11674/zwyf.2009.0611
    [14] 房玉林宋士任王华陈书霞史世梦周晓明 . NaCl胁迫对葡萄幼苗根际pH值及营养成分的影响 . 植物营养与肥料学报, doi: 10.11674/zwyf.2008.0431
    [15] 郭建华王秀孟志军赵春江宇振荣陈立平 . 主动遥感光谱仪Greenseeker与SPAD对玉米氮素营养诊断的研究 . 植物营养与肥料学报, doi: 10.11674/zwyf.2008.0107
    [16] 肖时运刘强荣湘民谢桂先廖育林 . 不同施氮水平对莴苣产量、品质及氮肥利用率的影响. 植物营养与肥料学报, doi: 10.11674/zwyf.2006.0625
    [17] 贾莉君范晓荣尹晓明沈其荣 . pH值对水稻幼苗吸收NO3--的影响. 植物营养与肥料学报, doi: 10.11674/zwyf.2006.0508
    [18] 谭和平陈能武黄苹叶善蓉 . 四川茶区土壤营养元素背景值研究. 植物营养与肥料学报, doi: 10.11674/zwyf.2006.0605
    [19] 王坷许红卫史舟 , . 土壤钾素空间变异性和空间插值方法的比较研究. 植物营养与肥料学报, doi: 10.11674/zwyf.2000.0312
    [20] 胡承孝庞静魏文学杜应琼王运华陈浩陈志伟 . 利用钼值作为冬小麦缺钼土壤诊断指标的探讨. 植物营养与肥料学报, doi: 10.11674/zwyf.1998.0411
  • 加载中
WeChat 点击查看大图
图(5)表(6)
计量
  • 文章访问数:  317
  • HTML全文浏览量:  135
  • 被引次数: 0
出版历程
  • 收稿日期:  2020-06-09

基于主动冠层光谱仪的植被归一化指数估测莴苣(Lactuca sativa L.)生物量及氮素营养状况的准确性研究

    作者简介:纪荣婷 E–mail:jirongting@nies.org
    通讯作者: 施卫明, wmshi@issas.ac.cn
  • 1. 生态环境部南京环境科学研究所,江苏南京 210042
  • 2. 中国科学院南京土壤研究所土壤与农业可持续发展国家重点实验室,江苏南京 210008
  • 3. 宜兴市蔬菜办公室,江苏宜兴 214206
  • 基金项目: 国家重点研发计划项目(2017YFD0800404),江苏省农业科技自主创新资金项目(CX(18)1005);山东省重大科技创新工程项目(2019JZZY010701)。
  • 摘要:   【目的】  研究冠层光谱技术在蔬菜氮素营养诊断中应用的可行性和提高其准确性的方法,为推进蔬菜氮素营养管理与施肥推荐提供快速无损检测技术。  【方法】  以茎菜类蔬菜—莴苣 (Lactuca sativa L.) 为研究对象进行田间试验。设置5个化肥年施用梯度:0、108、162、216、270 kg/hm2,在莴苣幼苗期、莲座期、茎形成期和收获期,利用GreenSeeker冠层光谱仪获取冠层光谱特征值—植被归一化指数 (NDVI) 和比值植被指数 (RVI),并测定植株生物量和含氮量。计算了用生育期NDVI和RVI值预测蔬菜生物量和氮素营养的可行性与准确性。并验证了用移栽天数校正提高全生育期光谱值预测精度的可行性。  【结果】  NDVI和RVI与莴苣地上部生物量 (AGB)、根冠比 (RTS)、植株吸氮量 (PNU) 和植株氮浓度 (PNC) 等指标间均存在显著相关关系,尤其以NDVI相关性更高。相关性分析结果表明,NDVI与AGB和PNU呈正相关,相关系数分别介于0.779~0.945和0.819~0.938;与RTS和PNC呈负相关,相关性系数介于–0.367~–0.844和–0.328~–0.732。对比不同时期,莲座期和茎形成期的NDVI值对莴苣生物量和氮素营养指标预测的准确性较高,对AGB、RTS、PNU和PNC预测准确性分别为0.76~0.92、0.37~0.71、0.77~0.88和0.34~0.54。利用两年NDVI值建立各时期莴苣生物量和氮素营养状况统一预测方程,莲座期方程最为准确,对AGB、RTS、PNU、PNC预测准确性分别为73%、48%、52%、31%。综合全生育预测方程,冠层光谱仪测定的NDVI值对莴苣生物量和氮素营养预测指标的准确性较高,基于NDVI值的AGB、RTS、PNU和PNC预测方程准确度分别为54%、43%、57%和26%。引入移栽天数 (DAT) 对该预测方程进行校正后,AGB、PNU和PNC预测方程的准确度分别提高至62%、71%和34%。  【结论】  基于冠层光谱仪测定的各生育期的植被归一化指标 (NDVI) 可准确预测莴苣的生物量和氮素营养状况,尤以莲座期的预测结果最为准确。经移栽天数 (DAT) 校正后,基于全生育期的NDVI值建立的预测方程对AGB、PNU的预测准确度可分别提高到62%和71%,基本满足莴苣类低覆盖度蔬菜作物的氮素营养管理。

