409 IIUM Engineering Journal, Vol. 14, No. 2, 2013 Tawfik et al. 133 DISEASE DETECTION FROM FIELD SPECTROMETER DATA O. H. TAWFIK 1 , H. Z. MOHD SHAFRI 1 AND A. A. MOHAMMED 2 1 Department of Civil Engineering, Faculty of Engineering, University Putra Malaysia UPM, 43400 Serdang, Selangor, Malaysia, 2 Department of Civil and Structural Engineering, Faculty of Engineering and Built Environment, National University of Malaysia UKM, 43600 Bangi, Selangor Darul Ehsan, Malaysia. aliukm@yahoo.com ABSTRACT: Oil palm plants have been planted in a large scale of areas. However, Ganoderma disease has been recognized and diagnosed in the oil palm plants that has infected almost a half of the oil palm plants in Malaysia. To deal with this problem, the use of vegetation indices analysis on hyper spectral field data, this paper examines the ability of spectral data in identifying the stages of Ganoderma disease. The favourable result will be helpful to control the spreading of the diseases. By using vegetation indices, the oil palm plants could be classified into three categories, namely: 1 (T1 healthy), 2 (T2 semi healthy) and 3 (T3 severe damage). The results show that the best vegetation index is the Modified Red Edge Simple Ratio (MSR705) among the vegetation indices to identify the oil palm health stages. Moreover, it has been observed that the index of Narrowband greenness VIs has been exhibited an acceptable outcome in differentiating between the oil palm plant stage 1 (T1 healthy) and stage 2 (T2 semi healthy). ABSTRAK: Tanaman kelapa sawit ditanam secara meluas. Penyakit ganoderma dikenali dan didiagnosikan menjangkiti hampir separuh tanaman kelapa sawit di Malaysia. Untuk mengawal penyakit ini daripada merebak, analisis indeks tanaman dijalankan ke atas data kawasan spektrum melampau di mana keupayaan data ini diuji dalam membezakan peringkat-peringkat penyakit Ganoderma terhadap tanaman kelapa sawit. Dengan menggunakan indeks tanaman, kelapa sawit dapat diklasifikasikan kepada 1 (T1 sihat), 2 (T2 separa sihat) dan 3 (T3 rosak); kelas tanaman dengan tepat. Keputusan menunjukkan indeks tanaman terbaik sebagai Modified Red Edge Simple Ratio (MSR705) yang merupakan indeks tanaman dalam membezakan peringkat kesihatan kelapa sawit. Adalah didapati pengubahsuaian terhadap indeks Modified Red Edge Simple Ratio (MSR705) yang juga indeks Jalur Sempit Hijau VI telah memberikan keputusan yang munasabah dalam membezakan peringkat tanaman kelapa sawit peringkat 1 (T1 sihat) dan peringkat 2 (T2 separa sihat). KEYWORDS: Oil palm; Vegetation indices; Spectrometer; MSR; Ganoderma. 1. INTRODUCTION Approximately, 70 percent of the Earth’s land surface is covered with vegetation [1]. Furthermore, vegetation provides a basic foundation for all living beings and it is one of the most important components of the ecosystem [1, 2]. Dealing with disease problem in oil palm plantation involves a variety of curative measures in which disease detection and mapping play a central role. Hyperspectral IIUM Engineering Journal, Vol. 14, No. 2, 2013 Tawfik et al. 134 remote sensing data offer a better chance of disease detection [3]. Vegetation indices are widely used for the estimation of crop and vegetation variables by using visible and Near Infrared Regions (NIR) of the electromagnetic spectrum. Healthy plant typically displays very low reflectance and transmittance in visible region and very high reflectance and transmittance in NIR [3]. Sixteen vegetation indices and four modifications were tested on hyper spectral field data. Thus the presence of stresses in oil palm trees will be associated with the chlorophyll absorption in reflectance and the normalized pigment chlorophyll vegetation indexes which will be showing a loss of chlorophyll pigment compared to healthy oil palm plants [3]. Basal stem rot in oil palm is caused by Ganoderma boninense and it is the most severe fungal disease of oil palm in Malaysia. It has the ability to infect oil palms from as young as 12-24 months [4] to over 24 years after field planting [5]. High incidences of this disease have been reported in oil palms planted on coastal soil and peat [6-8]. The incidence of Ganoderma inland soils was relatively low and confined to waterlogged area [6]. Ganoderma is a white rot fungus. The organism causes economic loss of oil palm (OP) in various regions around the world including Southeast Asia [9], where the current author has had considerable experience of the crop disease. The basic premise of this review is that it is important for the control of Ganoderma disease to consider it specifically as a white rot fungus. This can be integrated with other approaches [10, 11]. The term white-rot is derived from the fungus degrading specifically the lignin component of wood while leaving white cellulose exposed. Typically the fungus may attack already weakened oil palm plants as Ganoderma seldom seriously infects undamaged trees. A classic example is Ganoderma adspersum, which causes [12]. 2. MATERIALS AND METHODS The data that have been used in this study is taken by the APOGEE spectroradiometer (300 - 1000 nm) with spectral resolution of 0.5 nm. These data were offered from nursery managed by Malaysian Palm Oil Board (MPOB), Bangi on November 2007. The measurement taken from 1 healthy and 2 unhealthy oil palms leaves, and classified as T1 (Healthy), T2 (Light Symptom) and T3 (Severe Symptom). Each wavelength in the data has 24 samples. Sixteen vegetation indices were applied for 24 samples at three stages to extract the results of vegetation indices to know if the results are within the green vegetation range or not and also to find out which index can exhibit the best way to differentiate between the oil palm disease stages. Figure 1 shows the overall flow of the steps that had been implemented in this study [6, 8, 12]. 2.1 Determine the Best Vegetation Vegetation Indices (VIs) are combinations of surface reflectance at two or more wavelengths designed to highlight a particular property of vegetation [13, 14]. They are derived using the reflectance properties of vegetation described in Plant Foliage. Each of the VIs is designed to accentuate a particular vegetation property. Sixteen equations were applied for 24 samples at three stages to extract the results of vegetation indices to know if the results are within the green vegetation range or not and also to find out generally, this study is comprised of two important parts. The first determined the best vegetation index IIUM Engineering Journal, Vol. 14, No. 2, 2013 Tawfik et al. 135 Fig. 1: Flowchart of methodology. and the second was modified the best vegetation index [15, 11, 16]. The Matlab software package ver.6.5 was used to determine the results of the vegetation indices Broadband Greenness (4 indices) [12, 17, 14, 15]: NTR RED NTR RED P P NDV I P P − = + (1) NTR RED P SR P = (2) IIUM Engineering Journal, Vol. 14, No. 2, 2013 Tawfik et al. 136 2.5 6 7.5 1 NTR RED NTR RED BLUB P P EV I P P  − =   + − +  (3) (2 ) (2 ) NTR RED BLUB NTR RED BLUB P P P A RV I P P P − − = + − (4) which index can exhibit the best way to differentiate between the oil palm disease stages. The indices are grouped into categories that calculate similar properties. The categories and indices are Narrowband Greenness (6 indices) [10, 16, 18]: 750 705 705 750 705R P P NDV I P P − = + (5) 750 460 705 705 460 P P mSR P P − = + (6) 750 705 705 705 705 460 2 P P mNDV I P P P − = + − (7) 740 720 1 P V OG P = (8) 740 747 715 726 2 P P V OG P P − = + (9) 734 747 715 720 3 P P V OG P P − = + (10) Light Use Efficiency (2 indices) [19-21]: 910 460 553 