WPC`  
      2 .      B   =   P    Z      	       #| Z   10cpi Courier 10cpi CG Times (Scalable) CG Times Italic (Scalable) HP LaserJet III                      HPLASIII.PRS x 6X   @ 87,\,~`YX@+ USUK  3'                                          3'Standard                                  6&                                          6& Standard e   rJet III        +                                          #  Xw     P 7[hXP#  2 d     `  "   b      V     +Courier 10cpi CG Times (Scalable)  "  Sh
^;C]ddCCCdCCCCddddddddddCCȲY~~wCN~sk~CCCddCYdYdYCdd88d8ddddJN8ddddYYdYd dd  dddCCCCYdddddd ddd8 YYYYYY~Y~Y~Y~YC8C8C8C8ddddddddddYdddddsdddd  dddd~d~d~d~d  dddddddd8dddd  oddd  ~d~d~8~8  vdddd  dd  kNkdkd~d~d~dddddd  d  YC     dd   d      CC/Nd ddCYQQddddddFddddF Cd hhd44ddzzddd woo     Ch            F"Ȑdh             d岲  dCCȐzȲ CddodȐȅdCdYdsȐ`Ȑ Ȑ ȮzȐ       Uw                                                                                                  ŐdȐ      Y   Y                           C   C   C   C                                                               Őz~o  zoY~NYYYC8YooYdYzsdzd d~YYzozzz~CdzYz z zz CCddd d                                                                  ddd                         z                  C  zdYC  Courier 10cpi CG Times (Scalable) CG Times Italic (Scalable) CG Times Bold (Scalable)    8 w C ; ,  [hXw     P 7XP   3 m = 6 ,  #{&m     P 7&P x / c 8 1 ,  c     P 7P 2 
        .    Z   y
     
   #| G           USUK  3'                                          3'Standard                                  6&                                          6& Standard        HPLASIII.PRS x 6X    +                                          #  Xw     P 7[hXP# 6b6O9O9O9O9\)\)\)mImImImImImIjjI_A_A_A  mI\)mIb6O9\)jIjImIjImI1     II   1      11WxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxNjjjI II1A<<IIIIII4IIII4 11 LLI&&IIZZoIIjI WzzQQ     1L      I     4"jjIjLjj             I傂  I11jZjjjjjjjjjjx1IjIQIjbI1IjAITjr_jjjjjjjjjjDjjjj jjzbZrj       >W                                               jjjj                                                  jjjIjjjj   jj   jA   jjjjA   jjjjj   jjj   jjj   jjj   jjjjjjjjjjjjjj     j   jjjjjj    jjjjjjjj1   jjj1   jjj1   jjj1   jjjj   j                                                   j   jjjj   jjjjZ\Q  ZQbA\9_AmIjA1)gAbQQmAbIjAjZTIZI I\AjAbZjQrZZZZ9I))AAAZAZ Z ZZ 11IIIII          II         ZZZZZZZZZZZZZZZZZZZ9999999IIIIIIIIIIIIIIIIIIII))))))))))))AAAAAAAAAAAAAAAAAAAZZZZZZZZZZZZZZZZZZZZ1  T  j\m1jjZIA1 xHP LaserJet III                      HPLASIII.PRS x 6X   @ 87,\,42X@+ 2        i     v     p   ~   "  Sh
^6=U\\===\====\\\\\\\\\\==Qs~sm=Gsizbsw===\\=Q\Q\Q=\\33\3\\\\DG3\\\\QQ\Q\ \\  \\\====Q\\\\\\ \\\3 QQQQQz~QsQsQsQsQ=3=3=3=3\\\\\\\\\\Q\\\\\i\\\\  ~\~\~\\s\s\s\s\  \\\\\\\\3\\\\  f\\\  s\s\s3s3  m\\\\  z\z\  bGb\b\s\s\s\\\\\\  \  wQ=     \\   \      ==/N\ \\=QKK\\\\\\A\\\\A =\ __\00\\pp\\\ mff     =_            A"\_             \壣  \==p =\\f\z\=\Q\i~X  zp       Nm                                                                                                  ń\      Q   Q                           =   =   =   =                                                               ńpsf  pfzQsGwQQQ=3QzffQz\Qpi\p\ \sQQzpfppps=\pQp p pp ==\\\ \                                                                  \\\                         p                  =  p\Q=  ddddd888888888888YYYYYYYYYYYYYYYYYYYzzzzzzzzzzzzzzzzzzzzC  s  ~CzdYC xCourier 10cpi CG Times (Scalable) CG Times Italic (Scalable) CG Times Bold (Scalable) Univers (Scalable)       6&                                          6& Standard        HPLASIII.PRS x 6X    +                                          #  Xw     P 7[hXP# a8Document g        Document Style  Style                                       X X`	`	  `	

a4Document g        Document Style  Style                                      .   2   k      k          v     a6Document g        Document Style  Style                                    G  X  

a5Document g        Document Style  Style                                   }    X (#

a2Document g        Document Style  Style                                  < o  
   ?                    A.        

a7Document g        Document Style  Style                                   y    X  X`	`	 (#`	

 2   t   C       	   Q  
     Bibliogrphy          Bibliography                                             :   X 
 (#

a1Right Par         Right-Aligned Paragraph Numbers                        : ` S  @                   I.  
  X (#

a2Right Par         Right-Aligned Paragraph Numbers                        	C  	   @`	                  A.    `	`	 (#`	

a3Document g        Document Style  Style                                  
B 
 b 
   ?                     1.        
 2           y  
   .       a3Right Par         Right-Aligned Paragraph Numbers                        L ! 
   `	`	 @P
                  1.  `	`	   (#

a4Right Par         Right-Aligned Paragraph Numbers                        U  j   `	`	  @                  a.    `	 (#

a5Right Par         Right-Aligned Paragraph Numbers                        
_ o    `	`	   @h                  (1)    hh# (#h

a6Right Par         Right-Aligned Paragraph Numbers                        h     `	`	   hh# @$                  (a)  hh#  ( (#

 2                   +  \  a7Right Par         Right-Aligned Paragraph Numbers                        p fJ    `	`	   hh# ( @*                  i)  (  h- (#

a8Right Par         Right-Aligned Paragraph Numbers                        y W" 3!   `	`	   hh# ( - @p/                  a)  -  pp2 (#p

a1Document g        Document Style  Style                                  X q q
    
   l   ^)                       I.           ׃

Doc Init             Initialize Document Style                                  
 
               0*0*    I. A. 1. a.(1)(a) i) a)                 I. 1. A. a.(1)(a) i) a)                                     Document g                                            2 u#           !     @"     "  Tech Init             Initialize Technical Style                              .  
k    I. A. 1. a.(1)(a) i) a)                 1 .1 .1 .1 .1 .1 .1 .1                                      Technical                                            a5Technical         Technical Document Style                               ) W D                   (1)  .  a6Technical         Technical Document Style                               )  D                   (a)  .  a2Technical         Technical Document Style                               < 6  
   ?                    A.        

  2 <&     #     N$     $     %  a3Technical         Technical Document Style                               9 W g 
   2                    1.        
 a4Technical         Technical Document Style                               8 bv {    2                     a.        
 a1Technical         Technical Document Style                               F ! < 
   ?                         I.           

 a7Technical         Technical Document Style                               ( @ D                   i)  .   2 S0     n&    &     *    +  a8Technical         Technical Document Style                               (  D                   a)  .  Pleading              Header for numbered pleading paper                     P@  n                         $]        X    X`	hp x (#%'0*,.8135@8:<H?A                                         y    *                    d       d d                                                                         y y    *                    d       d d                                                                         y 