    English Abstract

    • 莴苣 (Lactuca sativa L.) 属一年生或二年生草本菊科,是一种重要的蔬菜作物,栽培莴苣起源于西南亚地区,目前已在全世界广泛种植[1]。莴苣是我国常见的蔬菜种植类型,味道鲜美,富含多种矿质元素,在太湖地区秋冬季广泛种植。因其产量和经济效益较高,且氮素需求量较大,在现实生产中人们通常投入大量肥料,特别是氮肥以达到高产目的[2]。充足的氮素供应可提高莴苣中矿质养分含量,改善碳水化合物和糖类的比例,提高维生素C和胡萝卜素含量[3],但过量氮肥施用不仅不能显著增产,且易造成植株硝酸盐过量累积,对水环境和大气环境也会造成负面影响[4-6]。因此,在保证产量的同时合理化氮肥用量是莴苣生产中亟需解决的问题,如何在生长期内快速、准确、无损地监测莴苣的氮素营养状况对氮肥合理施用具有重要意义。

      传统的氮素营养诊断方法有土壤测试、植株外部形态观测和化学成分分析法等[7-8],但其测试过程较为复杂,耗时较长,且需要破坏性采样,不适合在生产中快速诊断作物氮素营养状况。近年来,随着遥感技术的发展,多种光谱仪逐渐应用于作物氮素营养诊断与施肥推荐[9]。其中,GreenSeeker光谱仪因其具有主动光源,不易受外界环境干扰,精确性和灵敏性较高而备受关注,可进行大田尺度的作物氮素营养诊断和产量预测[10]。研究表明,作物冠层在可见光波段的反射率主要取决于叶栅栏层中的叶绿素含量,与叶片氮素营养状况及氮含量呈负相关;在近红外区域的反射率主要取决于叶肉细胞及细胞间空腔的结构,与叶片氮素营养状况及氮含量呈正相关[11]。因此,利用GreenSeeker冠层光谱仪可监测作物的氮素营养状况并进行进一步的氮素营养管理。目前,国内外研究已利用GreenSeeker冠层光谱仪进行小麦、玉米、水稻等大田作物的生物量和氮素营养状况预测[12-14]。Li等[15]研究表明,利用GreenSeeker冠层光谱仪测定的植被归一化指数 (normalized difference vegetation index, NDVI) 和比值植被指数 (ratio vegetation index, RVI) 与冬小麦生物量和植株吸氮量间决定系数分别为0.805~0.819和0.508~0.659。在春玉米中,GreenSeeker冠层光谱仪测定的光谱指数对氮营养指数 (NNI) 预测的准确性为33%~55%[16]。但由于蔬菜冠层结构的特殊性、品种的多样性和施肥制度的特殊性,目前GreenSeeker冠层光谱仪在蔬菜作物生物量和氮素营养预测方面研究较少。部分研究报道,GreenSeeker冠层光谱仪可用于监测胡萝卜健康状况,并可在盆栽试验中监测包心菜的氮素营养状况[17-18]。但上述研究未建立定量预测关系或研究尺度较小,无法代表实际生产情况。因此,在田间尺度下,GreenSeeker冠层光谱仪能否应用于莴苣及其他蔬菜作物生物量和氮素营养实时监测,其预测准确性如何仍有待明确。

      因此,以太湖地区常见的茎菜类蔬菜——莴苣为研究对象,利用两年不同施氮水平下的田间试验为基础,利用GreenSeeker光谱仪获取不同生长期 (幼苗期、莲座期、茎形成期、收获期) 的冠层植被指数:植被归一化指数 (normalized difference vegetation index, NDVI) 和比值植被指数 (ratio vegetation index, RVI),分析不同植被指数与莴苣地上部生物量 (aboveground biomass, AGB)、根冠比 (root to shoot ratio, RTS)、植株吸氮量 (plant N uptake, PNU)、植株氮浓度 (plant N concentration, PNC) 等指标间的相关和回归分析关系,以期为莴苣提供一种快速、实时、原位的生物量和氮素营养状况估测方法,并为GreenSeeker光谱仪更好的应用于蔬菜氮素营养管理与推荐施肥提供科学依据。

      • 试验地点位于宜兴市丁蜀镇农业科技示范园 (31°14′N,119°53′E),该地区属于亚热带季风性气候,年平均气温和降雨量分别为15.7℃和1177 mm。供试土壤0—20 cm理化性质为:pH 5.69,EC值0.28 mS/cm,土壤有机质、总氮、硝态氮、速效磷、速效钾含量分别为24.9 g/kg、1.04 g/kg、42 mg/kg、64 mg/kg和63 mg/kg。试验设5个施氮处理,3次重复,小区面积为8.75 m2(3.5 m × 2.5 m),随机区组排列。N1~N5处理化肥氮施用量分别为N 0、108、162、216、270 kg/hm2,各处理氮肥在移栽时和幼苗期分别施用50%,除N0外,其余各小区氮投入量中的有机N肥用量均为78 kg/hm2,磷钾肥施用量分别为P2O5120 kg/hm2和K2O 150 kg/hm2。种植方式采用大棚种植,供试莴苣品种为宜兴地区常见的大紫叶莴笋,试验于2016年 (Year I) 和2017年 (Year II) 秋季至冬季进行。

      • 莴苣冠层光谱特征值 (NDVI和RVI) 通过手持式冠层光谱仪GreenSeekerTM (Trimble Inc., Sunnyvale, CA, USA) 测定。两种冠层光谱特征值计算方式如下:

        $ NDVI = \frac{{{\rho _{NIR}} - {\rho _{Red}}}}{{{\rho _{NIR}} + {\rho _{Red}}}} $

        $ RVI = \frac{{{\rho _{NIR}}}}{{{\rho _{Red}}}} $

        式中,ρNIRρRed 分别代表光谱仪在近红外波段和红光波段的光谱反射率。

        各小区NDVI和RVI测定时间为上午9:00—10:00,测定时光谱仪垂直放置于作物冠层上方60 cm处。测定时手持光谱仪以匀速行进,各小区获取4个测定值并以其平均值作为小区测定结果[19]。各小区莴苣冠层光谱测定示意图如图1所示。

        图  1  小区布置图及莴苣冠层光谱测定示意图

        Figure 1.  Plot layout and canopy spectrum sensing diagram

        各小区光谱值测定及采样日期和相应生长阶段如表1所示。测定光谱值后立即采集各小区植物样品进行进一步分析。各小区采集3株植物,每个时期各采集1组植株样,生育期内共采集4组植株样品。将莴苣整株从土壤中挖出,分为地上部和根系两个部分。称取地上部生物量 (AGB) 和根重并计算根冠比 (RTS),然后在105℃下杀青30 min,继续在75℃下烘干72 h至恒重。随后用植物粉碎机将样品粉碎,浓H2SO4-H2O2消煮并通过凯氏定氮法[20]测定植株氮浓度 (PNC) 并计算植株吸氮量 (PNU)。

        表 1  Year I和Year II莴苣移栽、收获日期及各生长发育时期具体采样日期 (mm-dd)

        Table 1.  Detailed transplanting, harvesting and sensing date in each stage of lettuces in Years I and II

        年份
        Year
        移栽日期 (month-day)
        Transplanting date
        收获日期 (month-day)
        Harvest date
        测定日期 (month-day)
        Sensing date (growth stage)
        Year I09-2812-2510-21 (幼苗期Seedling),11-05 (莲座期Rosette),
        11-16 (茎形成期Stem formation),12-03 (收获期Harvest)
        Year II10-2901-1511-24 (幼苗期Seedling),12-07 (莲座期Rosette),
        12-19 (茎形成期Tuber formation),01-07 (收获期Harvest)
      • 相关性分析运用SPSS软件 (ver. 20.0 for Windows, Chicago, IL, USA),Pearson相关分析进行。回归分析利用Matlab (The MathWorks Inc., Natick, MA, USA) 软件进行。所有图形采用Origin 8.5 (OriginLab Corporation, Northampton, MA, USA) 绘制。

      • Pearson相关分析可衡量定距变量间的线性关系,其计算出的相关性系数绝对值越接近1,表明变量间的相关性越强。冠层光谱测定值与生物量和氮素营养指标的相关性分析结果如表2所示。分析结果表明,除部分时期 (生长后期) 外,植被归一化指数 (NDVI) 和比值植被指数 (RVI) 测定值在生育期内与地上部生物量 (AGB) 和植株吸氮量 (PNU) 均显著正相关,与根冠比 (RTS) 和植株氮浓度 (PNC) 呈负相关。

        表 2  Year I和Year II莴苣不同生长期冠层光谱仪测定的NDVI及RVI与植株生长和N利用指标间的相关系数值 (r)

        Table 2.  Correlation coefficients (r) of the sensed NDVI and RVI with plant growth and nitrogen use index of lettuce in Years I and II

        生长期
        Growth stage
        指标
        Index
        AGBRTSPNUPNCAGBRTSPNUPNC
        Year IYear II
        幼苗期
        Seedling stage
        NDVI0.803**–0.586*0.820**–0.557*0.944**–0.3670.925**–0.608*
        RVI0.820**–0.580*0.831**–0.599*0.953**–0.3610.933**–0.603*
        莲座期
        Rosette stage
        NDVI0.798**–0.703**0.819**–0.732**0.922**–0.582*0.908**–0.638*
        RVI0.862**–0.684**0.874**–0.677**0.958**–0.636*0.914**–0.689*
        茎形成期
        Stem formation stage
        NDVI0.945**–0.844**0.938**–0.720**0.895**–0.470*0.862**–0.586*
        RVI0.942**–0.852**0.938**–0.771**0.936**–0.478*0.889**–0.617*
        收获期
        Harvest stage
        NDVI0.879**–0.748**0.860**–0.617*0.779**–0.512*0.843**–0.328
        RVI0.880**–0.725**0.857**–0.616*0.866**–0.464*0.774**–0.495
        全生育期
        Pooled stages
        NDVI0.752**–0.594**0.840**–0.538**0.857**–0.674**0.918**–0.736**
        RVI0.729**–0.558**0.871**–0.544**0.891**–0.615**0.878**–0.706**
        注(Note):NDVI—植被归一化指数 Normalized difference vegetation index; RVI—比值植被指数 Rratio vegetation index; AGB—地上部生物量 Aboveground biomass; RTS—根冠比 Root to shoot ratio; PNU—植株吸氮量 Plant nitrogen uptake; PNC—植株氮浓度 Plant nitrogen concentration; *—P < 0.05;**—P < 0.01;各时期取样数为 n=15,全生育期为 n=60 Sampling number in each growing stage was n=15 and in pooled stages n=60.