460 4 P P Modification P P − = + (11) 800 460 800 680 P P SIPI P P − = + (12) Leaf Pigments (4 indices) [22, 23]: 510 550 1 1 1CRI P P     = −        (13) 510 100 1 1 2CRI P P     = −        (14) IIUM Engineering Journal, Vol. 14, No. 2, 2013 Tawfik et al. 137 550 100 1 1 1A R I P P     = −        (15) 800 550 100 1 1 2ARI p P P      = −           (16) The Broadband Greenness equations represent the surface reflectance in an image band with a center wavelength as follows: PNIR = 800 nm, PRED = 680 nm and PBLUE = 450 nm. 2.2 Modified Best Vegetation Index The best vegetation index can only differentiate between T1 (healthy plant) and T3 (severe damage) and cannot differentiate between T1 (healthy plant) and T2 (semi healthy). So to make this index able to differentiate between T1 (healthy plant) and T2 (semi healthy) a modification on this index is needed. Using the default best index to differentiate between T1 (healthy plant) and T2 (semi healthy) is not detectable and it is difficult to specify the right stage for the plant. By randomly selecting a different wavelengths and substituting in the default best index, it can be shown that a four modified best indices can be obtained. These four indices can differentiate clearly between T1 (healthy plant) and T2 (semi healthy) that’s mean that simply we can distinguish between the plant stages. The four obtained modified best indices are [23]: 910 460 690 460 1 P P Modification P P − = + (17) 738 460 554 460 2 P P Modification P P − = + (18) 910 460 554 460 3 P P Modification P P − = + (19) 910 460 553 460 4 P P Modification P P − = + (20) 3. RESULTS AND DISCUSSION The Modified Red Edge Simple Ratio (MSR705) index provides the best result than the other indices. The Modified Red Edge Simple Ratio (mSR705) index results and the average of each stage of this index are obtained for 24-samples. Table 1 contains the obtained results of (MSR705). Figures 2, 3 and 4 show the (MSR705) index range of each stage. Figures 5 and 6 show the graphs of (MSR705) range of the three stages with the 24-samples and their averages. The common range for green vegetation is 2 to 8. IIUM Engineering Journal, Vol. 14, No. 2, 2013 Tawfik et al. 138 Table 1: Results of (MSR705) index for the 24 samples and their average at each stage. MSR705 T1 T2 T3 1.9289 1.9542 1.6241 2.2812 2.1408 2.0042 1.8115 1.8775 2.1952 1.8115 1.9241 2.3484 1.8115 2.3953 2.132 2.333 2.3014 2.1979 1.9777 2.3463 1.7486 2.1524 2.0096 1.8433 1.964 2.0096 1.4375 1.8322 1.9871 1.9948 2.0387 2.0841 1.9395 2.4495 2.0074 2.1526 2.4867 2.1664 1.6989 2.0708 1.9931 1.611 1.8841 1.8922 1.6973 1.9262 2.0325 1.5515 2.3315 1.9102 1.5636 1.9963 1.9185 1.5636 1.9963 1.8371 1.8672 2.1426 2.1424 1.6576 2.1426 2.1137 1.5725 2.1426 1.9396 1.8166 2.1279 2.1267 1.7602 1.6596 1.772 1.6241 Average Average Average 2.0541 2.0367 1.8168 Fig. 2: The (MSR705) index results of T1. In d e x r a n g e Values of MSR equation IIUM Engineering Journal, Vol. 14, No. 2, 2013 Fig. 3: The (MSR Fig. 4: The (MSR Fig. 5: The (MSR705 In d e x r a n g e 0 0.5 1 1.5 2 2.5 3 1 3 5 In d e x r a n g e T1 Healthy plant g Journal, Vol. 14, No. 2, 2013 139 Fig. 3: The (MSR705) index results of T2. Fig. 4: The (MSR705) index results of T3. 705) index results of 24 samples of T1, T2 and T3. Values of MSR equation 7 9 11 13 15 17 19 21 23 T1 Healthy plant T2 Semi healthy T3 Severe damage Tawfik et al. ) index results of 24 samples of T1, T2 and T3. IIUM Engineering Journal, Vol. 14, No. 2, 2013 Tawfik et al. 140 Fig. 6: The average of 24 results of the (MSR705) index of T1, T2 and T3. By using the Jeffries-Matusita (JM) Distance method, it has been found that the MSR705 index is the best index to differentiate between T1 (healthy plant) and T3 (Severe damage), by providing the maximum distance 0.9900167 and the