Commercial tree species discrimination using airborne AISA Eagle hyperspectral imagery and partial least squares discriminant analysis (PLS-DA) in KwaZulu-Natal, South Africa

Abstract Discriminating commercial tree species using hyperspectral remote sensing techniques is critical in monitoring the spatial distributions and compositions of commercial forests. However, issues related to data dimensionality and multicollinearity limit the successful application of the technology. The aim of this study was to examine the utility of the partial least squares discriminant analysis (PLS-DA) technique in accurately classifying six exotic commercial forest species (Eucalyptus grandis, Eucalyptus nitens, Eucalyptus smithii, Pinus patula, Pinus elliotii and Acacia mearnsii) using airborne AISA Eagle hyperspectral imagery (393–900 nm). Additionally, the variable importance in the projection (VIP) method was used to identify subsets of bands that could successfully discriminate the forest species. Results indicated that the PLS-DA model that used all the AISA Eagle bands (n = 230) produced an overall accuracy of 80.61% and a kappa value of 0.77, with user’s and producer’s accuracies ranging from 50% to 100%. In comparison, incorporating the optimal subset of VIP selected wavebands (n = 78) in the PLS-DA model resulted in an improved overall accuracy of 88.78% and a kappa value of 0.87, with user’s and producer’s accuracies ranging from 70% to 100%. Bands located predominantly within the visible region of the electromagnetic spectrum (393–723 nm) showed the most capability in terms of discriminating between the six commercial forest species. Overall, the research has demonstrated the potential of using PLS-DA for reducing the dimensionality of hyperspectral datasets as well as determining the optimal subset of bands to produce the highest classification accuracies.

[1]  Roberta E. Martin,et al.  Brightness-normalized Partial Least Squares Regression for hyperspectral data , 2010 .

[2]  D. Roberts,et al.  Mapping two Eucalyptus subgenera using multiple endmember spectral mixture analysis and continuum-removed imaging spectrometry data , 2011 .

[3]  Onisimo Mutanga,et al.  A comparison of regression tree ensembles: Predicting Sirex noctilio induced water stress in Pinus patula forests of KwaZulu-Natal, South Africa , 2010, Int. J. Appl. Earth Obs. Geoinformation.

[4]  I. M. Hasmadi,et al.  Analysis of crown spectral characteristic and tree species mapping of tropical forest using hyperspectral imaging. , 2010 .

[5]  Jean-Michel Roger,et al.  Discrimination of Corn from Monocotyledonous Weeds with Ultraviolet (UV) Induced Fluorescence , 2011, Applied spectroscopy.

[6]  S. Wold,et al.  PLS-regression: a basic tool of chemometrics , 2001 .

[7]  Susan L. Ustin,et al.  Spectral and physiological uniqueness of perennial pepperweed (Lepidium latifolium) , 2006, Weed Science.

[8]  Solomon Tesfamariam,et al.  Earthquake induced damage classification for reinforced concrete buildings , 2010 .

[9]  Paul Geladi,et al.  The Importance of Balanced Data Sets for Partial Least Squares Discriminant Analysis: Classification Problems Using Hyperspectral Imaging Data , 2011 .

[10]  Candace Berrett,et al.  Spatial characteristics of the difference between MISR and MODIS aerosol optical depth retrievals over mainland Southeast Asia , 2009 .

[11]  David Riaño,et al.  Retrieval of Fresh Leaf Fuel Moisture Content Using Genetic Algorithm Partial Least Squares (GA-PLS) Modeling , 2007, IEEE Geoscience and Remote Sensing Letters.

[12]  John A. Gamon,et al.  Assessing leaf pigment content and activity with a reflectometer , 1999 .

[13]  D. Roberts,et al.  Comparison of various techniques for calibration of AIS data , 1986 .

[14]  J. Schjoerring,et al.  Reflectance measurement of canopy biomass and nitrogen status in wheat crops using normalized difference vegetation indices and partial least squares regression , 2003 .