        在4个采样时期,冠层光谱仪测定值 (NDVI和RVI) 与AGB和PNU指标间均呈显著正相关。不同生长期NDVI测定值与AGB和PNU的相关性系数介于0.779~0.945,RVI测定值与AGB和PNU的相关性系数介于0.774~0.953。RTS与冠层光谱仪测定值呈显著负相关,与NDVI的相关性系数介于–0.367~–0.844,与RVI的相关系数介于–0.361~–0.852。PNC与冠层光谱仪测定值NDVI的相关系数介于–0.328~–0.720,与RVI介于–0.495~–0.771。综合4个采样时期,冠层光谱仪测定值与AGB和PNU的相关性系数较各单一时期虽略有下降,仍显著相关 (P < 0.01),相关性系数r介于0.729~0.918;光谱测定值与RTS和PNC间相关性较各单一时期显著上升,综合4个采样时期数据后,相关性系数r分别介于–0.558~–0.674和–0.538~–0.736。综上,冠层光谱仪测定的NDVI值和RVI值与莴苣各生长期AGB和PNU显著正相关,与RTS和PNC显著负相关,因此,冠层光谱仪可用于莴苣氮素营养状况预测。

      • 为定量评估各生长阶段GreenSeeker光谱测定值与莴苣生物量和氮素营养指标间的关系,进一步利用冠层光谱仪进行不同时期氮素营养指标的预测。本研究选择3种不同类型的回归方程 (线性方程、指数方程和幂方程) 来分析各时期光谱测定值与生物量和氮素营养指标间的关系,并通过F检验和决定系数的大小来评估预测方程的准确性,不同类型方程拟合结果如表3所示。对比不同类型预测方程结果表明,方程类型对预测结果无显著影响,预测方程的决定系数以指数方程和幂函数较高。本研究中以决定系数最高的方程类型作为各指标的预测方程,并通过两年测定结果建立各时期生物量和氮素营养状况预测方程如表4所示。

        表 3  不同时期莴苣冠层NDVI和RVI值与其生长和氮素营养指标间3 种回归方程分析对比结果

        Table 3.  The correlation coefficients (R2) between canopy indexes (NDVI and RVI) and plant growth and nitrogen status indexes of lettuce using three different equations at different stages

        年份
        Year
        时期
        Growth stage
        指标
        Index
        AGBRTSPNUPNC
        LEPLEPLEPLEP
        Year I幼苗期
        Seedling stage
        NDVI0.65**0.68**0.66**0.34*0.35*0.35*0.67**0.69**0.68**0.31*0.300.26
        RVI0.67**0.70**0.68**0.34*0.34*0.34*0.69**0.71**0.70**0.36*0.35*0.31*
        莲座期
        Rosette stage
        NDVI0.64**0.76**0.75**0.49*0.51*0.54**0.67**0.77**0.76**0.54**0.54**0.54**
        RVI0.74**0.71**0.75**0.47*0.47*0.50*0.76**0.74**0.77**0.46**0.47**0.52**
        茎形成期
        Stem formation stage
        NDVI0.89**0.89**0.89**0.71**0.69**0.68**0.88**0.88**0.88**0.52**0.50**0.49**
        RVI0.89**0.85**0.88**0.73**0.73**0.71**0.87**0.85**0.87**0.60**0.58**0.53**
        收获期
        Harvest stage
        NDVI0.77**0.77**0.77**0.56**0.58**0.58**0.74**0.73**0.74**0.38*0.38*0.38*
        RVI0.77**0.76**0.77**0.53**0.52**0.57**0.74**0.71**0.73**0.37*0.37*0.38*
        Year II幼苗期
        Seedling stage
        NDVI0.89**0.91**0.92**0.110.110.010.86**0.88**0.88**0.37*0.37*0.36*
        RVI0.91**0.90**0.91**0.120.120.110.87**0.87**0.88**0.36*0.37*0.37*
        莲座期
        Rosette stage
        NDVI0.85**0.92**0.90**0.62**0.66**0.71**0.82**0.83**0.84**0.41*0.40*0.33*
        RVI0.92**0.91**0.93**0.53**0.58**0.64**0.83**0.78**0.82**0.47**0.46**0.42**
        茎形成期
        Stem formation stage
        NDVI0.80**0.87**0.84**0.36*0.37*0.37*0.74**0.79**0.770.34*0.34*0.29
        RVI0.88**0.87**0.88**0.32*0.34*0.36*0.79**0.77**0.79**0.38*0.38*0.35*
        收获期
        Harvest stage
        NDVI0.61**0.60**0.61**0.43*0.44*0.44*0.65**0.59**0.66**0.110.110.10
        RVI0.73**0.77**0.74**0.40*0.42*0.45*0.60**0.55**0.64**0.240.230.23
        注(Note):NDVI—植被归一化指数 Normalized difference vegetation index; RVI—比值植被指数 Rratio vegetation index; AGB—地上部生物量 Aboveground biomass; RTS—根冠比 Root to shoot ratio; PNU—植株吸氮量 Plant nitrogen uptake; PNC—植株氮浓度 Plant nitrogen concentration; L—线性方程 Linear equation:$y = a \times {x_{NDVI}} + b$; E—指数方程 Exponential equation: $y = a \times {e^{b \times {x_{NDVI}}}}$;P—幂函数 Power function equation:$y = a \times {x_{NDVI}}^b$,a、b 为方程的回归参数 a and b are the regression parameters of equations.

