REMOTE  SENSING  ON  GANODERMA  DISEASE

Nisfariza  Mohd  Noor  Maris  Senior  Lecturer  

Department  of  Geography,  University  of  Malaya  KUALA  LUMPUR,  MALAYSIA  

 

Outline  

�   Introduc.on  �   Ganoderma  BSR  

•   Pathological  symptom  •   Detec.on  of  Ganoderma  BSR  

�   Sensor  for  Disease  Detec.on  �   Remote  Sensing  Proper.es  of  Plant  �   Remote  Sensing  Research  on  Ganoderma  Disease  in  MPOB  

�   GER1500  Spectroradiometer  •   Data  &  Analysis  

�   AISA  Hyperspectral    �   Data  &  Analysis  

�   Unmanned  Aerial  Vehicle  Data  &  Analysis  

�   Integra.on  of  GIS  and  Remote  Sensing  

�   Future  Developments  �   Conclusion  and  Recommenda.on  

2  

Introduction •  Oil   palm   (Elaeis   guineensis   Jacq.)   the   most  important  agriculture  sector  in  Malaysia  

•  Damage  and  loss  due  to  pests  and  diseases  :  serious  problems   and   can   reduce   the   profitability   of   the  business;   jeopardise   the   growth   of   the   industry  over  Dme.    

•   Three  major  oil  palm  threats  for  the  crop  losses  can  be  classified  into  insects,  vertebrates  and  diseases.    

3  

… Cont

•   Monitoring  agricultural  areas  :    Dme  and  labour  consuming  –    dealing  with  large  area  of  homogenous  crop.    

•   Disease   and   pest   detecDon   and   control   are   important  phases  in  agricultural  management.    

•   Cultural   pracDces   combined   with   biological   and   chemical  control   have   been   considered   as   the   best   approach   for  controlling  diseases  and  pests.    

•   Remote  sensing  provides  a  possible  soluDon  to  the  intensive  sampling  required  for  site-­‐specific  pest  management.  

4  

Ganoderma  BSR  

•   Fatal   disease   –  most   serious   disease  of   oil   palm   in  Malaysia   for   over   80;  severe  economic  loss  in  Malaysia.  

•   The   soil-­‐borne   fungi   aNack   from   the  roots   and   progressively   spread   to  the  stem  and  bole  system.    

•   Prevents  water  and  nutrient  uptake  :  foliar   symptoms   and   gradually  affects   the   growth   and   yield,  eventually   leading   to   the   death   of  the  palm  5  

Detection  of  Ganoderma  BSR  

§ Visual   interpretaDon   -­‐   middle   or   late   stages   of   disease  pathology   commonly   Dme-­‐consuming,   destrucDve   and  expensive.  

§ Aerial   detecDon   –   enable   treatment   at   the   early   stage   of   the  infecDon  and  avoid  more  extensive  damage  and  losses.    

§ Wet   lab   detecDon   -­‐   subclinical   symptom   of  Ganoderma   GSM  test,   PCR-­‐DNA   molecular,   ELISA-­‐polyclonal   anDbody   and  GanoScan1  tomography.  

6  

Pathological  Symptom  

§  The  visible  symptom  of  is  seen  at  later  stage:  infesta.on  of  the  pathogen  has  damaged  the  stem  and  bole  of  the  oil  palm.    

§   Four   Ganoderma   species   to   be   associated  with   oil   palm   in  Malaysia:  G.  boninense,   G.   zonatum,   G.   miniatocinctum   and   G.   tornatum   but   G.  boninense  is  the  most  aggressive  (Idris,  1999).  

7  

...  Cont  

8  

Sensors  for  disease  and  stress  detec.on  in  plants  

Molecular  methods  

Bio-­‐chemical  analysis  

Biomarker-­‐based  sensor  

Remote  Sensing  

Remote  Sensing  

Visible-­‐near  

infrared  spectrosc

opy  

Fluorescence  

spectroscopy  

Mid  infrared  spectrosc

opy  

Fluorescence  

imaging  

Hyperspectral  

imaging  RADAR  

Mul.spectral  

Imaging  Others  

Spectroscopic  Techniques   Imaging  Techniques  

Modified  from  Reza  Ehsani  (2010)  9  

…Electromagnetic  Spectrum  (Light+)  