percentage 70.015 % between T1 and T3 shown in Table 2. By using the Jeffries-Matusita (JM) Distance method, it has been found that the Modification 1 index is the best index to differentiate between T1 (healthy plant) and T2 (Semi healthy), by providing the maximum distance and the highest percentage which was 70.715 % from among the four modification indices shown in Table 3. Table 2: The Jeffries-Matusita (JM) Distance calculation for MSR705. T1 T3 Sun 49.2993 43.6022 Average 2.0541375 1.816758333 aVT1-aVT3 0.237379167 Sum(T1+T2) 92.9015 DJM 0.990167233 % 70.015 Table 3: The Jeffries-Matusita (JM) Distance calculation of all modification indices. Modification 1 Modification 2 Modification 3 Modification 4 T1 T2 T1 T2 T1 T2 T1 T2 Sum 123.2413 119.8307 123.0741 120.4596 121.9408 119.6175 124.6445 122.067 average 5.135054 4.992946 5.128088 5.01915 5.080867 4.984063 5.193521 5.086125 Sum(T1+T2) 243.072 243.5337 241.5583 246.7115 avT1-avT2 0.142108 0.108938 0.096804 0.107396 DJM 1.000069992 1.000040997 1.000032825 1.000038976 % 70.715 70.713 70.712 70.713 IIUM Engineering Journal, Vol. 14, No. 2, 2013 Fig. 7: The (VOG1) index results of 24 samples of T1. Fig. 8: The (VOG1) index results of 24 samples of T2. Fig. 9: The g Journal, Vol. 14, No. 2, 2013 141 Fig. 7: The (VOG1) index results of 24 samples of T1. Fig. 8: The (VOG1) index results of 24 samples of T2. Fig. 9: The (VOG1) index results of 24 samples of T3. Tawfik et al. IIUM Engineering Journal, Vol. 14, No. 2, 2013 Tawfik et al. 142 4. CONCLUSION According to the results it was realized that the Modified Red Edge Simple Ratio (MSR705) index of Narrowband greenness VIs has been exhibited an acceptable results as well as it can differentiate between the oil palm plant stage 1 (T1 healthy) and stage 3 (T3 severe damage) using the default equation index because of the healthy plant has a high amount of chlorophyll than severe damage plant. Also the modification equation of Modified Red Edge Simple Ratio (MSR705) index of Narrowband greenness VIs was applied on stage 1 (T1 healthy) and stage 2 (T2 semi healthy) and exhibited an acceptable results as well as it can differentiate between them. REFERENCES [1] Apan, A., Held, A. Phinn, S. and Markley, J.,: Formulation and assessment of narrowband vegetation indices from EO-1 hyperion imagery for discriminating sugarcane disease. Proceedings of the Spatial Sciences Conference, Sep. 22nd-27th, 2003, Canberra. (2003):1- 13. [2] Jensen, J.R,. Remote sensing of the environment: an earth resource perspective. Prentice Hall series in geographic information science, USA, Chap.1 and 10,. (2000):1-28, 333-377. [3] Kumar, L., Schmidt, K., Dury, S. and Skidmor, A.,: Imaging spectrometriy and vegetation science. In: Imaging Spectrometry: Basic Principles and Prospective Applications, van der Meer, F.D. and de Jong, S.M. (Eds.), Kluwer Academic Publishers, Dordrecht.(2001): 111- 55. [4] Shafri, H.Z.M. and Nasrulhapiza Hamdan, Hyperspectral Imagery for Mapping DiseaseInfection in Oil Palm Plantation Using Vegetation Indices and Red Edge Techniques. Am. J. Applied Sci., ." American Journal of Applied Sciences 6.6(2009): 1031-35. [5] Singh, G. Singh, Gurmit, A. Darus, and J. Sukaimi. “ Ganoderma-the Scourge of Oil Palm in the Coastal Area.” In: Ariffin, D. and S. Jalani (Eds.). Proc. Ganoderma Workshop, 11 September 1990, Palm Oil Research Institute of Malaysia, Bangi, Selangor, Malaysia,. (1990): 7-35. Proceedings of Ganoderma workshop, Bangi, Selangor, Malaysia, 11 September 1990.. Palm Oil Research Institute of Malaysia, 1991. [6] Thompson, A.,. Stem-rot of the oil palm in Malaya. Bulletin Straits Settlement and FMS., Science Series, 6:23. US Government Printing Office, 1931. [7] Laudien, R., Bareth, G. and Doluschitz, R., 2003: “Analysis of hyperspectral field data for