Canopy SIF synergize with total spectral reflectance optimized by the MDBPSO algorithm to monitor wheat stripe rust
BAI Zong-Fan1, JING Xia,1,*, ZHANG Teng1, DONG Ying-Ying21College of Geometrics, Xi’an University of Science and Technology, Xi’an 710054, Shaanxi, China 2Institute of Remote Sensing and Digital Earth, Chinese Academy of Science, Beijing 100094, China
Abstract In order to extract the characteristic parameters sensitive to wheat stripe rust from total spectral reflectance and improve the operation efficiency and estimation accuracy of the wheat stripe rust remote sensing detection model, this paper improved the traditional discrete binary particle swarm optimization (DBPSO) algorithm from two aspects: inertia weight and particle update method. The modified discrete binary particle swarm optimization (MDBPSO) algorithm was used to select the characteristic parameters for the severity of wheat stripe rust from the total spectral reflectance. The selected characteristic variables were combined with canopy solar-induced chlorophyll fluorescence (SIF) data as independent variables to construct wheat stripe rust estimation model. Random forest (RF) algorithm and back propagation (BP) neural network algorithm were used as model construction method to compare and analyze the accuracy of the feature parameter construction model extracted by the correlation coefficient (CC) analysis method and the DBPSO algorithm. The MDBPSO algorithm had faster convergence speed and higher optimization accuracy than the DBPSO algorithm. The number of iterations before and after the improvement was reduced from 395 to 156. The optimal fitness function (OFV) value decreased from 0.145 to 0.127. When the MDBPSO algorithm was used to select feature variables, the accuracy of the models constructed by the two methods of RF and BP neural networks was higher than that by CC analysis and DBPSO. The validation set determination coefficient (R2V) between the predicted disease index (DI) value and the measured DI value of the RF algorithm was 9% and 3% higher than that of the CC analysis method and the DBPSO algorithm, the validation set root mean square error (RMSEV) was reduced by 28% and 11%, respectively. The R2V between the predicted disease index (DI) value and the measured DI value of the BP neural network algorithm was 13% and 6% higher than that of the CC analysis method and DBPSO algorithm, respectively, and the RMSEV was reduced by 21% and 10% respectively. The MDBPSO algorithm can improve the remote sensing detection accuracy of wheat stripe rust. Among the three feature selection algorithms of MDBPSO, DBPSO, and CC analysis, the accuracy of the model constructed by the RF algorithm was higher than that by the BP neural network algorithm. The R2V between the predicted DI value and the measured DI value of the RF model was at least 7% higher than that of BP neural network algorithm, with an average increase of 9%; RMSEV had reduced by at least 15%, with an average reduction of 20%. The MDBPSO-RF model for wheat stripe rust remote sensing detection constructed by RF method using the characteristic parameters selected by the MDBPSO algorithm as independent variables is an appropriate model for wheat stripe rust remote sensing detection. The research results provide new ideas for further realizing large-area high-precision remote sensing monitoring of crop health. Keywords:total spectral reflectance;modified particle swarm optimization;solar-induced chlorophyll fluorescence;stripe rust of wheat;feature band