…incoming  light  is  preferentially  absorbed  (reflected)  depending  on  plant  physiology          Species        Photosynthesis        Water  Content  

http://www.innovativegis.com/basis/  

Remote  Sensing  Properties  of  Plant  

10  10  

•   Elowitz  (2013)  @  h8p://www.markelowitz.com/Hyperspectral.html    

11  

Biochemical  Response  of  Leaf    

12  

•   Philpot,  2011  

Remote  Sensing  Research  on  Ganoderma  in  MPOB  

Spectroradiometer  (GER1500)  

Hyperspectral  (AISA  DUAL)  

Multispectral  (UAV  Swinglet)  

RADAR  

13  

Hyperspectral Approach Spectroradiometry  

Nursery  (seedlings)  

10  months  old    

24  months  old  

Planta.on  

5  years  old  young  palm  

17  years  old  mature  palms  

Hyperspectral  Imaging  

Planta.on  

5  years  old  young  palm  

17  years  old  mature  palms  

Visual  Classification    @  Nursery  trials  

Severity   Classifica.on   Visual  Assessments  

T1  Seedlings   not   inoculated   and   not  infected  with  Ganoderma   boninense,  seedling  is  healthy.  

Leaves   and   tree   looking   healthy.   Absence   of  white  mycelium  or  fruiDng  body  (Ganoderma)  at  stem  base.    

T2  

Seedlings   inoculated   and   infected  w i t h   Ganode rma   bon i nen se ,  WITHOUT   foliar   symptoms   but   with  white   mycelium   or   fruiDng   body   at  stem  base.  

Leaves   and   tree   looking   healthy   without   any  foliar   symptom   of   BSR   disease.   Presence   of  white  mycelium  or  fruiDng  body  (Ganoderma)  at  stem  base.  

T3  

Seedlings   inoculated   and   infected  with   Ganoderma   boninense,   WITH  foliar   symptoms   and   with   white  mycelium   or   fruiDng   body   at   stem  base.    

Yellowing,  browning  or  drying  of  some  leaves  due  to  Ganoderma  infecDon.  One  or  two  new  leaves  remain  as  unopened  spears.  Decline  of  older   leaves.   Presence   of   white  mycelium   or  fruiDng  body  (Ganoderma)  at  stem  base.    

15  

Visual classification @ Nursery Trials

T1 T2   T3  

Visual  Classification    @  Plantation  plots  

Severity   Classifica.on   Visual  Assessments  

T1   Healthy  palm.  Leaves  and  tree  looking  healthy.  Absence  of  white  mycelium   or   fruiDng   body   (Ganoderma)   at   base  stem  

T2  

Palm   infected   with   Ganoderma  spp.  w/o   any   foliar   symptoms   but  w  white  mycelium  or  fruiDng  body  at  base  stem.  

Leaves   and   tree   looking   healthy.   Presence   of  white  mycelium  or   fruiDng  body   (Ganoderma)  at  base  stem  

T3  

Palm   infected   with   Ganoderma  spp.   with   foliar   symptoms   and  white  mycelium  or  fruiDng  body  at  base  stem.  

Yellowing   or   drying   of   some   leaves.   One   or   two  new   leaves   remain   as   unopened   spears.  DeclinaDon   of   older   leaves.   Presence   of   white  mycelium   or   fruiDng   body   (Ganoderma)   at   base  stem.    17  

Visual  Classification  @  Plantation  Plots  

Visual   appearance   of   severity   classes   in   oil   palm   plantation   for  Ganoderma  BSR  study.  

T1   T2   T3  

18  

Data  Acquisition  &    Equipment  

GER1500  Handheld  Spectroradiometer  

§ Baseline   study   of   possibility   the   use   of   remote   sensing   in  detecting  Ganoderma  BSR.  

§ Massive  ground  works.  § Wavelength  400nm  to  1050nm.  § 1.5  nm  intervals  § 512  bands  

20  

Spectroradiometer  Data  

•  Spectral  data  processing:  

