收稿日期: 2009-01-08
修回日期: 2009-05-07
网络出版日期: 2009-07-10
基金资助
国家重点基础研究发展计划项目“基于地表参数知识库的遥感综合定量反演”（编号：2007CB714407）和“被动遥感反射、辐射机理与参数反演”（编号：2007CB714403）;中国科学院西部行动计划（二期）项目“黑河流域遥感—地面观测同步试验与综合模拟平台建设”（编号：KZCX2-XB2-09）联合资助.
Estimating Time Series Leaf Area Index Based on Recurrent Neural Networks
Received date: 2009-01-08
Revised date: 2009-05-07
Online published: 2009-07-10
叶面积指数LAI是众多气象、环境、农业等模型的关键输入参数。尽管具有多个传感器的全球LAI产品已经相继发布，但是由于受反演方法的局限性以及反射率产品质量的影响，这些由单一传感器数据得到的LAI产品在时间上表现出一定的不连续性，这与自然生长植被的LAI变化规律不能一致。而神经网络在对复杂的、非线性数据的模式识别能力方面具有出色的表现。如在3层神经网络中，只要对隐层采用非线性递增映射函数，输出层采用线性映射函数，就可以用于对任意连续函数进行逼近。对于具有相同植被覆盖类型的同一地点多年的LAI数据，在无自然灾害和人为破坏的前提下，可以构成一个非线性的、连续的时间序列。通过融合MODIS和VEGETATION两种传感器产品，在利用相同植被类型的LAI时间序列来建立自回归神经网络，即NARX神经网络的同时，引入红、近红外和短波红外3个波段上时间序列的反射率以及相应的太阳天顶角、观测天顶角和相对方位角作为NARX神经网络的外部输入变量，并最终达到估算时间序列LAI的目的。验证结果表明，NARX神经网络非常适用于时间序列的LAI估算，并且其预测的LAI比原始的MODIS LAI在时间序列上表现的更连续和平滑。因此，该方法在改进典型植被类型的LAI遥感数据产品质量方面具有一定的应用价值。
屈永华 , 张立新 , 梁顺林 , 王锦地 , 柴琳娜 . 基于自回归神经网络的时间序列叶面积指数估算[J]. 地球科学进展, 2009 , 24(7) : 756 -768 . DOI: 10.11867/j.issn.1001-8166.2009.07.0756
Leaf Area Index is a key parameter of many meteorological, environmental and agricultural models. At present, global LAI products of several sensors have been released. However, due to the limitations of the retrieval methods and the qualities of the reflectance products, the released LAI products, generally retrieved from a single sensor data, have been found maybe not continuous in time series and can not characterize the natural growing vegetation well. Many research results in different domain have found that the neural network has nearly perfect performance in recognizing patterns in complex, nonlinear data, e.g., a three-layered neural network could be used to approximate any continuous function if nonlinear increasing function and linear function are used respectively in its hidden layers and output layer. As to a span of the LAI values of an individual vegetation type in a same area, if there were no natural disasters or human destructions, they should present as a nonlinear continuous time series. In this study, by fusing the MODIS and VEGETATION products, time series LAI were used to construct recurrent neural networks, namely the NARX neural network, for six typical vegetation types. Meanwhile, time series reflectances in red, near infrared and shortwave infrared bands and the corresponding sun-viewing angles were introduced into the NARX neural networks as the exogenous inputs to estimate time series LAI. The validation results show that the NARX neural network is competent to estimate time series LAI and the predicted LAI of it is more continuous and smoother than that of the original MODIS LAI in time series. Thus the proposed method may be helpful to improve the quality of LAI products of the typical vegetation types.
Key words： Time series; Leaf area index; Data fusion; NARX neural network; MODIS; VEGETATION
[1] Bonan G B. Importance of leaf area index and forest type when estimating photosynthesis in boreal forests[J].Remote Sensing of Environment,1993,43:303-314.
[2] Bonan G B. The land surface climatology of the NCAR land surface model coupled to the NCAR community climate model[J].Journal of Climate,1998, 11: 1 307-1 326.
[3] Bonan G B, Oleson K W, Vertenstein M, et al.The land surface climatology of the community land model coupled to the NCAR community climate model[J].Journal of Climate, 2002, 15:3 123-3 149.
[4] Fang H, Liang S, Kuusk A. Retrieving leaf area index using a genetic algorithm with a canopy radiative transfer model[J].Remote Sensing of Environment,2003, 85: 257-270.
[5] Sellers P, Schimel D. Remote sensing of the land biosphere and biogeochemistry in the EOS era: Science priorities, methods and implementation-EOS land biosphere and biogeochemical cycles panels[J].Global and Planetary Change,1993, 7: 279-297.
[6] Kergoat L. A model for hydrological equilibrium of leaf area index on a global scale[J].Journal of Hydrology, 1998,(212/213): 268-286.
[7] Sakamoto T, Yokozawa M, Toritani H, et al. A crop phenology detection method using time-series MODIS data[J].Remote Sensing of Environment,2005, 96: 366-374.
[8] Liu R, Chen J M, Liu J, et al. Application of a new leaf area index algorithm to China′s landmass using MODIS data for carbon cycle research[J].Remote Sensing of Environment, 2007, 85: 649-658.
[9] Verger A, Baret F, Weiss M. Performances of neural networks for deriving LAI estimates from existing CYCLOPES and MODIS products[J].Remote Sensing of Environment,2008, 112: 2 798-2 803.
[10] Liang S. Quantitative Remote Sensing of Land Surface[M]. New York: John Wiley and Sons, Inc, 2004.
