tmnov tutorial : Bottom Typing

return to Model 2      next to Model 3


 
1 - NO NEED for field data, nor for atmospheric correction
2 - this is demonstrated in this website, using a variety of hyper/multi spectral data
 
Requirements are
1 - homogeneous water body and atmosphere
2 - some coverage of optically deep water
3 - some coverage of dry land
 
Problems are
1 - the precision on estimated depth is found wanting, because the noise-equivalent change in radiance  of accessible data is too high for shallow water column correction work 
2 - radiance data should be preprocessed by the provider at level 1 in order to improve S/N ratio
3 - exponential decay: the deeper/darker the bottom, the poorer the performances
 
So
I keep digging
until suitable data
become available
 
 
 
Bottom Typing
SyntaxClassify
 
Select an AOI
  • draw your first shapefile where you see a patch of homogeneous bottom
  • save it as mask_10.shp
  • then run -Classify/Typer10_0.2
  • and so on with type_11, type_12 etc



 
  
   type 10
a nice clearcut AOI
  • CoefSIG=0.2 : the white lines represent
    • UP........................LB*1.02
    • MIDDLE..............LB
    • Bottom..................LB/1.02
  • This opens a tolerance window, which may be set tight or wide at your whim 
 
  type 21
a more complex case
  • CoefSIG=0.2 : the white lines represent
    • UP........................LB*1.02
    • MIDDLE..............LB
    • Bottom..................LB/1.02
  • This opens a tolerance window, which may be set tight or wide at your whim 
Once you have enough,
then run the Classifier-Model
  • -Classify/Classifier_0.5_0_ef_LastChType2
  • choose a low value for CoefSIG, like 0.1 or 0.2
    • so that the need for more intermediate or outsider classes becomes obvious
  • You want your AOIs to be pure clearcut cases: they call then "endmembers"
  • ==> Do not hesitate to relocate any AOI where the provisory classified image exhibits a more comfortable/homogeneous candidate site
  • ==> See whether you need more classes
-Model
-Classify/Classifier_0.5_0_ef_LastChType2
  • this elaborate syntax is overbearing when proceessing a BGRN image, but it becomes handy when processing an hyperspectral image!!!
  • you need both -Model and -Classify... turned ON
  • 4SM runs as normal (don't mess with the calibration parameters though!)
    • except that it also classifies each pixel "on the fly" as they nicely say (takes more time)
  • Where two or more classes pretend they belong, well 4SM reduces CoefSIG until only one wins the case
  • With a hyperspectral signature, a number of MISS is tolerated
  • nbMISS here is set at 0, as only the Green and the Blue bands are used (LastChType2) for bottom typing (as Red is entirely determined by Blue and Green)
    • their average and their B/G ratio is enough to tell several candidates apart

What do you think?
Seems that Mahalanobis and colleagues
are a bit "too much" under water!

And there is this nice perspective of "contextual editing" to help you out!
 
 
 
 
 
CoefSIG
Syntax_Classify
  • a high value is very inclusive
    • achieves a high rate of classified pixels
    • and eventually of confusion
  • a low value is very restrictive
    • reveals areas which deserve a bottom type of their own
    • mostly shows thin transition between two well established bottom types
 

CoefSIG=0.1
-Classify/Classifier_0.1_0_ef_LastChType2
  • 15 classes and CoefSIG=0.1
  • two more classes shall be created
  • CoefSIG shall be increased to 0.5

CoefSIG=0.5
-Classify/Classifier_0.5_0_ef_LastChType2
  • two more classes have been added
  • there is still a small unclassified patch which might need a class of its own

CoefSIG=0.1 under Mask 4
  • bottom type AOIs 20, 21, 22 and 23 are shown
  • they are located where the three bands case applies

CoefSIG=0.5 under Mask 4
  • bottom type AOIs 20, 21, 22 and 23 are shown
  • they are located where the three bands case applies

 
 
 
 
 
 
 
RLBgb=1.22 for the whole image
-Z/MSL0.00/n_1/RLBgb1.220/mask_4
nice -20  ./4SM.4.08 -Process/nosu/14_Aug_2011/20_AUG_2011  -OriginFUGRO-NPA               @@
-dB/tmnov/31_1_0_0/4_1/1265_1559/208.365_2398.755/1_1               
               @@
-Mis/Bahamas/Caicos/Landsat/TM/FUGRO-NPA/UTM_19_008/0.030_0.030/22_NOV_1990                       @@
-LS/0255.0/255.0/255.0/255.0                           @@
-cWL/0.500/0.602/0.500/0.500_AutoCAL                       @@
-M/@000001/00002/00003/00004                           @@
-LsM/219.2/120.0/135.6/111.1_cLM=1.130                   @@
-Lsw/60.42/17.41/10.79/06.50                               @@
-dLsw000.0/000.8/000.0/000.0                   @@
-Lw/0010.6/001.5/000.0/000.0                           @@
-Lm/0001.0/001.0/002.0/255.0                   @@
-KK1_2_0.5195/Knir4.500/mask_3_AutoCAL                       @@
-Z/MSL0.00/n_1/RLBgb1.220/mask_4                   @@
-B/tclNe/Bmin0/cLM1 .13                           @@
-deglint/vRbaD/mDEGLINT19/GlintM255.0                   @@
-Model               @-DD324/122                 @@
-Smooth/5/D/Smart+                           @@
-Classify/Typer17_0.2 /Classifier_0.1_0_ef_LastChType2 /mode_5_11_16       
 