HH 1

HH 2

HH 3

HH 4

HH 5

HH 6

HH 7

HH 8

HH 9

H 10

H 11

H 12

H 13

H 14

H 15

H 16

H 17

H 18

H 19

H 20

H 21

H 22

H 23

H 24

H 25

H 26

H 27

H 28	 + 	 ӋCourier 10cpi CG Times (Scalable) CG Times Italic (Scalable) CG Times Bold (Scalable) Univers (Scalable) CG Times Bold Italic (Scalable)  "  Sh
^18MSS888S8888SSSSSSSSSS88Jxir{icx{8Aui{x`xoZi{xxxl888SS8JSJSJ8SS..S.SSSS>A.SSxSSJJSJS SS  SSS8888JSSSSSS SSS. xJxJxJxJxJorJiJiJiJiJ8.8.8.8.{SxSxSxSxS{S{S{S{SxSxJ{SxSxSxS{S`SxSxSxS  rSrSrS{SiSiSiSiS  xSxSxSxSxS{S{SS.SSSS  z]SSuS  iSiSi.i.  {c{S{SxSxSx  oSoS  ZAZSZSiSiSiS{S{S{S{S{S  xxS  lJ8     SS   S      88/NxxxS SS8JDDSSSSSS;SSSS; 8S VVS++SSffSSxS c]]     8V            ;"xxSxWxx             S唔  S88xfxxxxxxxxxx 8SxS]SxoS8SxJS`xrxxxxxxxxxxPxxxx xx ofx       Gc                                               xxxx                                                   xxxSxxxx   xx   xJ   xxxxJ   xxxxx   xxx   xxx   xxx   xxxxxxxxxxxxxx     x   xxxxxx    xxxxxxxx8   xxx8   xxx8   xxx8   xxxx   x                                                   x   xxxx   xxxxfi]  f]oJiAlJ{JxJ8.uJo]]{JoSxJxf`SfS SiJxJofx]fffxi{8SxxxfJf f ff 88SSS S                      x                         {                   SSS                xxx         f                  8  fSJ8  \\\\\333333333333QQQQQQQQQQQQQQQQQQQpppppppppppppppppppp=  i  s=p\Q= x 2 C    0  Z  X5    +:    >   "  Sh
^18PSS888S8888SSSSSSSSSS88Sffoxf`xx8Jo]oxfxfS]xff]]888SS8SSJSJ.SS..J.xSSSSAA.SJoJJAJSJS SS  SSS8888SSSSSSS SSS. fSfSfSfSfSooJfJfJfJfJ8.8.8.8.oSxSxSxSxSxSxSxSxS]JfSxSxSxS]JxSfSfSfSfS  oSoSoSxSfSfSfSfS  xSxSxSxSxSxSxSS.SSSS  SSSoS  ]S]S].].  ofoSoSxSxSo  fSfS  SASSSS]S]S]SxSxSxSxSxS  o]S  ]A8     SS   S      88/NxxxS SS8SMMSSSSSS;SSSS; 8S SSS..SSffSSxS ZSS     8S            ;"xxSxSxx             S唔  S88xfxxxxxxxxxx 8SfS]SxoS8SxJS`xrxxxxxxxxxxPxxxx xx ofx       Gc                                               xxxx                                                   xxxSxxxx   xx   xJ   xxxxJ   xxxxx   xxx   xxx   xxx   xxxxxxxxxxxxxx     x   xxxxxx    xxxxxxxx8   xxx8   xxx8   xxx8   xxxx   x                                                   x   xxxx   xxxfff]  f]oJfA]JxJxJ8.oJo]]oJoSxJxffSfS S]J]Joff]fffffx8Sx]]fJf f ff 88SSS S                      f                         x                   SSS                ]]]         f                  8  fSJ8  SSSSS............JJJJJJJJJJJJJJJJJJJffffffffffffffffffff8  `  xi{8xxfSJ8 x  ? x x x ,    gx 6X   @ 87X@   8 w C ; ,  [hXw     P 7XP   3 m = 6 ,  #{&m     P 7&P  x / c 8 1 ,  c     P 7P>  z - b 8 1 , " b &_  x $&7X    2 p = 6 , = h&p _    p ^7&d ! T , , , V QT x    P 7P88xfxxxxxxxxxxx8SfS]SxoS8SxJS`xlxxxxxxxxxxMxxxx xxofx       Gc                                               xxxx                                                  xxxSxxxx   xx   xJ   xxxxJ   xxxxx   xxx   xxx   xxx   xxxxxxxxxxxxxx     x   xxxxxx    xxxxxxxx8   xxx8   xxx8   xxx8   xxxx   x                                                   x   xxxx   xxxfff]  f]oJfA]JxSxJ8.oJo]]oJoSxJxffSfS S]J]Joff]ffffAS..JJJfJf f ff 88SSSSS          SS         fffffffffffffffffffAAAAAAASSSSSSSSSSSSSSSSSSSS............JJJJJJJJJJJJJJJJJJJffffffffffffffffffff8  `  ffx8x]fSJ8 x "  Sh
^6=U\\===\====\\\\\\\\\\==\zzpGXzpfzz===\\=\fQfQA\f3=f3f\ffQG=f\\\Q\\\\\\  \\\====\\\\\\\ \\f3 \\\\\QzQzQzQzQG3G3G3G3f\\\\ffff\\f\\\\pf\\\  \\\fz\z\z\z\  \\\\\ff\3\f\\Gfi\\f  z\z\zGz3  ff\\ψ  \\  fGf\f\z\z\z\ffffff\  zQ=     \\   \      ==/N\ \\=\NN\\\\\\A\\\\A =\ ii\00\\pp\\\ sff     =i            A"\i             \壣  \==p =\\f\z\=\Q\i~X  zp       Nm                                                                                                  ń\      \   \                           =   =   =   =                                                               ńpzf  pfzQzGzQQQG3QzffQz\Qpp\p\ \zQQzpfpppzG\pQp p pp ==\\\ \                                                                  \\\                         p                  =  p\Q=  QQQQQ333333333333QQQQQQQQQQQQQQQQQQQpppppppppppppppppppp=  p  zGp\Q= x "  Sh
^,,CTTe,,,T,,,,TTTTTTTTTT,,EcT^`MJc`%JYHy`eOhVVT```[Q,,,CC,HMHMH1MM  H tMMMM/C1MJtJJ@;C;CTT  TTT,,,,HTTTTTT CTQ  cHcHcHcHcHr^HMHMHMHMH% % % % `MeMeMeMeM`M`M`M`M[JcH`TeMeM[J`MOMcHcTcT  ^T^T^T`TTTTTTTTT  cMcTcTcTcT`T`TT TTTT%Ql@TTYT  TTTTH0H   `[`T`TeTeTy  VTVT  VCVTVTTTTTTT`M`T`T`T`M`Mt[T  Q@,     CC   C      ,,/N```C TT,EJJTTT66T44CCT4 ,C CCT11TTRReCC`C [{{QQ     ,C            4"``C`F``             C充  C,,`Q`````````` ,CcTOTmc@,T`JTM`h[``````````C```` `` tYQh`       9O                                               ````                                                   ```C````   ``   `;   ````;   `````   ```   ```   ```   ``````````````     `   ``````    ````````,   ```,   ```,   ```,   ````   `                                                   `   ````   ```cMTO  JCcMMCQ>`MeQ%#YCcCyO`J`@eMmQOMQV >T>[O{``J{`mhcM`,Te[[mMV h [` ,,TTT T                      c                         `                   TTT                [[[         m                  ,  QC;,  MMMMM            MMMMMMMHHHHHHHHHHHHeeeeeeeeeeeeeeeeeeee,  M  cM`%e[cC;, x 2 V    D  \  H    M   ^  |R   "  Sh
^6=X\\===\====\\\\\\\\\\==\ppzpi=Qzfzpp\fppff===\\=\\Q\Q3\\33Q3\\\\GG3\QzQQGQ\Q\ \\  \\\====\\\\\\\ \\\3 p\p\p\p\p\zzQpQpQpQpQ=3=3=3=3z\\\\\\\\\fQp\\\\fQ\p\p\p\p\  z\z\z\\p\p\p\p\  \\\\\\\\3\\\\  \\\z\  f\f\f3f3  zpz\z\\\z  p\p\  \G\\\\f\f\f\\\\\\  zf\  fG=     \\   \      ==/N\ \\=\UU\\\\\\A\\\\A =\ \\\33\\pp\\\ b\\     =\            A"\[             \壣  \==p =\p\f\z\=\Q\i~X  zp       Nm                                                                                                  ń\      Q   Q                           =   =   =   =                                                               pppf  pfzQpGfQQQ=3zQzffzQz\Qpp\p\ \fQfQzppfppppp=\ffpQp p pp ==\\\ \                      p                                            \\\                fff         p                  =  p\Q=  \\\\\333333333333QQQQQQQQQQQQQQQQQQQpppppppppppppppppppp=  i  pp=fp\Q= x  ? x x x ,    gx 6X   @ 87X@   8 w C ; ,  [hXw     P 7XP   3 m = 6 ,  #{&m     P 7&P	  x / c 8 1 ,  c     P 7P>  z - b 8 1 , " b &_  x $&7X    2 p = 6 , = h&p _    p ^7& d ! T , , , V QT x    P 7P>   1 k = 6 , " W"&k &_  x $&7&X     7 z C ; , = sXz _    p ^7X j ) W 1 + ,  ۏW     P 7P                        C   C   C   C                                                               ΐzo  zoYNYYYN8YooYdYzzdzd dYYzozzzzNY88YYYzYz z zz CCddddd          dd         zzzzzzzzzzzzzzzzzzzNNNNNNNYYYYYYYYYYYYYYYYYYYY888888888888YYYYYYYYYYYYYYYYYYYzzzzzzzzzzzzzzzzzzzzC  z  NzdYC x "  Sh
^;C]ddCCCdCCCCddddddddddCCȲdzN`zoȐCCCddCdoYoYFdo8Co8odooYNCodddYdddddd  dddCCCCddddddd ddo8 dddddϐYYYYYN8N8N8N8oddddooooddoddddzoddd  dddodddd  dddddood8doddNorddo  ddN8  oodd  dd  oNododdddooooooȐd  YC     dd   d      CC/Nd ddCdUUddddddFddddF Cd ssd44ddzzddd ~oo     Cs            F"Ȑds             d岲  dCCȐzȲ CddodȐȅdCdYdsȐ`Ȑ Ȑ ȮzȐ       Uw                                                                                                  ŐdȐ      d   d                           C   C   C   C                                                               Őzo  zoYNYYYN8YooYdYzzdzd dYYzozzzNdzYz z zz CCddd d                                                                  ddd                         z                  C  zdYC  QQQQQ333333333333QQQQQQQQQQQQQQQQQQQpppppppppppppppppppp=  p  zzGpp\Q= x  ? x x x ,    gx 6X   @ 87X@   8 w C ; ,  [hXw     P 7XP   3 m = 6 ,  #{&m     P 7&P	
  x / c 8 1 ,  c     P 7P>  z - b 8 1 , " b &_  x $&7X    2 p = 6 , = h&p _    p ^7& d ! T , , , V QT x    P 7P>   1 k = 6 , " W"&k &_  x $&7&X     7 z C ; , = sXz _    p ^7X ! j ) W 1 + ,  ۏW     P 7P O  A % ! ,  JA     P 7JPer fed Vc rdb'k1cj~right fed WUy ~left fed WWT
 T& )dbAk2cjVL rdb'k1chj~center fed Vn rdb'k1cj~ 2 [  	  W  ,      	  K      "  Sh
^+1DIIr111I1111IIIIIIIIII11Aj\dm\Wjm19g\mjTjbO\mjjj_111II1AIAIA1II))I)rIIII69)IIjIIAAIAI II  III1111AIIIIII III) jAjAjAjAjAbdA\A\A\A\A1)1)1)1)mIjIjIjIjImImImImIjIjAmIjIjIjImITIjIjIjI  dIdIdImI\I\I\I\I  jIjIjIjIjImImII)IIII  kQIIgI  \I\I\)\)  mWmImIjIjIj  bIbI  O9OIOI\I\I\ImImImImImI  jjI  _A1     II   I      11/NjjjI II1A<<IIIIII4IIII4 1I LLI&&IIZZoIIjI WzzQQ     1L            4"jjIjLjj             I傂  I11jZjjjjjjjjjj 1IjIQIjbI1IjAITjrdjjjjjjjjjjGjjjj jj bZrj       >W                                               jjjj                                                   jjjIjjjj   jj   jA   jjjjA   jjjjj   jjj   jjj   jjj   jjjjjjjjjjjjjj     j   jjjjjj    jjjjjjjj1   jjj1   jjj1   jjj1   jjjj   j                                                   j   jjjj   jjjjZ\Q  ZQbA\9_AmAjA1)gAbQQmAbIjAjZTIZI I\AjAbZjQrZZZj\m1IjjjZAZ Z ZZ 11III I                      j                         m                   III                jjj         Z                  1  ZIA1  777771111111111111111111CCCCCCCCCCCCCCCCCCCC%  ?  OEQ%OOC71% x 2     
  \      `                       "  Sh
^!%377bV%%%7b%%%%7777777777%%nbn1bOEKQEAOQ%+MEdQO?OI;EQOhOOG%%%77%17171%777V7777)+77O771171n7 77  777%%%%1777777 777 O1O1O1O1O1bIK1E1E1E1E1%%%%Q7O7O7O7O7Q7Q7Q7Q7O7O1Q7O7O7O7Q7?7O7O7O7  K7K7K7Q7E7E7E7E7  O7O7O7O7O7Q7Q777777  P=77M7  E7E7EE  QAQ7Q7O7O7bO  I7I7  ;+;7;7E7E7E7Q7Q7Q7Q7Q7  hOO7  G1%     77   7      %%/nnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnNOOO7 77%1--77a777bb7'7777b' %7 997bnn77CCTn7n7O7 A\\==b     %9         h   '"nnnnOOnnn7Onn9OObbnnn             7bbnnn  b7%%nnnnnOCnbOOOOOOOOOO %n7O7=7bnOnI7%7O17?OVKnnOOOOOOOOOOnnnnn5nOOOOnnnnnnnnnnnnn nnnnnnOOnn nnn`ICVnOn   nn    /A                                               OOOO                       nn          n         n         nOOO7OOnOOnnn   OO   O1   OOOO1   OOOOO   OOO   OOO   OOO   OOOOOOOOOOOOOO     O   OOOOOO    OOOOOOOO%   OOO%   OOO%   OOO%   OOOO   O                                                   O   OOOO   OOOOCE=  C=I1E+G1Q1O1%M1I=d=Q1I7O1OC?7C7 7E1O1ICO=VCCCOEQ%7OOOC1C C CC %%777 7                      O                         Q                   777                OOO         C                  %  C71%    # &m     P 7#{&P# 
     