[15]  L. Kumar,et al.  Estimating and mapping grass phosphorus concentration in an African savanna using hyperspectral image data , 2007 .

[16]  Wei-Yin Loh,et al.  Classification and regression trees , 2011, WIREs Data Mining Knowl. Discov..

[17]  J. V. van Aardt,et al.  Spectral–age interactions in managed, even‐aged Eucalyptus plantations: application of discriminant analysis and classification and regression trees approaches to hyperspectral data , 2008 .

[18]  Clayton C. Kingdon,et al.  Remote sensing of the distribution and abundance of host species for spruce budworm in Northern Minnesota and Ontario , 2008 .

[19]  John A. Silander,et al.  Delineating forest canopy species in the northeastern united states using multi-temporal TM imagery , 1998 .

[20]  M. Vohland,et al.  Remote Sensing Techniques for Forest Parameter Assessment: Multispectral Classification and Linear Spectral Mixture Analysis , 2007 .

[21]  O. Mutanga,et al.  Spectral discrimination of papyrus vegetation (Cyperus papyrus L.) in swamp wetlands using field spectrometry , 2009 .

[22]  Bernhard Lendl,et al.  Differentiation of walnut wood species and steam treatment using ATR-FTIR and partial least squares discriminant analysis (PLS-DA) , 2010, Analytical and bioanalytical chemistry.

[23]  T. Newby,et al.  The South African land-cover characteristics database: a synopsis of the landscape. , 2000 .

[24]  Roberta E. Martin,et al.  Leaf chemical and spectral diversity in Australian tropical forests. , 2009, Ecological applications : a publication of the Ecological Society of America.

[25]  G. A. Blackburn,et al.  Hyperspectral remote sensing of plant pigments. , 2006, Journal of experimental botany.

[26]  Mary E. Martin,et al.  Determining Forest Species Composition Using High Spectral Resolution Remote Sensing Data , 1998 .

[27]  Sofía Valenzuela,et al.  Multivariate strategies for classification of Eucalyptus globulus genotypes using carbohydrates content and NIR spectra for evaluation of their cold resistance , 2008 .

[28]  Andrew K. Skidmore,et al.  Estimation of green grass/herb biomass from airborne hyperspectral imagery using spectral indices and partial least squares regression , 2007, Int. J. Appl. Earth Obs. Geoinformation.

[29]  Terry Caelli,et al.  Hyperspectral discrimination of tropical dry forest lianas and trees: Comparative data reduction approaches at the leaf and canopy levels , 2007 .

[30]  Terry Caelli,et al.  Discrimination of lianas and trees with leaf-level hyperspectral data , 2004 .

[31]  M. C. Rutherford,et al.  The vegetation of South Africa, Lesotho and Swaziland. , 2006 .

[32]  Nathalie Dupuy,et al.  Comparison of PLS1-DA, PLS2-DA and SIMCA for classification by origin of crude petroleum oils by MIR and virgin olive oils by NIR for different spectral regions , 2011 .

[33]  M. Thompson,et al.  A standard land-cover classification scheme for remote-sensing applications in South Africa , 1996 .

[34]  M. Tenenhaus,et al.  Prediction of clinical outcome with microarray data: a partial least squares discriminant analysis (PLS-DA) approach , 2003, Human Genetics.

[35]  B. Turner,et al.  Estimating foliage nitrogen concentration from HYMAP data using continuum, removal analysis , 2004 .

[36]  Michael E. Schaepman,et al.  A review on reflective remote sensing and data assimilation techniques for enhanced agroecosystem modeling , 2007, Int. J. Appl. Earth Obs. Geoinformation.

[37]  K. R. Reddy,et al.  Narrow-waveband reflectance ratios for remote estimation of nitrogen status in cotton. , 2002, Journal of environmental quality.

[38]  Andrew K. Skidmore,et al.  Mapping beech (Fagus sylvatica L.) forest structure with airborne hyperspectral imagery , 2009, Int. J. Appl. Earth Obs. Geoinformation.