[11] Walthall C, Dulaney W, Anderson M, et al. A comparison of empirical and neural network approaches for estimating corn and soybean leaf area index from Landsat ETM+ imagery[J].Remote Sensing of Environment, 2004, 92: 465-474.
[12] Qu Y, Wang J, Wan H,et al. A bayesian network algorithm for retrieving the characterization of land surface vegetation[J].Remote Sensing of Environment, 2008, 112: 613-622.
[13] Walthall C L. A study of reflectance anisotropy and canopy structure using a simple empirical model[J]. Remote Sensing of Environment,1997, 61: 118-128.
[14] Brown L, Chen J M, Leblanc S G, et al. A shortwave infrared modification to the simple ratio for LAI retrieval in boreal forests an image and model analysis[J].Remote Sensing of Environment,2000, 71: 16-25.
[15] Jacquemoud S, Baret F. A model of leaf optical properties spectra[J].Remote Sensing of Environment,1990, 34: 75-91.
[16] Myneni R B, Nemani R R, Running S W. Estimation of global leaf area index and absorbed par using radiative transfer models[J].IEEE Transactions on Geoscience and Remote Sensing,1997, 35: 1 380-1 393.
[17] Dawson T P, Curran P J, Plummer S E. LIBERTY-modeling the effects of leaf biochemical concentration on reflectance spectra[J].Remote Sensing of Environment,1998, 65: 50-60.
[18] Kuusk A. A two-layer canopy reflectance model[J].Journal of Quantitative Spectroscopy & Radiative Transfer, 2001, 71: 1-9.
[19] Verstraete M M, Pinty B, Myneini R B. Potential and limitations of information extraction on the terrestrial biosphere from satellite remote sensing[J].Remote Sensing of Environment, 1996, 58: 201-214.
[20] Noble P A, Tribou E H. Neuroet: An easy-to-use artificial neural network for ecological and biological modeling[J].Ecological Modelling,2007, 203: 87-98.
[21] Fang H, Liang S. Retrieving leaf area index with a neural network method: Simulation and validation[J].IEEE Transactions on Geoscience and Remote Sensing,2003, 41: 2 052-2 062.
[22] Gong P, Wang S, Liang S. Inverting a canopy reflectance model using a neural network[J].International Journal of Remote Sensing, 1999, 20: 111-122.
[23] Trombetti M, Riano D, Rubio M A, et al. Multi-temporal vegetation canopy water content retrieval and interpretation using artificial neural networks for the continental USA[J].Remote Sensing of Environment,2008, 112: 203-215.
[24] Funahashi K.On the approximate realization of continuous mappings by neural networks[J].Neural Networks, 1989,2:183-192.
[25] Cotter N E, Guillerm T J. The DMAC and a theorem of Kolmogorov[J].Neural Networks,1992, 5: 221-228.
[26] Kürková V. Kolmogorov′s theorem and multilayer neural networks[J].Neural Networks,1992, 5: 501-506.
[27] Katsuura H, Sprecher D A. Computational aspects of Kolmogorov′s superposition theorem[J].Neural Networks,1994, 7: 455-461.
[28] Zhang G P.A neural network ensemble method with jittered training data for time series forecasting[J]. Information Sciences, 2007, 177: 5 329-5 346.
[29] Assaad M, Bon R, Cardot H. A new boosting algorithm for improved time-series forecasting with recurrent neural networks[J].Information Fusion,2008, 9: 41-55.
[30] Yu L,Wang S, Lai K K. A neural-network-based nonlinear metamodeling approach to financial time series forecasting[J].Applied Soft Computing, 2007, 9(2): 563-574.
[31] Hussain A J, Knowles A, Lisboa P J G, et al. Financial time series prediction using polynomial pipelined neural networks[J].Expert Systems with Applications, 2008, 35: 1 186-1 199.
[32] Zhang G P, Qi M. Neural network forecasting for seasonal and trend time series[J].European Journal of Operational Research, 2005, 160: 501-514.
[33] Chen Y, Zhang W, Yong B. Retrieving leaf area index using a neural network based on classification knowledge[J].Acta Ecologica Sinica, 2007, 27: 2 785-2 793.
[34] Sun W, Liang S, Xu G, et al. Mapping plant functional types from MODIS data using multisource evidential reasoning[J].Remote Sensing of Environment, 2008, 112: 1 010-1 024.
[35] Pisek J, Chen J M. Comparison and validation of MODIS and VEGETATION global LAI products over four BigFoot sites in north America[J].Remote Sensing of Environment,2007, 109: 81-94.
[36] Weiss M, Baret F, Garrigues S, et al. LAI and fAPAR CYCLOPES global products derived from VEGETATION. Part 2: Validation and comparison with MODIS collection 4 products[J].Remote Sensing of Environment,2007, 110: 317-331.
[37] Knyazikhin Y, Martonchin J V, Diner D J, et al. Estimation of vegetation canopy leaf area index and fraction of absorbed photosynthetically active radiation from atmosphere-corrected MISR data[J].Journal of Geophysical Research, 1998, 103: 32 239-32 256.
[38] Baret F, Hagolle O, Geiger B, et al. LAI, fAPAR and fCover CYCLOPES global products derived from VEGETATION. Part 1: Principles of the algorithm[J].Remote Sensing of Environment, 2007, 110: 275-286.
[39] Demuth H, Beal M, Hagan M. User′s Guide: Neural network toolbox TM 6[Z]. Natick: The Mathworks, MA, 2008.
[40] Wang Q, Tenhunen J, Dinh N Q, et al. Evaluation of seasonal variation of MODIS derived leaf area index at two European deciduous broadleaf forest sites[J].Remote Sensing of Environment,2005, 96: 475-484.
/
| 〈 |
|
〉 |
甘公网安备62010202000687