 
    
  6&                                          6& Standard        HPLASIII.PRS x 6X   6&                                          6& Standard        HPLASIII.PRS Xw     +                                         












a. Section 2


a Practical Lessons Using Marine Image Data   
         p-++                p-++   6&                                          6& Standard        HPLASIII.PRS Xw    6&                                          6& Standard        HPLASIII.PRS Xw     +                                         # &m     P 7#{&P#           N     j  z N    # c     P 7P# Marine and Coastal Image Data Module 3:`"5 Practical Lessons: 1ė  

a Practical Lessons Using Marine and Coastal Image Data


   T 8 %a 1: EXTENT AND CONTENT DIFFERENTIATION OF COASTAL PLUMESă


   T   a, Aim of Lessonă

To introduce the use of multispectral imageenhancement techniques for the identification of river
plumes and water quality in coastal regions.


   T  a- Objectivesă

After completing this lesson you should be able to:

1) enhance the visual distinction between river plume water and clear sea water

2) distinguish plumes dominated by inorganic suspended sediment from those dominated by dissolved
organic matter (yellow substance)

3) differentiate the thermal characteristics of the observed water types

4) use observations about the shape, size and dispersion of the plumes to estimate the relative
importance of other physical processes operating in the coastal environment.


   T { ia' Background information 

   S , The coastal environment of the study area

The lesson concerns a 25 kilometre stretch of coastline in southern Johor Province on the west coast
of Malaysia, approximately 50 kilometres north-west of Singapore (figure 1).  This is part of a rapidly
industrialising region which includes the Riau Archipelago of Indonesia, and is known as 'The
Singapore Growth Triangle'.  River water quality in the imaged area is typical of that throughout the
region.  A large proportion of the rivers in the region drain peat bogs, found inland of mangrove
forest and numerous darktoned plumes are evident at the mouths of rivers, streams and small
channels.  The most noticeable characteristic of the water is its dark brown colour, imparted by tannin
dyes from decayed organic matter.  It is often carried, in combination with inorganic sediment,
several kilometres offshore as distinct, buoyant plumes.  This results from the low salinity and the
fact that the higher temperature of the dark coloured water often inhibits mixing with the denser
seawater.  Dissolved organic matter is often referred to as 'yellow substance' having a low reflectance
at blue wavelengths and a relatively higher reflectance for green and yellow (figure 2).  It has been
observed (Mantoura and Woodward, 1983) that salinity is inversely related to yellow substance.

Most watercourses on the Johor coast are short, having their source in peat swamp, only one or two
kilometres inland (eg. streams 1 and 2 on figure 1).  A few larger rivers originate at higher elevations
on granite and pass through drained peat swamp only in the lower reaches (stream 3 on figure 1).    \)        p-@*@*@@  As the waters are shallow and mangrove dominated with an eroding coastline, the coastal waters are
highly turbid, with Secchi disk depths of less than 150 cm everywhere within 4 km of the coast. 
Thus, except at the coastline itself, there is no contribution of sea bed reflectance to image
characteristics.

Industrial, economic and tourist development in the region is dependent on water quality.  In this
context, river plumes act as tracers for a variety of oceanographic processes as well indicating water
quality in the catchment.  The river water in question has been found to fall short of minimum
environmental standards for drinking water, primary contact recreation, industrial processes, and fish
culture.  The low salinity inhibits mangrove regeneration, thus depriving marine fisheries of their
spawning habitat.  Local fishermen avoid the dark coloured water as it is acidic, nutrient-poor, and
thus low in productivity.


   S  Image Data

The image was obtained at approximately high tide during the inter-monsoon season, when winds and
ocean currents are relatively low.

aa2 
   T  a' Table 1  Image details.

h  
     d d x                           
 ! d d x        :                       h    	 N 
 
                  d B 9 # T x    P 7QP# Y Name          K

 Date          
# Sensor          n. Size          6 Waveband          'A Wavelength          |M Resolution N 
 
. 	 	              PONTIAN1   
       24.05.89   
       
! LANDSAT TM   
       512 x 512   
       : 1   
       450  520 nm   
       P 30 m . 	 	. 	 	              PONT1   ;          ;          ;       512 x 256   ;       : 1   ;       450  520 nm   ;       P 30 m . 	 	. 	 	
              PONT2   i          i          i       512 x 256   i       : 2   i       520  600 nm   i       P 30 m . 	 	. 	 	;              PONT3                              512 x 256          : 3          630  690 nm          P 30 m . 	 	^		i              PONT6                              512 x 256          : 6          10401250 nm          P 120 m ^		    # &m     P 7#{&P# 

Points you should bear in mind when examining the image are:-

1. LANDSAT Thematic Mapper band 6 is a thermal band and as its spatial resolution is coarser than
that of the visible wavebands, it is less smooth when displayed at the same screen resolution. The DN
values represent a temperature difference of 0.5$C.

2. The pixel data with which you are working is interval data in the range 0 to 255 and is not
calibrated to an absolute ratio scale of radiance values. Thus your observations are limited to relative
comparisons within and between bands.




     &       p-@*@*@@  Ԍ   S    a+ Lesson Outlineă

   T  1. Load and Display PONTIAN1 in VGA mode, subsampled by 2 in both X and Y directions. This
shows the extent of the three plumes with which you will be working. It may be necessary to refer
   T a to the lower portion of PONTIAN1 to view the whole of Plume 3, when using the full resolution
images of the upper half in the rest of the lesson.

   T  2. Use the Modify LUT option to enhance image contrast. Choose Linear Stretch to obtain the
maximum of 256 grey levels for the image histogram. Examine the image in relation to Figure 1 and 
identify Plumes 1, 2 and 3, belonging to streams 1, 2 and 3 respectively. As a guideline, Plume 2
is the small, very dark, plume which hugs the coast in the centre of the image.  Notice the tonal
differences between Plumes 1 and 2, and Plume 3; also between the plumes and non-plume water. 

3. Since the objective is to distinguish spectrally between different plume types, examine Figure 2 and
   T  prepare a table of DN values from the image data (use Table 2) by loading PONT1, and PONT3 in
   T  succession and using the Cursor function. Do the same for PONT6. This should not take you more
than about ten minutes. Land areas have been masked in order to facilitate location of the coastline
   T ] and to exclude extraneous detail on the land. (If you wish to view the land areas, Display
   T 6 PONTIAN1 in full resolution, the upper and lower halves consecutively). In order to locate the
   T  plumes on PONT6, you will need to note cursor coordinates from the other bands. 

From observation of your table and figure 2, consider the following:
   T  (a). Do you find any overlap in the range of pixel values for different water types on PONT1 and
   T q PONT3? 
(b). Bearing in mind that the relative nature of the data limits direct comparison of pixel values
between bands, can you see any relationship between your data and the idealised spectral reflectance
curves in terms of the water quality characteristics of each of the three river plumes? 
(c). What is the maximum distance offshore to which a distinct outline of the plumes can be detected?
To what do you attribute this distinct outline?

Note that the spectrum for clear blue water is unlikely to be accurately represented in the near-coastal
water due to high turbidity. 

4. You can visualise the spectral difference between pixel values for the different water types by using
   T  the Transect option for PONT1 and PONT3. Choose start and end points which include all three
water types. Comment on the numerical separability (or otherwise) of the different water types in
   T k PONT3. 

5. Band ratioing is a technique which accentuates any inherent differences in radiance characteristics
between two wavebands and diminishes features with similar radiance in the two bands. Since your
inter-band comparisons are relative, you are most likely to obtain a useful ratio image where an
inverse relationship exists between features of interest in different bands, eg. you may have found that
while Plumes 1 and 2 exhibit a low response in band 1 relative to the other water types, they exhibit
the highest response in the thermal band. Try ratioing any two bands which conform to these criteria
   T ,% using the Two band ratio in the Multi image function, then stretch the result in Modify LUT.
   T & Histogram Equalisation would be appropriate here since this stretch has the effect of increasing the
contrast at the ends of the histogram.

6. When you think you have obtained a good visual separation between the three different water
   T f) types, try the Cursor and Transect functions again. Are the plumes numerically more distinct?   f)        p-@*@*@@  Ԍ   T    ԙ7. To obtain a better visual effect, load an EGA palette, smooth the image, and use Modify LUT,
   T   Manual Stretch to reduce the range of pixel values to a few broad classes representing the different
water types, and allocate a distinctive colour to each. This procedure effectively classifies the image
by density slicing, producing a map of water quality synonymous with a map layer in a Geographic
Information System.