PDF (535KB)元数据多维度评价相关文章导出EndNote|Ris|Bibtex收藏本文 本文引用格式 白宗璠, 竞霞, 张腾, 董莹莹. MDBPSO算法优化的全波段光谱数据协同冠层SIF监测小麦条锈病[J]. 作物学报, 2020, 46(8): 1248-1257. doi:10.3724/SP.J.1006.2020.01004 BAI Zong-Fan, JING Xia, ZHANG Teng, DONG Ying-Ying. Canopy SIF synergize with total spectral reflectance optimized by the MDBPSO algorithm to monitor wheat stripe rust[J]. Acta Agronomica Sinica, 2020, 46(8): 1248-1257. doi:10.3724/SP.J.1006.2020.01004
Dong JH, Yang XD, Yang GJ, Wang BS. Inversion of wheat stripe rust disease index based on near ground hyperspectral data J Triticeae Crops, 2016,36:1674-1680 (in Chinese with English abstract). [本文引用: 3]
ZhaoY, JingX, Huang WJ, Dong YY, Li CJ. Comparison of sun-induced chlorophyll fluorescence and reflectance data on estimating severity of wheat stripe rust Spectrosc Spect Anal, 2019,39:2739-2745 (in Chinese with English abstract). [本文引用: 1]
LiangD, LiuN, Zhang DY, Zhao JL, Ding YW. Discrimination of powdery mildew and yellow rust of winter wheat using high-resolution hyperspectra and imageries , 2017,46:50-58. [本文引用: 1]
ZhengQ, Huang WJ, Cui XM, ShiY, Liu LY. New spectral index for detecting wheat yellow rust using sentinel-2 multispectral imagery , 2018,18:868-886. [本文引用: 1]
LiangK, Zhang XX, DingJ, Xu JH, Han DS, Shen MX. Discrimination of wheat scab infection level by Fourier mid-infrared technology combined with sparse representation based classification method Spectrosc Spect Anal, 2019,39:3251-3255 (in Chinese with English abstract). [本文引用: 2]
ShiY, Huang WJ, González-MorenoP, LukeB, Dong YY, ZhengQ, Ma HQ, Liu LY. Wavelet-based rust spectral feature set (WRSFs): a novel spectral feature set based on continuous wavelet transformation for tracking progressive host-pathogen interaction of yellow rust on wheat , 2018,252:1-19. [本文引用: 2]
ShiY, Huang WJ, Zhou XF. Evaluation of wavelet spectral features in pathological detection and discrimination of yellow rust and powdery mildew in winter with hyperspectral reflectance data , 2017,11:1-18. [本文引用: 2]
Jiang JB, Chen YH, Huang WJ, LiJ. Study on hyperspectral remote sensing retrieval models about winter wheat stripe rust severity J Nanjing Agric Univ, 2007,30(3):63-67 (in Chinese with English abstract). [本文引用: 1]
WangL, Qu JJ. Satellite remote sensing applications for surface soil moisture monitoring: a review , 2009,3:237-247. [本文引用: 1]
LiuQ, Gu YL, Wang CC, WangR, LiW, Ma ZH. Canopy hyperspectral features analysis of latent period wheat stripe rust based on discriminant partial least squares J Plant Prot, 2018,45:138-145 (in Chinese with English abstract). [本文引用: 2]
Huang MY, Huang WJ, Liu LY, Huang YD, Wang JH, Zhao CJ, Wan AM. Spectral reflectance feature of winter wheat single leaf infected with stripe rust and severity level inversion Trans CSAE, 2004,20(1):176-180 (in Chinese with English abstract). [本文引用: 1]
LiuQ, Wang CC, WangR, Gu YL, LiW, Ma ZH. Hyperspectral qualitative identification on latent period of wheat stripe rust J Plant Prot, 2018,45:153-160 (in Chinese with English abstract). [本文引用: 1]
YangC, Tan YL, BruzzoneL, Lu LJ, Guan RC. Discriminative feature metric learning in the affinity propagation model for band selection in hyperspectral images , 2017,9:782-798. [本文引用: 2]
KennedyJ, EberhartR. A discrete binary version of the particle swarm algorithm In: IEEE, eds. International Conference on Systems, Man, and Cybernetics. , 1997. pp 4104-4108. [本文引用: 2]
Shen LC, Huo XH, Niu YF. Survey of discrete particle swarm optimization algorithm J Syst Eng Electron, 2008,30:1986-1990 (in Chinese with English abstract). [本文引用: 1]
YinC, YeY, ZhaoH, Jiang YZ, WangH, Shang YZ, Wang JF. Remote sensing of water quality based on HJ-1A HSI imagery with modified discrete binary particle swarm optimization-partial least squares (MDBPSO-PLS) in inland waters: a case in Weishan Lake , 2018,44:21-32. [本文引用: 1]