        表 4  不同时期莴苣冠层植被指数 (NDVI和RVI) 与植物生长和氮素营养指标间预测方程及决定系数

        Table 4.  The prediction equation and coefficients of determination between canopy indexes (NDVI and RVI) and biomass and nitrogen status indexes (AGB, RTS, PNU and PNC) of lettuce using three different equations at different stages

        年份
        Year
        生长期
        Growth stage
        指标
        Index
        AGBRTSPNUPNC
        预测方程
        Equation
        R2RMSE预测方程
        Equation
        R2RMSE预测方程
        Equation
        R2RMSE预测方程
        Equation
        R2RMSE
        Year I幼苗期
        Seedling stage
        NDVIy = 41.6e4.20x0.6848.8y = 0.096x–0.430.350.021y = 0.19e3.42x0.690.14y = –2.14 x + 4.270.310.27
        RVIy = 33.8e0.79 x0.7047.5y = 0.23x–0.520.340.021y = 0.16e0.65x0.710.13y = –0.44 x + 4.470.360.26
        莲座期
        Rosette stage
        NDVIy = 3.84e7.82 x0.76235y = 0.057 x–1.280.540.029y = 0.026e6.10x0.770.40y = 2.78 x–0.360.540.22
        RVIy = 23.9 x2.080.75240y = 0.33 x–0.760.500.030y = 0.099 x1.680.770.40y = 4.48 x–2.040.520.22
        茎形成期
        Tuber formation
        stage
        NDVIy = 2489x5.870.8956.6y = –0.60 x–0.590.710.016y = 7.54 x5.910.880.18y = –6.68 x + 8.500.520.27
        RVIy = 107 x – 2950.8957.9y = –0.017 x + 0.260.730.015y = 0.32 x–0.900.870.18y = –0.20 x + 4.930.600.25
        收获期
        Harvest stage
        NDVIy = 4832 x7.480.7771.7y = 0.028 x–5.630.580.016y = 17.74 x–11.780.740.21y = 2.77 x–0.540.380.058
        RVIy = 175 x – 6640.7771.9y = 2.71 x–1.520.570.016y = -0.017 x + 0.260.740.21y = 4.28 x–0.140.380.058
        Year II幼苗期
        Seedling stage
        NDVIy = 5655 x2.630.9219.7y = –0.19 x + 0.210.110.028y = 0.030e9.89x0.880.057y = –4.51 x + 4.630.370.30
        RVIy = 10.1 x5.160.9121.0y = –0.061 x + 0.260.120.028y = 0.033 x4.650.880.057y = –6.55e-0.38x0.370.30
        莲座期
        Rosette stage
        NDVIy = 41.2e4.73x0.9247.8y = 0.063 x–0.750.710.019y = 4.38 x1.560.830.20y = –1.83 x + 4.140.410.24
        RVIy = 58.2 x1.820.9346.8y = 0.27 x–0.840.640.021y = 0.65 x–0.500.830.20y = –0.32 x + 4.180.470.24
        茎形成期
        Tuber formation
        stage
        NDVIy = 78.8e3.66x0.8757.1y = 0.23e–1.37x0.370.022y = 0.32e3.39x0.790.25y = –1.63 x + 3.870.340.21
        RVIy = 109 x1.360.8854.4y = 0.21 x–0.550.360.022y = 0.71 x–0.420.790.25y = –0.24 x + 3.790.380.21
        收获期
        Harvest stage
        NDVIy = 1765 x1.090.61198y = 0.092 x–0.370.440.013y = 3.90 x0.830.660.37y = –0.64 x + 3.160.110.25
        RVIy = 417e0.21x0.77150y = 0.16 x–0.280.450.013y = 1.15 x0.570.640.38y = –0.098 x + 3.170.240.23
        Year I &
        Year II
        幼苗期
        Seedling stage
        NDVIy = 766x–65.80.7643.9y = –0.14 x + 0.200.240.025y = 3.12x1.520.850.11y = 2.95 x–0.150.290.30
        RVIy = 169 x–1620.7445.4y = 0.23e–0.20x0.240.025y = 0.59 x–0.610.840.12y = 4.12 x–0.230.290.30
        莲座期
        Rosette stage
        NDVIy = 21.9e5.38x0.73195y = 0.075 x-0.640.480.027y = 0.27e2.77x0.520.46y = –1.15 x + 3.950.310.26
        RVIy = 54.5 x1.630.75188y = 0.22 x–0.530.450.028y = 0.52e0.23x0.550.44y = 3.87 x–0.120.310.26
        茎形成期
        Tuber formation
        stage
        NDVIy = 791 x0.850.55119y = 0.11 x–0.260.140.026y = 2.25 x0.580.330.43y = –0.67 x + 3.490.130.30
        RVIy = 329e0.077x0.49126y = –0.003 x + 0.140.120.026y = 1.24e0.052x0.260.45y = –0.040 x + 3.310.150.30
        收获期
        Harvest stage
        NDVIy = 923 x0.290.07238y = 0.11 x–0.200.120.019y = 2.38 x0.230.080.51y = 2.87 x–0.0580.010.19
        RVIy = 709 x0.0810.02244y = 0.14 x–0.100.100.019y = 1.93 x0.0640.020.52y = 0.0036 x + 2.860.010.19
        注(Note):NDVI—植被归一化指数 Normalized difference vegetation index; RVI—比值植被指数 Rratio vegetation index; AGB—地上部生物量 Aboveground biomass; RTS—根冠比 Root to shoot ratio; PNU—植株吸氮量 Plant nitrogen uptake; PNC—植株氮浓度 Plant nitrogen concentration.