   T r a2 Referencesă

Mantoura R.F.C. and E.M.S. Woodward, 1983. Conservative behaviour of riverine dissolved organic
   T 
 carbon in the Severn estuary: chemical and geochemical implications. Geochimica and Cosmochimica
   T  Acta, 47, 1293-1309.

Rimmer J.C., M.C. Collins and C.B. Pattiaratchi, 1987. Mapping of water quality in coastal water
   T _ using Airborne Thematic Mapper data. International Journal of Remote Sensing, 81, 85-102.

   T  Robinson I., 1983. Satellite Observations of Ocean Colour. Phil. Trans. Roy. Soc. London, A309
415-432.

      y    b                     d       d d                                                                         y 


This lesson was written by J.Nichol.

   R  Comments or suggestions relating to this lesson should be sent to:

Dr.J.Nichol
Department of Geography( hh. 5 Fax:< 657773091
National University of Singaporehh. 5 Telex:< NUSPER RS51111
10, Kent Ridge Crescent
Singapore 0511
            p-@*@*@@     Y    # Xw     P 7[hXP# 
a+ Table 2  Data Entry Table

# T x    P 7QP# w  
 ! d d x        :                      
 A d d x        |                    w  ^	N 
 
              COORDINATES   !       PONT1   !       PONT3   !       PONT6   !       RATIO1   !       RATIO2 N 
 
. 	 	                      X    Y   O       P1&2  P3  NP   O       P1&2  P3  NP   O       P1&2  P3  NP   O       P1&2  P3  NP   O       P1&2  P3  NP . 	 	. 	 	!              1   }          }          }          }          }        . 	 	. 	 	O              2                                                   . 	 	. 	 	}              3                                                   . 	 	. 	 	              4   	          	          	          	          	        . 	 	. 	 	              5   5
          5
          5
          5
          5
        . 	 	. 	 		              6   c          c          c          c          c        . 	 	. 	 	5
              7                                                   . 	 	. 	 	c              8   
          
          
          
          
        . 	 	. 	 	              9                                                   . 	 	. 	 	
              10                                                   . 	 	P		              
Minimum
Maximum
Mean   k         k         k         k         k       P		    # &m     P 7#{&P# 
   C        p-@*@*@@  










































   T H$ a) Figure 1  Location of Images

   %         p-@*@*@@  










































   T H$ 
a$ Figure 2  Idealised Reflectance Curves
   !%	         p-@*@*@@     T    a- Answers for Lesson 1ă

   T  3(a). You should find that on PONT1, Plumes 1 and 2 are just separable from Plume 3, but that
   T  Plume 3 values overlap with those for non-plume water. On PONT3, the mean, and range of values
for both plume types are similar (not spectrally separable), but values for non-plume water are
considerably lower, thus separable.

3(b). Plumes 1 and 2 exhibit consistently low reflectance in the short wavelength region, with
moderate values in band 3, characteristic of yellow substance-dominated water. Plume 3 has higher
reflectance in band 1, similar to the values for non-plume water. In band 3 its reflectance values are
lower and closer to those of Plumes 1 and 2. Thus Plume 3 resembles water containing suspended
sediment. Since it is highly probable that all the imaged plumes contain significant quantities of both
suspended sediment and yellow substance, and the coastal non-plume water is highly turbid, Figure
2 may not be closely represented by any of the water types on the image, eg. non-plume water on
   T  PONT3 is far from exhibiting the zero reflectance expected of clear blue water.

3(c). Plume 1 extends to a distance of approximately 4 kilometres offshore. River plumes are
characterised by low salinity. Moreover, salinity is inversely related to yellow substance. There is a
slow mixing rate between saline and non- saline water, the former being denser. Another factor
contributing to the distinction between the water masses is temperature, since dark coloured water
absorbs heat and may be warmer. These factors make the plumes buoyant, with a sharp surface
boundary.

   T l 4. As in your pixel data, PONT3 should exhibit lower separability, thus show a smoother transect
than PONT1.

   T  6. A ratio of PONT1 and PONT6 should give much clearer separation between the three main water
types and this will be evident both on the transect and in the data you have entered into table 2.




   
         p-@*@*@@     N    .      z N    # c     P 7P# Marine and Coastal Image Data Module 3:`"5 Practical Lessons: 2ė  

a Practical Lessons Using Marine and Coastal Image Data


   S 8 
a" 2:  USING LANDSAT FOR COASTAL MANAGEMENT1
a MAPPING AND DETERMINING MAJOR SUBSTRATE CATEGORIES.


   T   a1 Aim of Lessonă

This lesson illustrates how to map a remotely and rarely visited, inaccessible coastline.  The mapping
includes estimating, accurately and quantitatively, the locations of important features such as coral
reefs, the exact shoreline and major terrain and habitat types.  Preliminary bathometric information
is also obtained.


   T Y a2 Objectives

After completing this lesson you should be able to:

1) separate land from sea (ie record the precise shoreline)

2) distinguish shallow but submerged features from land

3) record the major terrestrial and marine habitats and discontinuities

4) understand the purpose and effect of filtering images.


   T R ~a, Background Informationă

   S  The BILKO image processing software

Some experience of using previous versions of the BILKO software to manipulate satellite image data
is assumed.  This lesson provides experience in using some of the new capabilities introduced in
version 1.3 of BILKO.  The online help facility can be used at any stage by typing <H>.  An
explanation of the current command operation is given and <ESC> returns to the main program.

   S   The Task

The scene provided is of a desert coastline along the Red sea.  It is an image from the Thematic
   T s# Mapper (TM) of Landsat, whose pixels are 30 x 30 m. It is already geometrically "corrected", that
is, it corresponds to a vertical view with Mercator projection.  The region concerned is rarely visited
and is inaccessible, but some ground observations have been obtained and photographs taken on land. 
The scene is taken at mid tide, but the area has a total tidal range of less than 0.5m.

The scene would lie conformally over a map or chart of the area of the same scale, were one
available.  However, such a map is not available, to this scale or any other; naval charts of the area
are over 50 years old and not accurate, and terrestrial maps are mere sketches which bear little   [)        p-@*@*@@  resemblance to the actual terrain.

The satellite image is thus the only accurate "map" of this region,  a condition which is very common
in tropical countries.

There is a requirement for a map of the area, and for identifying the major habitat types, including
offshore reefs and onshore vegetation stands, if any, for coastal management requirements.  A single
site visit lasting a few hours was made into this inaccessible area.

   S  The Image Data

   T H	 Bands 1 to 4 of the Landsat TM image of the study area are provided in image files named RAS1,
   T !
 RAS2, RAS3 and RAS4, respectively.  The size of the image portions is 512 pixels ' 256 lines ,
corresponding to an area of size 15.36 ' 7.68 km on the ground.

The spectral characteristics of the Landsat TM bands are as follows:

c  
 A d d x        |                  
 a d d x     Z   7           c   
 
 
 
Z              % TM Band          0 Colour          L; Wavelength / nm  
 
` 	 	Z              
( 1   :       Z1 Blue   :       L> 450  520 ` 	 	P                
( 2           1 Green          L> 520  600 P  P  :              
( 3          p2 Red          L> 630  690 P  P                
( 4   *       / Infrared   *       L> 760  900 P  P                
( 5   z       / Infrared   z       L= 1550  1750 P  P  *              
( 6          0 Thermal          %L< 10400  12500 P  z              
( 7   J       / Infrared   J       L= 2080  2350     

   S  Notes about aspects of the lesson

Distinguishing land and sea  The lesson is split into three parts, examining first the whole image, then
the sea areas and finally the land areas.  It is helpful to be able to separate the land from the sea
because different procedures are applied to land and sea areas.  If the complete scene is used,
procedures required for the land will be applied to the sea as well, and viceversa.  This can cause
some confusion.  Thus the first part of the lesson demonstrates how to isolate land and sea pixels on
separate image files.  The most useful information needed for separating land from sea is that infrared (band 4) is totally absorbed by even a few cm of water, but is strongly reflected from most land
surfaces.

Coral Reefs grow to form morphological shapes which have steep slopes facing into deep water, and
have extensive, flat tops which lie more or less at the low tide level.  Parts of their rims may be
exposed at low or mid tide either because they have attached algae which float, or because rocks line
their edges.

Variance is a measure of the spread of values within a group of numbers.  For example the two
groups of numbers 3,4,5 and 1,4,7 each have an average (mean) of 4, but the second group has a   R)       p-@*@*@@  greater variance.

Saving Files  In this lesson, you are asked to save processed images for use later.  This requires space
on your computer's hard disk.  If this is not available, you should store the files on a floppy disk in
   T ` drive A by adding A: to the beginning of each file name.  To load the image again, make sure the
   T 9 appropriate disk is in drive A and at the Load Image prompt enter the complete filename including
   T  A:.  Note that when filters are applied, the filtered image is stored on the hard disk and has a default
filename according to the filter applied (the lesson tells you what it is).  The filtered image is loaded
   T  by entering the default name at the prompt.  If you wish to use the same Filters option several times,
you will need to change the name of the resulting file after each filtering operation to prevent it being
   T t overwritten.  This can be done by using the Shell option in the Miscellaneous Menu.  See also the
Introductory Tutorial for Bilko 1.3.



   S  a0 Lesson Outlineă

   S ] A. The Whole Scene

   T 
 1. Load the image for band 4 (RAS4).  View it by selecting Display image from the main menu. 
If you have a VGA display the image will appear in grey tones.  If you want a colour display you
   T  must use the Load palette function, selecting either the EGA or TEMP1 palette.  Examine the
histogram for the channel.
 X Why is it bimodal ?"
 X If sea absorbs all the infrared radiation, why is the reflectance not zero for pixels covering
the sea ?"

   T  2. The best way to distinguish land from sea is by using infrared (channel 4).  Enter the Multi image
   T  option and select radiometric masking from the menu.  Use RAS4 to define a mask to eliminate the
   T  sea pixels from the image RAS1.  When you have entered the filenames, the radiometric masking
function presents you with a histogram screen.  Move the left cursor to a suitable point between the
   T 2 two main peaks of the histogram.  This has the effect of masking out those pixels in RAS1 at the
   T  locations where the corresponding pixels in RAS4 have values lower than the cutoff defined by the
   T  cursor.  Since the left peak of the RAS4 histogram corresponds to sea pixels this causes sea pixels
   T  to be masked.  To complete the masking, press <RETURN>.  You will be returned to the main menu. 
   T  Selecting Display image reveals the masked RAS1 image.  Save this image to a file called LAND1.