type 10

type 11

type 12

type 13

type 14

type 15

type 16

type 17
 
=========== STATISTICS OF ALL SHALLOW BOTTOM TYPING: BOA radiance 0-200
=========== CoefSIG=0.10 nbMISS=0 Fast=0 -E1/1265/1/1559  on Sun Aug 21 19:08:15 2011

             6.08% Type_0 is Land pixel
             2.11% Type_1 is Unclassified Shallow pixel
            20.12% Type_2 is Optically Deep water pixel
             2.66% Type_3 is No Data pixel
             0.00% Type_4 is Wave Breaker pixel
             0.00% Type_5 is Cloud/Shadow pixel
                   Type_6 and above is for Classified Shallow pixel

Classified by signature_number from 6 to 17
 5.24% as Type_10 B=178.0 G/B=1.028  178.0 183.0 180.5   0.0
21.67% as Type_11 B=146.2 G/B=1.008  146.2 147.3 146.7   0.0
26.47% as Type_12 B=113.0 G/B=1.173  113.0 132.6 122.8   0.0
 9.97% as Type_13 B=81.0  G/B=1.168   81.0  94.6  87.8   0.0
 1.25% as Type_14 B=61.8  G/B=1.372   61.8  84.8  73.3   0.0
 2.49% as Type_15 B=56.5  G/B=1.163   56.5  65.7  61.1   0.0
 1.19% as Type_16 B=45.3  G/B=1.296   45.3  58.7  52.0   0.0
 0.75% as Type_17 B=34.4  G/B=1.148   34.4  39.5  36.9   0.0

Classified from the brightest to the darkest
 5.24% as Type_10 B=178.0 G/B=1.028  178.0 183.0 180.5   0.0
21.67% as Type_11 B=146.2 G/B=1.008  146.2 147.3 146.7   0.0
26.47% as Type_12 B=113.0 G/B=1.173  113.0 132.6 122.8   0.0
 9.97% as Type_13 B=81.0  G/B=1.168   81.0  94.6  87.8   0.0
 1.25% as Type_14 B=61.8  G/B=1.372   61.8  84.8  73.3   0.0
 2.49% as Type_15 B=56.5  G/B=1.163   56.5  65.7  61.1   0.0
 1.19% as Type_16 B=45.3  G/B=1.296   45.3  58.7  52.0   0.0
 0.75% as Type_17 B=34.4  G/B=1.148   34.4  39.5  36.9   0.0

Classified from the bluest to the greenest
21.67% as Type_11 B=146.2 G/B=1.008  146.2 147.3 146.7   0.0
 5.24% as Type_10 B=178.0 G/B=1.028  178.0 183.0 180.5   0.0
 0.75% as Type_17 B=34.4  G/B=1.148   34.4  39.5  36.9   0.0
 2.49% as Type_15 B=56.5  G/B=1.163   56.5  65.7  61.1   0.0
 9.97% as Type_13 B=81.0  G/B=1.168   81.0  94.6  87.8   0.0
26.47% as Type_12 B=113.0 G/B=1.173  113.0 132.6 122.8   0.0
 1.19% as Type_16 B=45.3  G/B=1.296   45.3  58.7  52.0   0.0
 1.25% as Type_14 B=61.8  G/B=1.372   61.8  84.8  73.3   0.0

Classified from the most abundant to the least abundant
26.47% as Type_12 B=113.0 G/B=1.173  113.0 132.6 122.8   0.0
21.67% as Type_11 B=146.2 G/B=1.008  146.2 147.3 146.7   0.0
 9.97% as Type_13 B=81.0  G/B=1.168   81.0  94.6  87.8   0.0
 5.24% as Type_10 B=178.0 G/B=1.028  178.0 183.0 180.5   0.0
 2.49% as Type_15 B=56.5  G/B=1.163   56.5  65.7  61.1   0.0
 1.25% as Type_14 B=61.8  G/B=1.372   61.8  84.8  73.3   0.0
 1.19% as Type_16 B=45.3  G/B=1.296   45.3  58.7  52.0   0.0
 0.75% as Type_17 B=34.4  G/B=1.148   34.4  39.5  36.9   0.0