        回归分析结果表明,在Year I,RVI值在幼苗期和茎形成期对氮素营养指标的预测准确性较高,基于RVI测定值的AGB、RTS、PNU、PNC预测方程的R2分别为基于NDVI值的1~1.03、0.97~1.03、0.99~1.03和1~1.22倍;在莲座期和收获期,基于NDVI值的AGB、RTS、PNU、PNC预测方程的准确性分别比基于RVI值的预测方程高0~1.33%、1.75%~8.00%、0和0~17.39%。在Year II,除收获期外,NDVI值对AGB预测的准确性较RVI值提高24.24%,其余时期两种植被指数预测准确性差异不大;对于RTS,NDVI值预测的准确性显著优于RVI值,尤其在莲座期,基于NDVI值预测方程的准确性为RVI方程的1.11倍;对于PNU,两指标预测准确性无明显差异;对于PNC,基于RVI值的预测方程准确性优于NDVI值,基于RVI值的预测方程准确性为NDVI方程的1~2.18倍。综上,对于各时期氮素营养指标预测,NDVI值和RVI值预测准确性差异不大,均可准确预测莴苣氮素营养指标。对比不同生长期冠层光谱仪对莴苣生物量和氮素营养指标的预测方程结果及准确性可知,莲座期和茎形成期预测准确性较高,尤其莲座期最佳,莲座期和茎形成期,冠层光谱测定值对AGB、RTS、PNU、PNC预测准确性分别为0.75~0.93、0.36~0.73、0.77~0.88和0.34~0.60。而幼苗期和收获期预测准确性相对较低,该时期对AGB、RTS、PNU和PNC的预测准确性分别比莲座期和茎形成期低11.03%、35.09%、9.04%和38.33%。利用两年光谱测定数据建立莴苣产量和氮素营养状况统一预测方程结果来看,莲座期为最佳预测时期,对AGB、RTS、PNU、PNC预测准确性分别为75%、48%、55%、31% (表4),因此,可利用冠层光谱仪进行不同年份莴苣产量和氮素营养状况估测分析。

      • 为提高预测结果的科学性和实际应用的便利性,在实际生产中更好的实时预测整个生长期莴苣的氮素营养状况,有必要建立全生育期氮素营养指标预测方程来预测不同年份和不同生长期莴苣的氮素营养状况。

        在全生育期预测方程中,NDVI值和RVI值均与AGB成正相关 (图2)。对比两种冠层植被指数,基于NDVI的线性方程预测效果较好,而幂方程更加适用于RVI-AGB的预测。基于NDVI值预测方程的准确性和RMSE值分别比基于RVI值预测方程高22.7%和低9.16%。因此,NDVI值更适合于全生育期AGB的预测。整个生育期,冠层光谱仪测定的植被指数 (NDVI和RVI) 与RTS呈负相关 (图3)。基于两种植被指数的幂函数的预测准确性较高,两个预测方程的R2值分别为0.43和0.36。对比两种冠层植被指数,基于NDVI值预测方程的准确性和RMSE值分别比基于RVI值预测方程高19.44%和低7.41%。因此,NDVI值对全生育期RTS的预测准确性更高。同样,基于冠层光谱仪的植被指数与全生育期PNU均显著正相关,基于线性方程的NDVI-PNU和幂方程的RVI-PNU预测方程的R2值分别为0.57和0.47(图4)。基于NDVI测定值的PNU预测方程的准确度和RMSE值分别比RVI值高21.3%和低10.34%。从全生育期结果来看,整个生育期冠层光谱仪测定的植被指数 (NDVI和RVI) 与PNC呈负相关 (图5),基于两种植被指数的幂函数的预测准确性较高,两个预测方程的R2值分别为0.26和0.17,因此,NDVI值对全生育期RTS的预测准确性更高。因此,综合不同氮素营养指标,NDVI值的预测准确性优于RVI值,且AGB和PNU预测方程以线性方程为佳,RTS和PNC预测方程以幂函数方程准确性较高。

        图  2  冠层植被指数NDVI(a) 和RVI(b) 与Year I和Year II莴苣地上部生物量 (AGB) 全生育期预测结果分析 (n=120)

        Figure 2.  Relationships between active canopy sensor-based normalized difference vegetation index (NDVI) (a) and ratio vegetation index (RVI) (b) and the aboveground biomass (AGB) of lettuce at all stages in year I and year II (n=120)

        图  3  冠层植被指数NDVI(a) 和RVI(b) 与Year I和Year II莴苣根冠比 (RTS) 全生育期预测结果分析 (n=120)

        Figure 3.  Relationships between active canopy sensor-based normalized difference vegetation index (NDVI) (a) and ratio vegetation index (RVI) (b) and the root to shoot ratio (RTS) of lettuce at all stages in year I and year II (n=120)

        图  4  冠层植被指数NDVI(a) 和RVI(b) 与Year I和Year II莴苣植株吸氮量 (PNU) 全生育期预测结果分析 (n=120)

        Figure 4.  Relationships between active canopy sensor-based normalized difference vegetation index (NDVI) (a) and ratio vegetation index (RVI) (b) and the plant N uptake (PNU) of lettuce at all stages in year I and year II (n=120)