   T F 3. Use the Linear stretch in the Modify LUT option to investigate the image.  Enter the Multi
   T   image menu and select the Toggle option.  Enter LAND1 and the original image RAS1.  Why is the
coastline so much more clearly defined in the masked image?  Could you have detected the coastline
   T ! using RAS1 by itself? 

4. Perform the complement of stage 3.  That is, mask out the higher mode and save the image to a
   T Y$ file named SEA1.  Now toggle between SEA1 and LAND1.  You will see an exact coastline
according to where you set the mask.

5. In the histogram there is a "valley" between the modes.
 X Is it important whether the mask point is at the bottom of one mode or the other, or  exactly
half way ?"   (
        p-@*@*@@  Ԍ X What happens if you move the mask point ?"

   T  6. Repeat stages 3 to 5, applying the mask based on channel 4 to RAS2, the band 2 image, and to
   T  RAS3, the band 3 image.

   T : 7. An alternative way to identify the landsea boundary is to apply a gradient filter.  Load RAS1 and
   T  enter the Filters function.  Select the gradient filter.  This function operates on the displayed image
   T  and rewrites the filtered image to the screen.  Save the result under the name GRADIENT.  Note that
you may need to change the lookup table to be able to see the gradient image.  Notice the bright
lines.  Each line represents the boundary between two very different pixels.
 Is it possible to tell where the landsea boundary is?

   T %
 8. Load LAND1 which you created earlier.  Again select the gradient filter and view the result.  Is
   T 
 it easier now to sea the landsea boundary?  Now load SEA1 again and repeat the gradients.  It is
easier still to see the landsea boundary.  Why should this be?

   T 
 9. Use the toggle function in Multi image and load SEA1 as the first image and GRADIENT
(created in 7) as the second.  Compare the landsea boundaries indicated by each of them.  Which
is the best for determining the "exact" landsea interface?

   S  B. The Sea scene

   T  10. With SEA1 loaded and displayed, enter the Filters submenu and select the Smooth image option. 
This applies a coarse   smoothing filter which has the form of a 3 x 3 matrix of numbers applied to the
image.  The form is;

a2 1    1    1
a2 1    1    1
a2 1    1    1

This filter replaces each pixel with the average value of the 9 of which it is the centre.

Examine the edge of the shore and the outline of the offshore reefs.

   T  Finally save the smoothed image to a file SMOOTH1

   T j   11. Again enter the Filters submenu and select the option 3 ' 3 in which you specify your own
filter.  Note that this filter operates from the stored file on disk and returns the result to a new file
   T   named 3X3 which has to be loaded in order to view it.  (For further information see the introductory
tutorial in section 1 of this module.)  Enter the following filter;

a2 1    3    1
a2 3    9    3
a2 1    3    1

This is a more gentle filter which "weights" the replaced pixel such that its own value is worth 9
times the value of the corner pixels.

   T ( 12. Enter the Multi Image option and select Toggle.  Use this to toggle between your two smoothed
   T e) images SMOOTH1 and 3X3.   e)        p-@*@*@@  ԌWhat are the differences in the effect of each filter ?  Which filter removes the wave "speckle"
without changing the reef outlines ?  Why ?


   S ` C. The Land Scene.

   T  13. Examine the land scene.  Enter the Modify Palette option and change the colours representing
   T  different reflectance.  Using the option to apply a Manual stretch in the Modify LUT menu you can
change the colours representing different reflectance bands.

   T r 14. Apply the Filters, texture function which replaces each pixel on the specified image with the
variance of the 3 x 3 pixels around it.  Note that this also operates on the file stored on disk and
   T #
 returns the answer to an image file named TEXTURE which must be loaded before it can be viewed. 
You will probably also need to change the lookup table to see the patterns on the texture image
clearly.  What does the texture filter signify in this case ?  You may find it helpful to apply the
smoothing filter once to the texture image, in order to assist interpretation.



      y    b L                    d       d d                                                                         y 


This lesson was prepared by C.Sheppard.

   R | Comments or suggestions relating to this lesson should be sent to:

Dr.C.Sheppard
Department of Marine Science and Coastal Management
Ridley Building
The University
NewcastleuponTyne! ( hh. 5 Fax:< (44) 91 222 7891
   R d NE1 7RU,  UNITED KINGDOMhh. 5 Telex:< 53654 UNINEW G
   <         p-@*@*@@  

   S  /a. Answers to Lesson 2ă


2.  The histogram for channel 4 is bimodal because part of the image is sea, part is land, and both
these two substrates have major differences in reflectance characteristics.  

The reflectance is nonzero over the sea because there is dust in the air which reflects infrared in the
pixels over the sea, rather than the sea itself.  The lower mode of the bimodal distribution is
comprised of pixels over the sea.

3. If you are not sure where to set the mask edge, try 56.  However, there is no absolutely correct
place to set the mask  see section 5.

   T  The original image RAS1 does not show the coastline at all clearly becasue the bluegreen light to
which this waveband responds reflects nearly as strongly from the sea bed through shallow water as
it does from the surrounding dry land.

5. It is important to place the mask boundary carefully on the histogram because this determines
where the edge will appear.  There is no position which is exactly right, but you should note that the
further you move the cutoff to the right on the histogram the more wet or damp land is masked as
if it were sea.  

As you move the masking point the position of the boundary will move because progressively drier
pixels will be masked.  In fact, along the edge many pixels contain some water and some dry land. 
These may be called "mixels".

   T  6. Note that as you examine images of progressively longer wavelengths (RAS2 and RAS3), the
reflectance from pixels defined as sea becomes less, because the water tends to absorb the longer
wavelengths of light.

7. It is not possible to tell exactly where the landsea boundary is, because gradient lines caused by
the boundary between two different ground types are evident as well as those between land and sea. 
Overall the image is fairly cluttered with boundary lines.

8. The premasked images show up the landsea boundary well because the masking has emphasised
the difference between land and sea pixels which generates a strong gradient.  This boundary stands
   T   out best on SEA1 because it contains fewer confusing gradients from other boundaries in the sea
   T   compared with LAND1.

   T " 9. You will probably decide that the plain masked image SEA1 is the best for delineating the landsea
   T u# boundary.  It gives a thin clear definition of the boundary whereas GRADIENT gives a thicker,
sometimes more diffuse, line.  Thus gradient filters may not necessarily give the sharpest distinction
of a boundary if alternative methods are available.

12. Observe how the weighted filter is more "gentle".  In many instances small scale details remain
after applying the 3 ' 3 filter but are removed by the application of the stronger smoothing filter.

14. The result of applying the texture filter is to produce an image which has its greatest values where   ^)         p-@*@*@@  the variance of a block of 3 ' 3 pixels is greatest.  The variance is large where the block of 9 pixels
straddles a masked edge, and thus the coastline shows up as a thick band of high values.  Elsewhere
on the land the texture value is highest where the original image was most variable.  If the variability
on the original image is indicative of a type of landcover, the texture image can be interpreted
accordingly.









            p-@*@*@@  
 z  .   f      z N    # c     P 7P# Marine and Coastal Image data Module 3:z              p-@*@*@@  # &m     P 7#{&P#    f   f      z N    # c     P 7P# Marine and Coastal Image Data Module 3:`"5 Practical Lessons: 3ė  


a Practical Lessons Using Marine and Coastal Image Data.



   S  	a" 3:  USING LANDSAT FOR COASTAL MANAGEMENT2 
a. VEGETATION MAPPING.


   T p a0 Aim of Lesson.ă

The lesson develops simple classification procedures to illustrate how vegetation can be identified
from a Landsat TM image, and how the area of important vegetation stands in the image can be
quantified.


   T Y a2 Objectives.ă

After completing this lesson, you should be able to:

1) identify the principal characteristics shown by coastal vegetation in satellite images

2) show vegetation on an image in a highlighted form by assigning the vegetation to a particular
colour

3) measure the area of ground covered by stands of vegetation.


   T z ~a, Background Informationă

   S + The BILKO image processing software

It is suggested that some experience of using the BILKO image processing software package is gained
by working through previous lesson modules.  This lesson demonstrates the use of some of the new
features added to version 1.3.  To load the software, enter the directory containing the programme
and type UNESCO13.  The online help facility can be used at any stage by typing <H>.  An
explanation of the current command operation will be given and <ESC> returns to the main
program.

   S ! The Task

This lesson uses the same image data as the previous lesson on "Mapping and determining major
substrate categories" and is a continuation of it.

The scene provided is of a desert coastline along the Red sea.  It uses thematic mapper (TM) data
   T & from Landsat TM,  whose pixels are 30 x 30 m.  These have already been geometrically "corrected", 
that is each image corresponds to a vertical view with Mercator projection.  The region concerned
is rarely visited and is inaccessible, but some ground observations have been made and photographs
taken on land.  The scene is taken at mid tide, but the area has a total tidal range of less than 0.5 m.   [)        p-@*@*@@  ԌThe scene would lie conformally over a map or chart of the area of the same scale,  were one
available.  However, such a map is not available, of this scale or any other;  Naval charts of the area
are over 50 years old and not accurate, and terrestrial maps are mere sketches which bear little
resemblance to the actual terrain.  The satellite image is thus the only accurate "map" of this region,
a condition which is very common in tropical countries.

There is a requirement for a map of the area, and for identifying the major habitat types, including
offshore reefs and onshore vegetation stands, if any, for coastal management requirements.  One site
visit lasting a few hours was made into this inaccessible area.


   S H	 The Image Data

   T 
 Bands 1 to 4 of the Landsat TM image of the study area are provided in image files named RAS1,
   T  RAS2, RAS3 and RAS4, respectively.  The size of the image portions is 512 pixels ' 256 lines ,
corresponding to an area of size 15.36 ' 7.68 km on the ground.

The spectral characteristics of the Landsat TM bands are as follows:

T 
 a d d x     Z   7         
  d d x     
   7           T   
 
 
 

              % TM Band          0 Colour          L; Wavelength / nm  
 
` 	 	
              
( 1          Z1 Blue          L> 450  520 ` 	 	P                
( 2   :        1 Green   :       L> 520  600 P  P                
( 3          p2 Red          L> 630  690 P  P  :              
( 4          / Infrared          L> 760  900 P  P                
( 5   *       / Infrared   *       L= 1550  1750 P  P                
( 6   z       0 Thermal   z       %L< 10400  12500 P  *              
( 7          / Infrared          L= 2080  2350  z   

   S  Notes about aspects of the lesson

Vegetation contains the green, photosynthetic, chemical compound chlorophyll.  This substance
strongly absorbs red and strongly reflects infrared.  Use is made of this in showing which pixels have
a high cover of chlorophyll.