detection of sugar beet diseases. In: Information technology for a better agri-food sector, environment and rural living.” Proceedings 4th Conference of the European Federation for Information Technology in Agriculture, Food and Environment, Harnos, Z., Herdon, , M. and Wiwczaroski, T.B. (Eds.), Debrecen, Budapest, Hungary, I (2003) 375-81. [8] Blackburn GA. “Hyperspectral remote sensing of plant pigments.” J Exp Bot, 2007, 58:844- 867.." Journal of Experimental Botany 58.4 (2007): 844-67. [9] Gitelson, A.A., M.N. Merzlyak, and O.B. Chivkunova, 2001. “Optical Properties and Nondestructive Estimation of Anthocyanin Content in Plant Leaves.” Photochemistry and Photobiology 71:38-45..74.1 (2001): 38-45. [10] Kaufman, Y.J. Yoram J and D. Didier Tanre, 1996. “Strategy for Direct and Indirect Methods for Correcting the Aerosol Effect on Remote Sensing: from AVHRR to EOS- MODIS.” Remote Sensing of Environment 55:65-79. 55.1 (1996): 65-79. [11] Kumar, L., Schmidt, K., Dury, S. and Skidmor, A., 2001: “Imaging spectrometriy andvegetation science. In: Imaging Spectrometry: Basic Principles and Prospective Applications. van der Meer, F.D. and de Jong, S.M. (Eds.), Kluwer Academic Publishers, Dordrecht,, (2001) 111-155. [12] Ariffin, D., A.S. Idris, 2003. “Progress and Research on Ganoderma Basal Stem Rot of Oil Palm.” In: Mohd Basri, W., K. W. Chan, D. Mohd Tayeb and S. Sundram (Eds.). Proc. IIUM Engineering Journal, Vol. 14, No. 2, 2013 Tawfik et al. 143 “Seminer on Elevating National Oil Palm Productivity and Recent Progress in the Management of Peat and Ganoderma. Bangi, Selangor, Malaysia, (2003): 167-205. [13] Rouse, J.W., R.H. Haas, J.A. Schell, and D.W. Deering, 1973. “Monitoring Vegetation Systems in the Great Plains with ERTS.” Third ERTS Symposium, NASA SP-351 I: 309- 317. NASA special publication 351 (1974): 309. [14] Sellers, P.J., 1985. “Canopy reflectance, photosynthesis and transpiration.” International Journal of Remote Sensing 6:1335-1372. 6.8 (1985): 1335-1372. [15] Huete, A.R., H. Liu, K. Batchily, and W. van Leeuwen, 1997. “A Comparison of Vegetation Indices Over a Global Set of TM Images for EOS-MODIS.” Remote Sensing of Environment 59(3):440-451.59.3 (1997): 440-451. [16] Sims, D A. and Gamon (2002). “Relationships between leaf pigment content and spectral Reflectance across a wide range of species, leaf structures and development stages.” Remote Sensing of Environment 81 (2-3): 337-354. 81.2 (2002): 337-54. [17] Tucker, C.J., 1979. “Red and photographic infrared linear combinations for monitoring vegetation.” Remote Sensing of the Environment 8:127-150. 8.2 (1979): 127-50. [18] Datt, B., 1999. “A new reflectance index for remote sensing of chlorophyll content in higher plants: tests using eucalyptus leaves.” Journal Of Plant Physiology 154 (1999):30-36. [19] Vogelmann, J.E., B.N. Rock, and D.M. Moss, 1993. “Red Edge Spectral Measurements from Sugar Maple Leaves.” International Journal of Remote Sensing 14 (1993):1563-575. [20] Gamon, J.A., J. Penuelas, and C.B. Field, 1992. “A Narrow-Waveband Spectral Index That Tracks Diurnal Changes in Photosynthetic Efficiency.” Remote Sensing of Environment 41(1992):35-44. [21] Gitelson, A.A., Gritz, Y. and Merzylak, M.N. (2003). “Relationships between leaf chlorophyll content and spectral reflectance and algorithms for nondestructive chlorophyll assessment in higher plant leaves.” Journal of Plant Physiology, 160(2003): 271–82. [22] Penuelas, J., F. Baret, and I. Filella, 1995. “Semi-Empirical Indices to Assess Carotenoids/Chlorophyll-a Ratio from Leaf Spectral Reflectance.” Photosynthetica 31(1995): 221-30. [23] Gitelson, A.A., Y. Zur, O.B. Chivkunova, and M.N. Merzlyak, 2002. “Assessing Carotenoid Content in Plant Leaves with Reflectance Spectroscopy.” Photochemistry and Photobiology 75 (2002):272-281.