        图  5  冠层植被指数NDVI(a) 和RVI(b) 与Year I和Year II莴苣植株氮浓度 (PNC) 全生育期预测结果分析 (n=120)

        Figure 5.  Relationships between active canopy sensor-based normalized difference vegetation index (NDVI) (a) and ratio vegetation index (RVI) (b) and the plant N concentration (PNC) of lettuce at all stages in year I and year II (n=120)

      • 值得注意的是,尽管基于冠层光谱仪的全生育期氮素营养诊断方程可为实际应用提供有力的支撑。但与各生长时期预测方程相比,全生育期预测方程的准确性显著下降。此外,由于部分时期恶劣天气的干扰和其他农事操作的影响,很难确保各年份间采样时期的一致性。因此,我们尝试在全生育期预测方程中引入移栽天数 (DAT) 来提高预测方程的准确性 (表5)。研究表明,引入DAT并采用二元线性回归分析,光谱测定值与氮素营养状况指标间拟合准确性显著提高 (表6),尤其是地上部生物量 (AGB) 和植株吸氮量 (PNU)。光谱测定值 (NDVI和RVI) 对AGB和PNU预测准确度提高了14.81%~40.43%,均方根误差 (RMSE) 值降低了10.09%~20.69%。同时,基于NDVI值的校正方程准确性最高,对AGB和PNU预测的准确性分别达到62%和71%。但校正方程对RTS的预测准确性未见明显变化,校正前后方程准确性下降13.95%~22.22%。对于植株氮浓度 (PNC),基于NDVI的预测方程准确性较高,校正后方程准确性提升至0.34,较校正前方程准确性提升了30.77%~94.12%,RMSE值下降了–6.25%~–8.82%。

        表 5  引入移栽天数 (DAT) 与冠层植被指数 (NDVI和RVI) 估测莴苣地上部生物量 (AGB)、根冠比 (RTS)、植株吸氮量 (PNU) 和植株氮浓度 (PNC) 的预测方程

        Table 5.  Equations for estimating aboveground biomass (AGB), root to shoot ratio (RTS), plant nitrogen uptake (PNU) and plant nitrogen concentration (PNC) using sensor-based measured normalized difference vegetation index (NDVI) and ratio vegetation index (RVI) and days after transplanting (DAT)

        预测指标 Predicted index预测方程 Predicted equation
        NDVI-AGBAGB = 830 × NDVI + 7.81 × DAT-303
        RVI-AGBAGB = 50.0 × RVI + 9.91 × DAT − 179
        NDVI-RTSRTS = –0.0094 × NDVI − 1.38 × 10–4 × DAT + 0.18
        RVI-RTSRTS = –0.0051 × RVI − 4.23 × 10–4 × DAT + 0.17
        NDVI-PNUPNU = 1.87 × NDVI-0.023 × DAT – 0.65
        RVI-PNUPNU = 0.11 × RVI − 0.028 × DAT + 0.37
        NDVI-PNCPNC = –0.30 × NDVI − 0.011 × DAT + 3.96
        RVI-PNCPNC = –0.0066 × RVI − 0.013 × DAT + 3.91

        表 6  引入移栽天数 (DAT) 校正后莴苣生物量和氮素营养状况预测方程参数变化

        Table 6.  Change of the regression parameters in the model for lettuce biomass and N status predictions before and after modified with transplanting days (DAT)

        预测方程
        Predicted equation
        校正前 Original parameter校正后 Modified parameter校正效果 Change rate (%)
        R2RMSER2RMSER2RMSE
        NDVI-AGB0.542280.6220514.81–10.09
        RVI-AGB0.442510.5721929.55–12.75
        NDVI-RTS0.430.0250.370.026–13.954.00
        RVI-RTS0.360.0270.280.028–22.223.70
        NDVI-PNU0.570.520.710.4224.56–19.23
        RVI-PNU0.470.580.660.4640.43–20.69
        NDVI-PNC0.260.320.340.3030.77–6.25
        RVI-PNC0.170.340.330.3194.12–8.82
        注(Note):校正效果 Change rate (%) = (校正后值 Modified parameter-校正前值 Original parameter) /校正前值 Original parameter × 100.
      • 研究表明,多种方法可用于监测蔬菜作物的氮素营养状况,如叶片氮浓度测定、组织硝酸盐含量测试、SPAD仪和冠层光谱仪等[21]。在这些方法中,冠层光谱仪因其成本较低、无需破坏性采样、准确性和可操作性较高而备受关注[10,22]。但因蔬菜品种多样、冠层结构相对复杂、生育期相对较短,目前利用冠层光谱仪进行莴苣作物氮素营养诊断的研究较少,冠层光谱仪能否在田间尺度下进行莴苣等蔬菜作物生物量及氮素营养状况的准确性预测仍有待研究[19]。本研究通过相关性分析和回归分析表明,GreenSeeker测定的NDVI值和RVI值均与AGB、PNU显著正相关,与RTS和PNC呈负相关。不同生长期冠层植被指数与AGB和PNU指标间的相关性系数介于0.774~0.953;RTS和PNC与冠层光谱仪的相关性系数分别介于–0.361~–0.852和–0.328~–0.771(表2)。在以往研究中,SPAD叶绿素仪对春黄瓜和秋黄瓜产量预测的准确性分别为55%和83%[23]。在马铃薯中,于静等[24]研究发现,在块茎形成期、块茎膨大期和淀粉积累期,马铃薯冠层NDVI测定值与植株吸氮量呈显著相关关系,y = 0.397x0.15(R2 = 0.57)。与Xia等[16]研究结果一致,在作物冠层封闭前的生长阶段,作物生物量的增长速度快于植物对氮素的吸收,并主导了作物的冠层反射率,因此冠层光谱值与AGB间的相关性较PNC更好。此外,氮素营养供应状况对作物根冠生长有明显的调节作用,随着施氮水平的提高,根冠比降低[25]。目前,国内外未见对作物根冠比与NDVI测定值之间的相关关系进行研究,本试验从蔬菜作物的根冠比与NDVI测定值的相关性角度进行分析,在一定程度上反映GreenSeeker光谱仪可表征蔬菜作物的氮素状况,利用NDVI值进行莴苣的氮素营养诊断是可行的。综上,冠层光谱仪对莴苣氮素营养指标预测的准确性与其他仪器基本一致甚至更好,因此,冠层光谱仪可用于莴苣氮素营养诊断。