[39]  C. S. Jha,et al.  Forest fragmentation and its impact on species diversity: an analysis using remote sensing and GIS , 2005, Biodiversity & Conservation.

[40]  Anne-Laure Boulesteix,et al.  Partial least squares: a versatile tool for the analysis of high-dimensional genomic data , 2006, Briefings Bioinform..

[41]  A. Gitelson,et al.  Optical Properties and Nondestructive Estimation of Anthocyanin Content in Plant Leaves¶ , 2001, Photochemistry and photobiology.

[42]  Sonia Czarnes,et al.  Variable selection in near infrared spectra for the biological characterization of soil and earthworm casts , 2008 .

[43]  M. Lawes,et al.  Forest landscape pattern in the KwaZulu–Natal midlands, South Africa: 50 years of change or stasis? , 2004 .

[44]  Philippe Lagacherie,et al.  Continuum removal versus PLSR method for clay and calcium carbonate content estimation from laboratory and airborne hyperspectral measurements , 2008 .

[45]  P. Gong,et al.  Conifer species recognition: Effects of data transformation , 2001 .

[46]  Brian R. Sturtevant,et al.  Estimation of forest structural parameters using 5 and 10 meter SPOT-5 satellite data , 2009 .

[47]  Russell Turner,et al.  Classifying Eucalyptus forests with high spatial and spectral resolution imagery: an investigation of individual species and vegetation communities , 2005 .

[48]  Bjoern H. Menze,et al.  A comparison of random forest and its Gini importance with standard chemometric methods for the feature selection and classification of spectral data , 2009, BMC Bioinformatics.

[49]  R. Lucas,et al.  Classification of Australian forest communities using aerial photography, CASI and HyMap data , 2008 .

[50]  Lorenzo Bruzzone,et al.  The role of spectral resolution and classifier complexity in the analysis of hyperspectral images of forest areas. , 2007 .

[51]  Lin Li,et al.  Retrieval of vegetation equivalent water thickness from reflectance using genetic algorithm (GA)-partial least squares (PLS) regression , 2008 .

[52]  R. Wynne,et al.  Examining pine spectral separability using hyperspectral data from an airborne sensor: An extension of field‐based results , 2007 .

[53]  Onisimo Mutanga,et al.  Examining the utility of random forest and AISA Eagle hyperspectral image data to predict Pinus patula age in KwaZulu-Natal, South Africa , 2011 .

[54]  D. Tewari.  Is commercial forestry sustainable in South Africa? The changing institutional and policy needs , 2001 .

[55]  A. R. Griffin,et al.  Discrimination between seedlings of Eucalyptus globulus, E. nitens and their F1 hybrid using near-infrared reflectance spectroscopy and foliar oil content , 2008 .

[56]  D. Roberts,et al.  Hyperspectral discrimination of tropical rain forest tree species at leaf to crown scales , 2005 .

[57]  R. Leardi,et al.  Genetic algorithms applied to feature selection in PLS regression: how and when to use them , 1998 .

[58]  S. Franklin,et al.  Remote sensing of forest environments : concepts and case studies , 2003 .

[59]  Christine Stone,et al.  Chlorophyll content in eucalypt vegetation at the leaf and canopy scales as derived from high resolution spectral data. , 2003, Tree physiology.

[60]  C. Jun,et al.  Performance of some variable selection methods when multicollinearity is present , 2005 .

[61]  A. Brenning.  Benchmarking classifiers to optimally integrate terrain analysis and multispectral remote sensing in automatic rock glacier detection , 2009 .

[62]  Fuan Tsai,et al.  Spectrally segmented principal component analysis of hyperspectral imagery for mapping invasive plant species , 2007 .

[63]  Lori M. Bruce,et al.  Why principal component analysis is not an appropriate feature extraction method for hyperspectral data , 2003, IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477).