Strong infrared reflectance is also a characteristic of hot sand and rock, even if covered by no
chlorophyll at all.  However, sand and rocks tend not to absorb red as much as chlorophyll does.

Vegetation distribution  Site visits over some of the area revealed two small mangrove stands, in
addition to extensive mud flats which commonly support rich photosynthetic "bluegreen" algae as
well.  In addition, the edge of the reef crest in this southern Red Sea site supports the thick stands
   T & of the large alga called Sargassum.  It is not practical to provide the photographs with this lesson, but
if this were an operational research project, you would benefit from having access to photographs,
particularly if they were taken close to the time of the satellite overpass.
     T)       p-@*@*@@  Ԍ   T    ԙa0 Lesson Outline

   T  1. Load the image for band 3 (RAS3).
 X Plot the histogram for this channel."
 X Note the value of N where the darkest pixels start."
 Display the scene.  Note that if your computer has a VGA graphics capability the default
display has 16 grey tones.  If you wish to display images in colour you should change the palette to
   T  one containing colours, such as EGA or TEMP1, by selecting Load palette from the main menu.

   T  2. Load the image for band 4  infra red (RAS4).
 X Plot the histogram for the channel."
 X Note the value of N where the darkest pixels start."
 Display the scene.

   T  3. From our site visits and map, we know roughly where the mangroves, algal flats and Sargassum
are located.  It has therefore been possible to identify four positions, in terms of x and y coordinates
on the image, where the ground cover is known.  These points are representative of their surrounding
area, and are defined as follows:

Y 
  d d x     
   7         
  0d d x        #y	              Y  
 
 
z                        x position          y position          G Notes  
 
` 	 	              Mangroves 1          * 175          )5 228          Centre of mangrove stand 1 ` 	 	` 	 	              Mangroves 2   N       * 323   N       )5 123   N       Centre of mangrove stand 2 ` 	 	` 	 	              Algal mats          * 433          )5 136          large area ` 	 			N                 R 6 Sargassum   >       * 493   >       )5 228   >       in large strip on reef edge 		    
   T  Load in turn RAS3 and RAS4, in each case examining the displayed image using the Cursor
function.  Note the values of pixels at the points defined in the table, and compare them with their
surroundings.  Is there anything about the channel 3 and 4 values which distinguishes these points
from the surrounding areas several pixels away?

4. Our intention is to identify the mangrove plants which appear green because of their chlorophyll
content.  Why are we not interested in screening the green channel?  Try loading and displaying the
   T  green band (RAS2) and so determine whether the green of the mangroves can be identified clearly
on this channel.

As you can see, this does not work.  Instead, we shall use a combination of the red and infrared
information to pick out where the vegetation occurs, as explained in the following steps.

   T # 5. Determine the minimum value of the pixels in the red band (RAS3).  You could do this using the
   T # cursor function but it is easier and more reliable to examine the histogram of the red band, as you
did in step 1 earlier in the lesson.  This is known as the "dark value".  A "safe" dark value is one
   T % or two less than the minimum values  use 26 for RAS3.  Subtract it from every pixel in the red band
   T s& using the function Add constant to be found in the Miscellaneous submenu.  Save the resulting
   T L' image to a file called RAS3M.

6. Similarly determine the minimum value of the pixel in the infrared band.  A good dark value for   (        p-@*@*@@     T    RAS4 is 13.  Subtract the dark value from every pixel in the infrared band and save the image to
   T   a file called RAS4M.

   T  7. Enter the Multi Image function and select the Image Ratio option.  Perform a ratio of infrared
   T c over red using RAS4M/RAS3M.  When the scattergram appears place the cursors approximately at
the positions 150/64 (numerator = y coordinate = 154, denominator = x coordinate = 64) and
100/192.  The position of the cursors determines the maximum and minimum of the range of values
of the ratio which are to be scaled to fit on the ratio image between 255 and 0.  Examine how the
ratio image is changed when you set the cursors at 150/64 and 50/128.

 X Bearing in mind that chlorophyll reflects infrared and absorbs red, after the ratio procedure
is complete, what range of values will be found in pixels which include dense mangrove
cover?"

 X Check the sense of your answer.  What would have been the result if you had entered the
numerator and denominator the other way round?  Are the pixels with vegetation in them
higher or lower in value, i.e. did you enter them the correct way round?"

8. Why was it necessary to subtract the dark values from red and infrared first?  Use a pencil and
calculator to show that the "sensitivity" of the ratio is greatly increased if dark values are subtracted
first.

   T  9. Examine the image resulting from the ratio.  Note that the image ratio function delivers its result
into the image memory ready for display.  You will probably need to change the Lookup table before
   T E you can see anything clearly on the image.  Then use the cursor function to investigate the values
in detail.  Since the ratio function scales the results it is not easy to gain direct information from the
absolute pixel values.  The usefulness of the ratio image lies in the way it shows up patterns of high
and low values.  Move the cursor to an area where you know the mangroves to be, and then to an
area of adjacent, bare desert.  What range of values is encountered?  Is this consistent with what you
   T ~ would expect?  Save the ratioed image as RATIO.

   T / 10. Enter the Modify LUT function and change the widths of the bands which relate to each colour
   T  using the Manual Stretch option.  Using the manual stretch set the leftmost "plateau" of pixel values
in the histogram to dark blue.  Then set all of the big peak to light blue.  Set the small tail to the right
of the peak to green and set all the rest of the area on the right of the histogram to red.  Remember
to answer "No" when asked whether you wish to overlay the stretch.

You will now see that the sea and land correspond approximately to the dark and light blue. There
are patches of green where the ratio is high and patches of red where the ratio is very high.

   T ! 11. Return to the Manual stretch function and move the values defining the boundaries of the
colours, particularly between yellow and red.  Do this several times and each time observe the
resulting image.  What exactly are you accomplishing?

Is it fair to assume that other patches on land which now show up the same colour are also covered
with vegetation ?

Can you conclude how useful this procedure might be for determining the extent of vegetation in
areas where you have not visited, i.e. for which there is no ground data?
   b)         p-@*@*@@  Ԍ   T    12. Using the histogram function to examine the image RATIO, and your judgement of which values
correspond to dense vegetation, count the number of pixels defined as mangrove stand.  Multiply the
number of pixels suggesting vegetation cover by 900 to arrive at the actual area of the mangrove
stand.  (Pixels are 30 x 30 = 900 m in Landsat TM).

13. Examine the edge of the reef.  There is a line of apparently chlorophyllrich pixels lining the
edges of some reefs.  These are clearly not mangroves!  Can you tell whether these are a different
type of plant from your image?


     q y    b                     d       d d                                                                         y 


This lesson was prepared by C.Sheppard


Comments or suggestions relating to this lesson should be sent to:

Dr.C.Sheppard
Department of Marine Science and Coastal Management
Ridley Building
The University
NewcastleuponTyne! ( hh. 5 Fax:< (44) 91 222 7891
   R  NE1 7RU,  UNITED KINGDOMhh. 5 Telex:< 53654 UNINEW G            p-@*@*@@     S    a- Answers for Lesson 3ă

   T  1. The value of N where the darkest pixels start on RAS3 is around 28.

   T a 2. The value of N where the darkest pixels start on RAS4 is around 15.

   T  3.  On RAS3, consider the site designated Mangrove 1.  You see that the values of pixels are lower
here and in the immediate zone of about 10'10 pixels corresponding to a small mangrove stand, than
in the surrounding areas.  In this way the mangrove stand shows up clearly on channel 3.  However,
   T  the channel 4 values, for the corresponding "Mangrove 1" pixels on RAS4, are both higher and lower
than their surroundings, implying that little can be concluded from this waveband alone about
vegetation cover.

7. Pixels with dry vegetation will have higher infrared / red ratios than pixels without vegetation.
Vegetation pixels will have lower ratios if the numerator and denominator are mixed up.

8. By subtracting the dark value from each image before performing the ratio, you increase the range
of the numbers in the image and thus improve the potential for discrimination between different ratios. 
   T 4 However, moving the box cursor in the image ratio histogram can have a similar effect.

9. Values on the ratioed image are higher for pixels corresponding to vegetated stands than for pixels
in the surrounding area.

11.  There is a broad spread of pixels which have high to very high ratios.  With this four colour 
display the red pixels represent the locations containing the most chlorophyll, and the yellow pixels
those probably containing some chlorophyll.  By making a small change to the boundary value which
increases the number of red pixels you are reclassifying moderate chlorophyll pixels as high
chlorophyll.  Shifting the boundary to lower values has the opposite effect.  By this means you can
effectively change the area declared to be covered by vegetation.  Of course this does not change in
reality  interpretation based on ground knowledge is indispensable.

In principle it is fair to assume that pixels showing up the same colour as known vegetation areas are
also covered with vegetation.  But in practice you would need some ground investigations, and you
would need to prepare a more refined display than this four colour one.

You might assume that pixels in a generally similar terrain which you have not visited, and which
have similar characteristics to those in your scene, can be classified as ground and vegetation in the
same way.  In fact, remotely sensed images are used for this purpose.

13. It is not possible from a simple four colour image to tell what causes the high chlorophyll along
the edge of the reef.  However, more refined multispectral procedures can help to distinguish between
different types of plant.  For these, more powerful techniques of image processing are required than
can be provided using Bilko.
   M$         p-@*@*@@  # &m     P 7#{&P#    f   ?      z N    # c     P 7P# Marine and Coastal Image Data Module 3:`"5 Practical Lessons: 4ė  

a Practical Lessons Using Marine and Coastal Image Data


   T 8 a 4: DETECTION OF POTENTIAL AREAS FOR COASTAL AQUACULTURE USING
   S  %
a THE INDIAN REMOTE SENSING SATELLITE  IRS 1Aă



   T q  a# Aim of Lessonă

To introduce a method of data interpretation for a newly emerging area in coastal research using the
Indian remote sensing satellite.