Denoising  –    5pt  smoothing  

Original  Reflectance  

First  Derivative  

Stats  –  Denoise    spectra  

Stats  –  1st    Derivative  

Stats  -­‐  Original  Ref  

Compare    Statistically  

21  

Raw  Reflectance  Data  

T1  –  Inoculated  looking  healthy,  no  foliar,  

mycelium  or  fruiting  body  

T2  –  Inoculated  no  foliar  symptom  but  with  white  

mycelium  or  fruiting  body  at  base  stem  

T3  –  Inoculated  with  foliar  symptom  and  white  

mycelium  or  fruiting  body  at  base  stem  

22  

First  Derivative  Analysis  

T1  –  Inoculated  looking  healthy,  no  foliar,  

mycelium  or  fruiting  body  SFS  =  9%  

T2  –  Inoculated  no  foliar  symptom  but  with  white  

mycelium  or  fruiting  body  at  base  stem  SFS  =  14%  

T3  –  Inoculated  with  foliar  symptom  and  white  mycelium  or  fruiting  body  at  base  

stem  SFS=49%  

EXPERIMENT  1  23  

Red  Edge  Position  Analysis    

�   The  inflection  point  on  the  slope  between  the  red  absorption  and  near  infrared  reflectance   -­‐   used   to   correlate   chlorophyll   content   with   reflectance.  Decreasing  chlorophyll   causes   the   red  edge  position   to  shift   towards  shorter  wavelengths.  

�   Shifts  in  the  wavelength  position  of  the  red  edge  and  the  intensity  of  the  peak  slope   have   been   related   to   stress   factors   in   vegetation   such   as   chlorophyll  decline  and  senescence.  The  red  edge  feature  is  therefore  an  early  indicator  of  disease  symptoms  in  leaves.  

24  

24  

AISA  Dual  Hyperspectral  System  

§ A  scale  up  study  from  ground  to  aerial  data.  § Wavelength  400nm  to  2500  nm  § 5nm  interval  § 244  bands    § 0.68m  resolution  § Specim,  Finland.  

25  

Data  Acquisition  and  Equipment  

26  

Hyperspectral  Airborne  Image  Processing  

Raw  Data  

Classification  

Cross  Track  Illumination  

Processed  Data  

Radiometric  

Spectra  Statistics  

Continuum  Removed  

Atmospheric  

Geometric    

MNF    Transform  

Vegetation  Indices  

Accuracy  Assessment  

Pre-­‐Processing  

27  

Study  Area  –  Planted  Palm  /  Airborne  

Mature  palm  17  years  old  

Young  palm    5  years  old  

¡   Oil   palm   plantation   in   Seberang   Perak,  Perak,  Malaysia  (4°  6’  N  100°  53’  E).    

¡   Data  acquired  October,  2010  

28  

AISA  Hyperspectral  Image  29  

29  

The  hyperspectral  image  acquired  20  October  2008,  effect  of   fluctuations   in   solar   illumination   during   flight.  [Unfavourable   weather   in   October,   the   month   for  monsoon  transition  period  in  the  country].  

30  

Best  Band  Selection  •   Significant   bands   to   discriminate   palms   infected   with  Ganoderma  BSR,  it  shows  that  early  detection  is  possible  using  derivative  spectral  distinction.    

•  Wavelengths   bands   suitable   for   discriminating   different  levels   of   Ganoderma;   the   original   reflectance   spectra  and   first   derivative   spectra   to   obtain   the   best  wavelengths  in  discriminating  the  disease.    

•   The   results   showed   that   for   Ganoderma ,   the  discriminating  bands  for  mature  and  immature  oil  palms  were  found  in  ranges  of  515  –  586  nm,  615  –  622  nm,  633  –  644   nm,   690   –   708   nm,   727   –   758   nm  and   773   –   784   nm  respectively.  

31  

Hyperspectral    Vegetation  Indices  

§  Vegetation   Indices   –     indicate   the   physiological   and   pathological  parameters   of   vegetation   –   interpret   the   status   of   vigour   of   the  plants.  

 

§  It   is   not   always   possible   to   detect   disease   or   other   stresses   in   crops  from   reflectance   values.   A   disease   may   be   difficult   to   detect   at   an  early   stage,   e.g.   fungi   that   attack   the   roots   of   plants   or   the  base  of  their  stem,  but  do  not  at  first  cause  any  reaction  in  the  plants  itself.    

 

§  Several   indices   was   tested:   the   Normalised   Difference   Vegetation  Index   (NDVI),  Modified  Chlorophyll  Absorption  Ratio   Index   (MCARI),  Structural   Independent   Pigment   Index   (SIPI),   Optimised   Soil  Vegetation  Index  (OSAVI)  and  Transformed  Chlorophyll  Absorption  in  Reflectance  Index  (TCARI).    