ZhangJ, Tian HQ, Zhao ZY, Zhang LN, ZhangJ, LiF. Moisture content detection in silage maize raw material based on hyperspectrum and improved discrete particle swarm Trans CSAE, 2019,35(1):285-293 (in Chinese with English abstract). [本文引用: 3]
Wang LA, Zhou XD, Zhu XK, Guo WS. Inverting wheat leaf area index based on HJ-CCD remote sensing data and random forest algorithm Trans CSAE, 2016,32(3):149-154 (in Chinese with English abstract). [本文引用: 1]
Hu XP, Yang ZW, Li ZQ, Deng ZY, Ke CH. BP neural network prediction of wheat stripe rust in Huazhong region Acta Agric Boreali-Occident Sin, 2000,9(3):28-31 (in Chinese with English abstract). [本文引用: 1]
Rodriguez-GailanoV, Mendes MP, Garcia-SoldadooM J, Chica-OlmoM, RibeiroL. Predictive modeling of groundwater nitratepollution using Random Forest and multisource variables related to intrinsic and specific vulnerability: a case study in an agricultural setting (Southern Spain) , 2014,476:189-206. DOI:10.1016/j.scitotenv.2014.01.001URLPMID:24463255 [本文引用: 1] Watershed management decisions need robust methods, which allow an accurate predictive modeling of pollutant occurrences. Random Forest (RF) is a powerful machine learning data driven method that is rarely used in water resources studies, and thus has not been evaluated thoroughly in this field, when compared to more conventional pattern recognition techniques key advantages of RF include: its non-parametric nature; high predictive accuracy; and capability to determine variable importance. This last characteristic can be used to better understand the individual role and the combined effect of explanatory variables in both protecting and exposing groundwater from and to a pollutant. In this paper, the performance of the RF regression for predictive modeling of nitrate pollution is explored, based on intrinsic and specific vulnerability assessment of the Vega de Granada aquifer. The applicability of this new machine learning technique is demonstrated in an agriculture-dominated area where nitrate concentrations in groundwater can exceed the trigger value of 50 mg/L, at many locations. A comprehensive GIS database of twenty-four parameters related to intrinsic hydrogeologic proprieties, driving forces, remotely sensed variables and physical-chemical variables measured in
Ashour looD, Mobasheri MR, HueteA. Developing two spectral disease indices for detection for wheat leaf rust ( Puccinia triticina) , 2014,6:4723-4740. [本文引用: 1]
SongL, GuanterL, GuanK, YouL, HueteA, JuW, ZhangY. Satellite sun-induced chlorophyll fluorescence detects early response of winter wheat to heat stress in the Indian Indo-Gangetic Plains , 2018,24:4023-4037. [本文引用: 1]
JingX, Bai ZF, GaoY, LiuL Y. Wheat stripe rust monitoring by random forest algorithm combined with SIF and reflectance spectrum Trans CSAE, 2019,35(13):154-161. (in Chinese with English abstract). [本文引用: 1]
Chen SY, JingX, Dong YY, Liu LY. Detection of wheat stripe rust using solar-induced chlorophyll fluorescence and reflectance spectral indices Remot Sens Technol Appl, 2019,34:511-520 (in Chinese with English abstract). [本文引用: 1]
SunH, ZhengT, LiuN, ChengM, Li MZ, ZhangQ. Vertical distribution of chlorophyll in potato plants based on hyperspectral imaging Trans CSAE, 2018,34(1):149-156 (in Chinese with English abstract). [本文引用: 1]
Lu JJ, Huang WJ, Jiang JB, Zhang JC. A comparative study on the severity of stripe rust disease in winter wheat estimated by little baud sign and traditional spectral features J Triticeae Crops, 2015,35:1456-1461 (in Chinese with English abstract). [本文引用: 1]
Liu XJ. Retrieval of Solar-Induced Chlorophyll Fluorescence by Remote Sensing PhD Dissertation of University of Chinese Academy of Sciences, Beijing, China, 2016. pp 17-19 (in Chinese with English abstract). [本文引用: 1]