   T 
 a$ Objectivesă

On completion of this lesson you should be able to:

1) understand how information about estuarine water content can be derived from multispectral data

2) use a method called the Weighted Index Scheme (WIS) to discover the potential of regions for
coastal aquaculture

3) compare the results of traditional classification techniques with those of WIS

4) detect changes in areas suitable for aquaculture using time series satellite data.


   T S ~a Background Informationă

   S  General information on IRS 1A

The Indian Remote Sensing Satellite  IRS 1A is the first of a series of operational Indian earth
resources satellites.  It was launched on March 17th 1988, into a polar sunsynchronous orbit of
904km by the Soviet launcher VOSTOK from a cosmodrome at Baikanur in the USSR.  With an
inclination of 99.049$ and a period of 103.19 minutes, the satellite has a mean equator crossing time
of 10.25 am on a descending node and a repeat cycle of 22 days.

IRS 1A has a payload of two sensors, namely LISS 1 and LISS 2, with ground resolution of 72.5 m
and 36.25 m respectively.  The corresponding swath widths are 148 and 74 km, the swath of LISS
1 being matched by two individual LISS 2 sensors.  The wavebands listed in Table 1 apply to both
sensors and are similar to those of the Landsat Thematic Mapper (TM): indeed LISS is closely
comparable with the TM.

   &         p-@*@*@@     T    a Table 1  Waveband Definitions

T 
  0d d x        #y	            
  d d x        AX        T  `	 	`	 	               Band          ( Spectral range / m `	 	`	 	              ! 1   q       U, 0.45  0.52 `	 	P               ! 2          U, 0.52  0.59 P P q              ! 3          U, 0.62  0.68 P `	              ! 4   q       U, 0.77  0.86 `	    

   S !
 Source of lesson

This lesson is based upon images of an estuary on the Arabian Sea coast of India.  It results from
work originally carried out as part of the technology development programme in India (Sudarshana
1990), the data analysis being performed at the Indian Institute of Remote Sensing on a VAX 11/780
super mini computer with VIPS 32 software.  The images show a major estuarine system, the
Aghanashini estuary on the west coast of India.  The area extends from 14$28- to 14$32-N and
74$21- to 74$32-E (see figure 1).


   S  Oceanography of the area

The estuaries of the west coast of India are shallow and partially mixed with a strong seasonality due
to the SW monsoon.  Because of their location on the coastal plains, they contain considerable back
water and creek systems.  The Aghanashini estuary has a semidiurnal tide of approximately 2 metres
and presents low lying, intertidal systems like wide mud flats, mangroves and semienclosed
miniature lagoons.  These sedimentary environments are highly productive, but remote for access. 
Consequently, many areas remain unexplored and unexploited.  The aim of this study was to discover
the potential for aquaculture in the area. 


   S  Basis of the Classification Algorithms

The research work encompassed the collection of ground data and digital image processing.  Several
ground stations in the estuary were established and systematic sampling for the analysis of water and
sediment was carried out, based on 24 important ecological factors.  Simultaneously, a hand held
radiometer, compatible with the spectral bands of IRSLISS recorded spectral signatures for each
location.  The data were statistically analyzed on a personal computer and four significant field
parameters were shown to influence the spectral radiance emanating from the stations.  These are
benthic biomass, phaeopigment quantity in the water, sediment median diameter and humic substance
load in the sediment.  In fact, the tidal wetlands are as shallow as 40 to 50 cm most of the time and
the spectral radiance is strongly influenced by the water column characteristics, benthic bioturbation
and the sediment quality.  However, to take care of the tidal influence on sediment moisture,
vegetation in the stations and the burden of yellow substance in the water, three further relationships
were included in the analysis.

The LISS image for 14th January 1990 was selected for comparison with the synchronous ground
sampling.  To identify temporal changes, data from 1989 and 1984 were also obtained.  The 1984   Y)       p-@*@*@@  data were taken from Landsat MSS as LISS was not available then.


   S  Preprocessing of satellite data

The satellite data were analyzed on a VAX 11/780 computer system.  The images were registered
with topographic maps and geometrically corrected for location accuracy.   The field stations were
later registered on the map and the satellite radiance at the field positions was extracted from the
image.  The radiance values from different spectral channels on the 14th Jan 1990 image were used
to develop algorithms for each of the seven field parameters.  Table 2 shows the algorithms that were
developed.

Before applying the algorithms to the image data the land and sea portions were masked, leaving only
the shallow estuarine region.  The Landsat MSS data of 1984 were scaled to match the equivalent
radiance of the LISS.

   T 
 a Table 2  Algorithms for selected ecological parameters based on LISS radiance

J 
  d d x        AX       
  d d x     1           J  `	 
              Algorithm No.          * Model  
` 	1               1          Benthic biomass  =  0.3876+1.3338 log(band 1 radiance) ` 	` 	               2   q       log (phaeopigment) =  0.2188+4.83 log(band 2 radiance) ` 	` 	               3          log (median diameter) = 0.4989+0.6427 log(band 3 radiance) ` 	` 	q               4   1       Humic substances = 4.9133+5.8831 log (band 4 radiance)  ` 	8 	               5          Soil br. index  =  0.332(band 1) + 0.603(band 2)
                 + 0.577(band 3) + 0.388(band 4) 8 	8 	1               6          Green veg. index  =   0.283(band 1)   0.66(band 2)
                    + 0.577(band 3) + 0.388(band 4) 8 	h	i               7   1       Yellow substance index  =   0.899(band 1) + 0.428(band 2)
                          + 0.76(band 3)  0.041(band 4) h	    

   S  Weighted index scheme.

Further processing was carried out on the geometrically corrected and masked images which are
provided for this lesson.  The seven algorithms were applied to the LISS images of 1990 and 1989
and to the MSS image of 1984: consequently 7 images were obtained for each year, each image
pertaining to one significant ecological parameter.  After algorithm application, the original pixel
values were replaced by an absolute measure of the field parameter concerned.

In the next stage local expert knowledge and reviews of research data were used to assess the
optimum ranges of the field parameters best suited to the development of aquaculture.  These are
presented in table 3.  Next each image was considered separately and all pixels falling in the optimum
range were assigned a weight 1.  Pixels falling on either side of the optimum range were assigned a
weight 0.  At this stage, there were seven images for each year and in each image the pixels had the   (       p-@*@*@@  values 1 or 0 depending on whether or not they fell within the optimum range.  Finally, for each
year, the seven images were summed pixel by pixel to generate a single image.  These final images
therefore had pixel values ranging from 0 to 7, the former corresponding to locations having no
potential for aquaculture and the latter having the optimal conditions for aquaculture.  The WIS is
therefore a gradation scheme delineating aquaculture potential.


   T  Sa Table 3  Optimal ranges of parameters used in the weighted index scheme

J  
  d d x     1         
 	d d x        	
        J   
 
              s Parameter   	       - Optimum range  
` 	              benthic biomass   y
          T 	 0.025  0.082 g mé2 ` 	` 		              phaeopigments             T  1.815  49.0212 mg mé3 ` 	` 	y
              Sed. med. diameter   9
       2.78  4.55 Phi ` 	` 	              Sed. humic substance          2.78  5.22 mg/g ` 	` 	9
              Soil brightness index          38.35  46.01 ` 	` 	              Green vegetation index   Y        0.62   7.62 ` 		              Yellow substance index           22.48   23.97 	 Y   
A schematic sketch of the weighted index scheme is provided here as figure 2 and a key to interpret
the weighted gradation is provided as table 4.


   T ! a Table 4  Key to interpretation of weighted index images

J 
 	d d x        	
       
 !d d x        A!        J  	 
Y              Zh Weight   R       Interpretation  
` 	              :h 0          No aquaculture potential ` 	` 	R              :h 1          Essentially deep water pixel with no aquaculture potential ` 	` 	              :h 2   r       Deep water with negligible potential ` 	` 	              :h 3          Relatively shallow pixels with marginal aquaculture potential ` 	` 	r              :h 4   2!       Moderate potential ` 	` 	              :h 5   "       High potential ` 	` 	2!              :h 6   #       Near optimal pixels with very high potential ` 		"              :h 7   %       Most ideal pixels 	 #   
A simple unsupervised classification (Jensen, 1988) of images was also carried out to determine the
principal features of the image.  These classification images are also provided in the lesson, but these
do not give such an accurate indication of areas suitable for aquaculture.
   (        p-@*@*@@  Ԍ   S    ęImage data

The images used in this lesson, listed in table 5, have dimensions  320 pixels ' 240 lines, each pixel
corresponding to a square of 72 m on the ground.  The four bands of the 1990 image are given as
raw counts.  The digital numbers may be converted to true radiance by the following formulae:
 `	`	 band 1:  (grey value ' 15.7)/127
 `	`	 band 2:  (grey value ' 24.5)/127
 `	`	 band 3:  (grey value ' 16.1)/127
 `	`	 band 4:  (grey value ' 15.8)/127
The image resulting from the application of the Weighted Index Scheme is given for 1984, 1989 and
1990, and the unsupervised classification image is included for all three years.

   T  
 Fa  Table 5  Image Data

O  
 !d d x        A!      
 Ad d x        AX"
           O  		 
 
#              c	
 Filename   Q
       # Sensor   Q
       4 Description  
 
` 	 	              I
 AGNA901.DAT          >$ LISS          Raw data, 1990, band 1 ` 	 	` 	 	Q
              I
 AGNA902.DAT          >$ LISS          Raw data, 1990, band 2 ` 	 	` 	 	              I
 AGNA903.DAT   q       >$ LISS   q       Raw data, 1990, band 3 ` 	 	` 	 	              I
 AGNA904.DAT          >$ LISS          Raw data, 1990, band 4 ` 	 	` 	 	q              
 THAGNA90.DAT   1       >$ LISS   1       Classification image, 1990 ` 	 	` 	 	              "
 FIAGNA90.DAT          >$ LISS          Final WIS image, 1990 ` 	 	` 	 	1              
 THAGNA89.DAT          >$ LISS          Classification image, 1989 ` 	 	` 	 	              "
 FIAGNA89.DAT   Q       >$ LISS   Q       Final WIS image, 1989 ` 	 	` 	 	              
 THAGNA84.DAT            Landsat, MSS          Classification image, 1984 ` 	 			Q              "
 FIAGNA84.DAT   A         Landsat, MSS   A       Final WIS image, 1984 		
    

The images are in geographic northsouth orientation with the estuary in the image opening at its far
western point to the Arabian Sea.  Only the tidally active estuarine part has been extracted through
masking.  When investigating the spectral variability of the estuarine features consider the following
facts about the ecology of the region:  the system consists of deep water, shallow water, exposed tidal
flats, mangrove vegetation and stabilised islands;  in general the radiance increases in this sequence,
although in some bands the order is reversed between mangroves and tidal flats.