        本研究中对比分析了GreenSeeker冠层光谱仪测定的两种植被指数 (NDVI值和RVI值) 对莴苣氮素营养状况预测的准确性。研究表明,进行各时期氮素营养指标预测时,不同时期和不同指标预测时两指标表现不同 (表3表4)。在全生育期预测方程中,NDVI值对AGB、RTS、PNU、PNC的预测准确性分别较RVI值高22.7%、19.44%、21.3%和52.9%(图2图5)。综上,NDVI值对莴苣氮素营养状况预测的准确性优于RVI值。该结果与Cao 等[26]研究结果一致,当作物生物量低于2 m2/m2时,NDVI对冬小麦叶面积指数预测的准确性更高。但Yao等[13]的研究结果表明,RVI值对高产水稻产量的预测效果较NDVI值更加敏感。这主要由两种植被指数的内在特征决定的。由于NDVI值的准确性和灵敏性,其对低地表覆盖度下植被指数预测更加敏感,但当地表覆盖度超过80%或者饱和时,NDVI值易发生饱和,因而会造成当作物叶面积指数较高时产量等指标估测结果偏低[27-28]。本研究中选取的作物为莴苣,其种植密度低于一般的水稻、小麦等大田作物,因此,当作物植被覆盖度相对较低时,NDVI值测定的准确性更高。

        已有研究结果表明,不同生长时期会显著影响光谱测定值对作物氮素营养状况的预测效果[29-30]。对比4个不同生长期,幼苗期和收获期预测准确性相对较低,莲座期和茎形成期,冠层光谱测定值对AGB、RTS、PNU和PNC预测准确性分别为0.75~0.93、0.36~0.73和0.77~0.88和0.34~0.60 (表4)。从利用两年测定数据建立莴苣产量和氮素营养状况统一预测方程结果来看,莲座期为最佳预测时期,对AGB、RTS、PNU、PNC预测准确性分别为75%、48%、55%、31% (表4)。该结果与水稻等大田作物中研究结果类似,生长早期冠层光谱测定值易受土壤背景值影响,而生育后期时,光谱测定值饱和等情况会影响预测准确性[10,31]。因此,莲座期和茎形成期为莴苣生物量和氮素营养指标预测的较适宜时期。冠层光谱仪对莴苣全生育期方程对AGB、RTS、PNU、PNC的预测准确性分别为44%~54%、36%~43%、47%~57%和17%~26%,显著低于各时期方程的准确性 (图2图5)。Liu等[32]、Xue等[33]水稻研究中,也发现全生育期产量预测方程较各生育时期预测方程的精度显著下降,其原因可能是作物需氮量随着生长阶段的变化而不断变化,而且基于光谱仪的植被指数在各生长阶段受到的环境和植物等外界条件影响显著不同。但从实际应用考虑,全生育期方程较各生长时期预测更加便捷,因此,本研究通过引入移栽天数DAT对莴苣氮素营养预测方程进行校正。表5表6结果表明,引入DAT后,莴苣氮素营养诊断方程的准确性显著提高,尤其是AGB和PNU预测准确性,分别提高至62%和71%,该结果也与水稻中结论类似,引入DAT参数可提高水稻全生育期SPAD测定值对叶片N浓度预测的精度[34],因此可考虑利用DAT进行氮素营养预测的校正,但DAT校正后RTS预测准确性无显著变化,需进一步考虑其他方法进行RTS的预测校正。

      • 冠层光谱仪测定的NDVI值和RVI值均可预测莴苣的地上部生物量 (AGB)、根冠比 (RTS)、植株氮吸收量 (PNU) 和植株氮含量 (PNC),尤以莲座期和茎生长期预测准确性更佳。采用全生育期预测方程,基于NDVI的预测方程对AGB、RTS、PNU和PNC的预测精度分别为54%、43%、57%和26%,高于基于RVI的预测精度。在全生育期预测方程中引入移栽天数DAT进行校正可进一步提高预测的准确性,使AGB和PNU预测的准确性提高到62%和71%,基本满足莴苣类低地表覆盖度蔬菜的氮素营养管理及施氮推荐。

    参考文献 (34)
    WeChat 关注分享

    返回顶部

    目录

      /

      返回文章
      返回