These statistics include all pixels located under Mask
 
 
 
 
RLBgb=1.09 under mask_4
-Z/MSL0.00/N_1/RLBgb1.090/Mask_4
nice -20  ./4SM.4.08 -Process/nosu/20_Aug_2011/21_AUG_2011  -OriginFUGRO-NPA                @@
-dB/tmnov/31_1_0_0/4_1/1265_1559/208.365_2398.755/1_1               
                @@
-Mis/Bahamas/Caicos/Landsat/TM/FUGRO-NPA/UTM_19_008/0.030_0.030/22_NOV_1990                       @@
-LS/0255.0/255.0/255.0/255.0                           @@
-cWL/0.500/0.602/0.500/0.500_AutoCAL                       @@
-M/@000001/00002/00003/00004                           @@
-LsM/219.2/120.0/135.6/111.1_cLM=1.130                   @@
-Lsw/60.42/17.41/10.79/06.50                               @@
-dLsw000.0/000.0/000.0/000.0                                 @@
-Lw/0010.6/001.5/000.0/000.0                           @@
-Lm/0001.0/001.0/002.0/255.0                   @@
-KK1_2_0.5195/Knir4.500/mask_3_AutoCAL                       @@
-Z/MSL0.00/N_1/RLBgb1.090/Mask_4                     @@
-B/tclNe/Bmin0/cLM1 .13                           @@
-deglint/vRbaD/mDEGLINT19/GlintM255.0                   @@
-Model                                @@
-Smooth/5/D/Smart+                           @@
-Classify/typer23_0.2/Classifier_0.2_0_ef_LastChType2/mode_5_11_16                       @@


 

type 20

type 21

type 22

type 23
 
=========== STATISTICS OF ALL SHALLOW BOTTOM TYPING: BOA radiance 0-200
=========== CoefSIG=0.50 nbMISS=0 Fast=0 -E1/1265/1/1559  on Sun Aug 21 20:37:49 2011

             0.00% Type_0 is Land pixel
             0.29% Type_1 is Unclassified Shallow pixel
            13.47% Type_2 is Optically Deep water pixel
             0.00% Type_3 is No Data pixel
             0.00% Type_4 is Wave Breaker pixel
             0.00% Type_5 is Cloud/Shadow pixel
                   Type >5 is for Classified Shallow pixel

Classified by signature_number from 6 to 23 under Mask_4
 3.46% as Type_20 B=114.8 G/B=1.122  114.8 128.8 121.8   0.0
17.80% as Type_21 B=72.6  G/B=1.255   72.6  91.1  81.8   0.0
28.36% as Type_22 B=30.8  G/B=1.240   30.8  38.2  34.5   0.0
36.62% as Type_23 B=19.0  G/B=1.279   19.0  24.3  21.6   0.0

Classified from the brightest to the darkest under Mask_4
 3.46% as Type_20 B=114.8 G/B=1.122  114.8 128.8 121.8   0.0
17.80% as Type_21 B=72.6  G/B=1.255   72.6  91.1  81.8   0.0
28.36% as Type_22 B=30.8  G/B=1.240   30.8  38.2  34.5   0.0
36.62% as Type_23 B=19.0  G/B=1.279   19.0  24.3  21.6   0.0

Classified from the bluest to the greenest under Mask_4
 3.46% as Type_20 B=114.8 G/B=1.122  114.8 128.8 121.8   0.0
28.36% as Type_22 B=30.8  G/B=1.240   30.8  38.2  34.5   0.0
17.80% as Type_21 B=72.6  G/B=1.255   72.6  91.1  81.8   0.0
36.62% as Type_23 B=19.0  G/B=1.279   19.0  24.3  21.6   0.0

Classified from the most abundant to the least abundant under Mask_4
36.62% as Type_23 B=19.0  G/B=1.279   19.0  24.3  21.6   0.0
28.36% as Type_22 B=30.8  G/B=1.240   30.8  38.2  34.5   0.0
17.80% as Type_21 B=72.6  G/B=1.255   72.6  91.1  81.8   0.0
 3.46% as Type_20 B=114.8 G/B=1.122  114.8 128.8 121.8   0.0
 
 
 
 
Final bottom typing image
 

RLBgb1.22 : it is clearly seen that
  • the above series of bottom type signatures are not convenient for the outer reef area under mask_4, where many pixels remain unclassified

FINAL
  • the area under mask_4 has been processed separately
  • there is hardly any need for modal filtering
  • contextual editing might quite deservedly help in reducing this "diversity" from just a Landsat TM image