   S # a" Lesson Outlineă

   T a% 1.  Load and display AGNA901.DAT.  Describe the contrast of this image and consider why it takes
   T :& the form it does.  Using the Histogram option find the maximum and minimum values of the grey
levels.  In spite of the possibility for 256 grey levels, the histogram is narrow.  Why is this so?  Does
this provide a reason for the contrast observed?
   (        p-@*@*@@  Ԍ   T    Use the Linear stretch option in Modify LUT to stretch the image between the minimum and
maximum grey levels.  How does this affect the contrast of the image?
  
   T  2.  Enter the Transect option and plot the transect between the given default cursor positions.  The
curve that appears on the screen shows the variation of radiance over the transect.  Try to relate the
various features of the transect to the ecology described above.  Flick between image and transect to
aid your interpretation.

   T  Use the Cursor option to note the different grey levels across the image: can you relate the pixel
   T  values to the water type and the ecological zone in which the pixel is.  Use the Histogram option to
investigate whether there is a tendency for the grey levels to cluster about discrete values which might
denote distinct water types.

   T 
 3.  Repeat stages 1 and 2 for the other spectral bands, i.e. AGNA902/3/4.  Try to identify the
differences between the different bands and whether this helps you to differentiate between water
   T  types.  The Toggle function can be used to compare two bands at a time.

   T ^ 4.  Load and display THAGNA90.DAT.  Use the Linear stretch option in Modify LUT to stretch
the image and note the number of available grey levels.  Remember that these are not true grey levels
but are different classes as obtained in a classification technique using the spectral bands.  Observation
of the image will show that there are too many classes to give a clear general understanding: they
should therefore be regrouped into a small number of classes.  

Such a regrouping was performed with the benefit of local knowledge of the region.  You can find
   T G the information to repeat this regrouping in table 6.  Select Manual stretch to regroup them on the
histogram.  Assign one colour for each group.  If you find a colour noise covering the screen, you
may have to mask class 0 or 1 with black.  Save this look up table.

   T  Use Modify palette to alter the colour of the classes to contrasting colours and save this new palette. 
Note the changes produced when the image is smoothed.

   T 1 a Table 6. Proposed grouping of class numbers from the THAGNA90 image.

O 
 Ad d x        AX"
          
 ad d x        	E        O  	 
              =V	 Feature   b       N Class numbers to be regrouped in THAGNA90.dat  
` 	              Deep zone          1 to 4, 18 to 21, 28 to 32, 37, 39 ` 	` 	b              Mangrove   "        9, 12 to 16 ` 	` 	              Sites for aquaculture   !       5 to 8, 10 to 11, 17, 22 to 27, 33 to 36, 38 ` 		"               Tidal flat   #       40 to 60 	 !   

   T $ 5.  Use the histogram option and the table of data provided above to estimate the number of pixels
in each of the four grouped classes and prepare a table of results.  If each pixel is 72 x 72 metres,
calculate the area of each grouped class in the Aghanashini system and add this to the results.  (A
factor of 0.005184 would yield area in square kilometres).

   T ( 6. Repeat the above for THAGNA89.DAT and THAGNA84.DAT using the class regrouping data   (        p-@*@*@@  given in Tables 7 and 8.

  
   T  xa Table 7. Proposed class groups for 1989 data.

J  
 ad d x        	E       
 d d x     9   /W        J   
 
9               Feature          z Class numbers to be regrouped in THAGNA89.DAT  
` 	9              Deep zone          1, 4, 7, 9, 15, 18, 22 to 23 ` 	` 	              Mangrove   y       5 to 6, 14 ` 	` 	              Sites for aquaculture   	       2 to 3, 8, 10 to 13, 16 to 17, 19 to 21, 24 to 26 ` 		y              Tidal flat   i       27 to 60 	 	   

   T 
 xa Table 8. Proposed class groups for 1984 data.

J  
 d d x     9   /W       
 d d x        /W        J  	 
	               Feature   J       z Class numbers to be regrouped in THAGNA84.DAT  
` 	              Deep zone          1, 2, 4 to 5, 8 to 9, 12 to 14, 16 to 21, 23, 38 ` 	` 	J              Mangrove   
       6, 15 ` 	` 	              Sites for aquaculture    j       3, 7, 10 to 11, 22, 24 to 26, 34 to 37 ` 		
              Tidal flat          27 to 33, 39 to 60 	 j   

Using the above classifications determine the change in the area of the four estuarine classes between
1984, 1989 and 1990 and identify which parts, if any, of the estuary appear to have undergone
substantial changes.  Can you detect any trends or do the changes seem to be random?

   T 
 7.  Load and display FIAGNA90.DAT.  This is a weighted index image.  There is nothing on the
screen except the colour bar.  Why is this?  How many grey levels would you expect to see in this
image?

   T k Use the Linear stretch option to enhance the image between its maximum and minimum values. 
   T D Now use Modify palette to assign contrasting colours to the weights.

   T   8.  Interpret the image with reference to table 4.  Using the Histogram option, note the number of
pixels in each weight and multiply by a factor of 0.005184 to calculate the area of each weight in
square kilometres.

Table 4 suggests that weights 4, 5, 6 and 7 define areas increasingly suitable for aquaculture: assign
a single colour to all these weights and mark the others grey.  This will produce a very clear map of
the aquaculturable area.  Is this distributed throughout the estuary or localised?

   T ' 9.  Repeat the above exercises for FIAGNA89.DAT and FIAGNA84.DAT.  Evaluate the temporal
changes in total aquaculturable area in the estuary.  Does this method suggest it is increasing or
decreasing with time?   g)       p-@*@*@@  Ԍ10.  Compare the two methods for estimating aquaculturable area (i.e. compare the outcome of stages
8 and 9 with the results of steps 5 and 6.  In particular, compare the magnitude of the areas for
aquaculture predicted by each method and their distribution within the estuary.







     p y    b                     d       d d                                                                          y 


This lesson was written by R. Sudarshana.


   R  Any comments or suggestions relating to this lesson should be sent to:

Dr. R. Sudarshana $ hh* 0 Telex:6 0585224
Indian Institute for Remote Sensing
4, Kalidas Road, P.B. 135
Dehra Dun 248001
INDIA


   P          p-@*@*@@  













































   T & ia9 Figure 1  Location Map   
   '!         p-@*@*@@  













































   T & `	a& Figure 2  Data Analysis Procedure for the Weighted Index Scheme
   '"         p-@*@*@@     T    a; Answers for Lesson 4ă


1.  The image contrast is very low as the brightness of the ground elements are in their original
unstretched form.  The histogram is restricted to a very small area on the scale of contrast.  The
image is masked to remove the land portions and so the contrast is restricted to the variability of
elements within the estuarine bounds.  The minimum and maximum grey levels are 30 and 47
respectively.  The contrast observed is so because the channel chosen for the experiment lies between
0.45 and 0.52 m, a waveband within which the radiance variability is low for tropical estuaries.

Using the linear stretch the contrast is enhanced.  This does not necessarily imply that new features
have been delineated: to be certain you would need ground observations to pick out features which
correlate with the contrast observed on the enhanced image.

2.  The transect shows two nodata areas which account for the lowering of values on the grey scale. 
Ignoring this, there is not much variation in the profile except in the region before the second nodata
area.  This means that the transect line passes across an almost homogeneous zone except in a small
region where radiance is high.

The cursor option shows high DN values  about 44  which corresponds to exposed mudflats, and
low values  about 34  which shows shallow water to be present.  Roaming with the cursor shows
that most central parts belong to the range 32  36, while near the margins values are in the range
3945.  This confirms the distinct delineation between water bodies and tidal areas.  However, using
a single channel it is not possible unambiguously to distinguish between different water classes.

3.  There are evidently differences in the patterns of pixel values between the different bands, and
these can help in distinguishing between water types.  For example the regions on the channel 3 and
4 images (red and infrared wavebands) where the reflectance is greater normally indicate exposed
or very shallow vegetation.  It is, however, not easy, given the limitations of the Bilko image
processing toolkit, to systematically extract clear information from multichannel data.  The best you
can do is to examine the spectral signatures of a few particular locations; that is, note the pixel values
in each band at specific locations.  To extract the full information content of multispectral data
requires classification procedures which can be performed by more powerful software.  You can find
the results of such procedures in the classification images introduced in section 4 of the lesson outline.

4.  There are 60 grey levels.

5,6.  The following table is a sample for 2 years and on these lines you should calculate all of the
three data sets.

   !#         p-@*@*@@  
a+ Distribution of Feature space in West coast estuaries
 W
a1 as per maximum likelihood classification
   T  a@ Area in km2ă

^  
 d d x        /W       
 d d x     8$   \rnJ                    ^  X 
 
X 
 
8              Yv Estuary          Z) Year          `}8 Estuary
#}7 (channel)         )H Mangroves         Z Tidal
Z flats         k Sites for
j Aquaculture X 
 
` 	 	8              v Aghanashini          W) 1990          y}8 16.169          S)J 4.683          Y 29.761          l 14.355 ` 	 			              v Aghanashini   	       W) 1989   	       y}8 12.851   	       S)J 4.578   	       Y 27.952   	       m 19.59 		    

7.  There are only 7 grey levels in the image as it contains only values from the weighted index
scheme.  As 7 is a very small number on the scale of 256 possible grey levels, the contrast is not
visible until the lookup table is changed.

8,9.  The following table is a sample for 2 years; on these lines you should calculate all of the three
data sets.



a3 Distribution of Areas for aquaculture
a7 as per Weighted Index Scheme
   T  xa? Area in km2 
# c     P 7P# 
  
 d d x     8$   \rnJ                   

 d d x     @$   QQHHzzQ@                                  p 
 
p 
 
@              Year          wt. 0          wt. 1          wt. 2          wt. 3          wt. 4          wt. 5          wt. 6          7   	       Total (47) p 
 
P 	 	@              1990           6.37           8.28           10.93           8.53           5.40           6.86           3.67           4.33   	        20.26 P 	 			              1989          11.87          6.42          12.06          5.82          5.93          7.48          3.85          1.13   	       18.39 		    
 