Maier SW, Günther KP, StellmesM. Sun-induced fluorescence: a new tool for precision farming In: McDonald M, Schepers J, Tartly L, eds. Digital Imaging Spectral Techniques: Applications to Precision Agriculture Crop Physiology. , 2003. pp 209-222. [本文引用: 1]
Liu XJ, Liu LY. Improving chlorophyll fluorescence retrieval using reflectance reconstruction based on principal components analysis , 2015,12:1645-1649. [本文引用: 1]
CordonG, Lagorio MG, Paruelo JM. Chlorophyll fluorescence, photochemical reflective index and normalized difference vegetative index during plant senescence , 2016,199:100-110. DOI:10.1016/j.jplph.2016.05.010URLPMID:27302011 [本文引用: 1] The relationship between the Photochemical Reflectance Index (PRI), Normalized Difference Vegetation Index (NDVI) and chlorophyll fluorescence along senescence was investigated in this work. Reflectance and radiance measurements were performed at canopy level in grass species presenting different photosynthetic metabolism: Avena sativa (C3) and Setaria italica (C4), at different stages of the natural senescence process. Sun induced-chlorophyll fluorescence at 760nm (SIF760) and the apparent fluorescence yield (SIF760/a, with a=irradiance at time of measurement) were extracted from the radiance spectra of canopies using the Fraunhofer Line Discrimination-method. The photosynthetic parameters derived from Kautsky kinetics and pigment content were also calculated at leaf level. Whilst stand level NDVI patterns were related to changes in the structure of canopies and not in pigment content, stand level PRI patterns suggested changes both in terms of canopy and of pigment content in leaves. Both SIF760/a and PhiPSII decreased progressively along senescence in both species. A strong increment in NPQ was evident in A. sativa while in S. italica NPQ values were lower. Our most important finding was that two chlorophyll fluorescence signals, PhiPSII and SIF760/a, correlated with the canopy PRI values in the two grasses assessed, even when tissues at different ontogenic stages were present. Even though significant changes occurred in the Total Chlr/Car ratio along senescence in both studied species, significant correlations between PRI and chlorophyll fluorescence signals might indicate the usefulness of this reflectance index as a proxy of photosynthetic RUE, at least under the conditions of this study. The relationships between stand level PRI and the fluorescence estimators (PhiPSII and SIF760/a) were positive in both cases. Therefore, an increase in PRI values as in the fluorescence parameters would indicate higher RUE.
CaoY, Ye YT, Zhao HL, Jiang YZ, WangH, Yan DM. Satellite hyperspectral retrieval of turbidity for water source based on discrete particle swarm and partial least squares Trans CSAM, 2018,49(1):173-182 (in Chinese with English abstract). [本文引用: 1]
Yang HC, Zhang SB, Deng KZ, Du PJ. Research into a feature selection method for hyperspectral imagery using PSO and SVM , 2007,17:473-478. [本文引用: 1]
Zhang HJ, Zhang DM, YanW, Chen ZY, Xin ZY. Throughput optimization based resource allocation mechanism in heterogeneous networks Compute Sci, 2019,46(10):109-115 (in Chinese with English abstract). [本文引用: 1]
Li JL, Dong YY, ShiY, Zhu YN, Huang WJ. Remote sensing monitoring of wheat powdery mildew based on random forest model J Plant Prot, 2018,45:395-396 (in Chinese with English abstract). [本文引用: 1]
YaoX, Yu KY, Yang YJ, ZengQ, Chen ZH, LiuJ. Estimation of forest leaf area index based on random forest model and remote sensing data Trans CSAM, 2017,48(5):159-166 (in Chinese with English abstract). [本文引用: 1]
ErshatA, MamatS, BaidengshaM, An SQ, Ma CY. Estimation of leaf chlorophyll content in cotton based on the random forest approach Acta Agron Sin, 2019,45:81-90 (in Chinese with English abstract). [本文引用: 1]