32  

Normalised  Difference    Vegetation  Index  (NDVI)  

Brighter  -­‐  Healthier  33  

…  Cont  

Classified  /  reconciliation  based  on  the  index  value  of  NDVI  

NDVI  Values  Non  Vege  /  Bare  Soil  :  0  –  0.7  (Red)  T1  :  0.85  –  0.9  (Green)  T2:  0.85  –  0.8  (Yellow)  T3:  0.7  –  0.8  (Blue)  

34  

34  

Unmanned  Aerial  Vehicle  (UAV)  

§ UAV   has   become   commercially   available   nowadays   –   have  the   advantages   of   high   spatial   and   temporal   resolution  characteristics  compared  to  airborne  and  satellite  platforms.  

§     § The  advancement  of  remote  sensing  technology  in  providing  a   synoptic   view   –   provides   possibility   to   monitor   a   large  plantation   state   of   vigorousness   and   stress   due   to   disease,  pest  and  nutrient  deficiency.  

§ UAV  promotes   the   remote   sensing   technology   in  detection  and   monitoring   of   existing   plot   infected   with   Ganoderma  BSR.    

 35  

Unmanned  Aerial  Vehicle  

§ Same   visual   detection   concept   is   applied   to   the   RGB   image;  since  it  corresponds  to  the  optical  concept  of  a  human’s  eye.    

§ The   use   of   NIR   cameras   in   the  UAV   is   additional   whereby   it  implies   the   high   reflectance   of   vegetation   in   the   NIR   region  invisible   to   the  human  eye   to  demonstrate   the  health  of   the  plant.    

§ The   RGB   and   NIR   images   are   coupled   with   GIS   analysis   to  achieve   the   aim   to   monitor   Ganoderma   BSR   in   oil   palm  plantation.    

36  

Data  Acquisition  and  Equipment  

37  

Flight  Planning  

38  

UAV  Swinglet  Data    

The  RGB  and  NIR  images.    39  

Resolu.on  :    •   RGB  (10cm)  •   NIR  (4cm)    

Integration  of  GIS  and  Remote  Sensing  

40  

GIS  Spatial  -­‐  Temporal  Analysis  

•   To   incorporate   all   data   collected   by   remote   sensing  with   the   census   to   develop   a   time   series   of   disease  spread  over  time.  Multi  sensor  and  multi  temporal.  

41  

What is next? •   Ganoderma   Workshop   2014   (MPOB   and   Malaysian   Oil  Palm  Stakeholders)  :    

•   Early   detecDon   tools   –   more   sensiDve,   precise   and  localize    

•   Standardize  the  methods  and  classificaDon  –  visualize    •   υFaster,  cheaper  and  precise  remote  sensing  and  GIS  •   Aerial   detecDon   –   need   esDmaDon   for   the   staDsDcal  analysis  using  hyperspectral  remote  sensing.    

•   Estate  experience  1  UAV   for  500  ha  per  day,  depend  on  certain  limitaDon    

Future  Developments  •   Develop   specific   vegetation   index   for   Ganoderma   based   on  significant  bands.  

•   Modeling   on   the   relationship   of   Ganoderma   disease   with  factors  such  as    soil  type,  moisture  content,  nutrient.  

•   Index   mapping   of   Ganoderma   disease   for   areal   based  identification  for  management  and  planning  purposes.  

•   New  UAV  hyperspectral  system  –  using  band  selection.  

•   GIS  mapping,  hotspots  and  prediction  of  disease  dispersion.    

•   RADAR   (Radio   Detection   and   Ranging)   fusion   with  hyperspectral  imagery.  

43  

Conclusion  and  Recommendation  

§  Identification  based  on  percentage  over   total  area  and  assessment  of  the  disease  symptoms  

§  Early   detection   of   the   disease   is   vital   for   profitable   agriculture  business.    

§  Plantation   management:   early   identification   ameliorate   on   plan  and  strategy  to  overcome  the  problem  in  a  cost  effective  manner.    

§  Expands   the  Ganoderma  BSR  detection  survey  options  available   to  plantation   managers   and   a   viable   data   source   for   detecting   and  monitoring  damage  due  to  pest  and  disease  in  oil  palm  plantation.  

§  Cheaper  technology  for  detection  

44  

Acknowledgement  

•  This  research  is  funded  by:  •   Innovative  Technology  Research  Cluster  (ITRC),  University  of  Malaya  http://itrc.um.edu.my  

•   Ganoderma  and  Disease  Research  of  Oil  Palm  Unit    (GANODROP)  –  Malaysian  Palm  Oil  Board,  MPOB  

•  Many  thanks  to  :  •   Department  of  Geography,  University  of  Malaya  

45  

Thank  You  

Muchas  Gracias