4SM negril85-tutorial-calibration1

see textfile autoCAL.txt

#Following is the simplified AutoCAL CommandLine=====mBPL=1====================
nice -20 4SM.4.08 -Process/17_Sep_2011/26_AUG_2011 -OriginUofUpsala
-DB/negril85/31_1_0_0/4_1/500_500/98.362_192.214/1_1
-Mis/Jamaica/Negril/Landsat/TM/UpsalaU/UTM_19_008/0.030_0.030/22_NOV_1990
-LS/0255.0/255.0/255.0/255.0
-cWL/0.500/0.622/0.500/0.500_AutoCAL
-M/@000001/00002/00003/00004
@CP/0000.0/000.0/000.0/000.0_1.26m
-LsM/166.9/083.4/117.9/096.7_AutoCAL
-Lsw/066.0/016.9/011.0/004.0
-dLsw000.0/000.0/000.0/000.0
-Lw/0007.5/001.5/000.0/000.0
-Lm/0001.0/001.0/001.0/255.0
-KK1_2_0.8076/Knir4.500/mask_3
-Z/MSL0.00/n_1/RLBgb1.000_1.00/mask_3
-B/tclNe/Bmin0/cLM1.00
@deglint/vRbaD/GlintM25.0
-extract/v/mBPL1/mask_3
-Calibrate/v/BdSNpZG/BDh_12_13_34
-Model/mask_3
-Smooth/0/d/smart-
@-E1/500/1/500    @-E1/500/1/500
@-DN300/100/3/comment @-DD550/125/4/comment

disable the AutoCAL script in negril85.sh
and copy this script into negril85.sh

#nice -20 4SM.4.08 -Process/nosu/26_AUG_2011 -Origin/UofUpsala
-DB/negril85/31_1_0_0/4_1/1265_1559/208.365_2398.755/1_1
-Mis/Jamaica/Negril/Landsat/TM/UpsalaU/UTM_19_008/0.030_0.030/22_NOV_1990
-MakePIX
-Import/data*negril85/dbnc_4_0_0_0/R500_L500/Origin_98.347_192.229_0_0/chIn1,4/chOut1,4 @/import.shp
-AutoCAL/NIRmax255/mBPL1/NoSmooth

#exit

#Following is the simplified AutoCAL CommandLine=====mBPL=1====================
nice -20 4SM.4.08 -Process/17_Sep_2011/26_AUG_2011 -OriginUofUpsala
-DB/negril85/31_1_0_0/4_1/500_500/98.362_192.214/1_1
-Mis/Jamaica/Negril/Landsat/TM/UpsalaU/UTM_19_008/0.030_0.030/22_NOV_1990
-LS/0255.0/255.0/255.0/255.0
-cWL/0.500/0.622/0.500/0.500_AutoCAL
-M/@000001/00002/00003/00004
@CP/0000.0/000.0/000.0/000.0_1.26m
-LsM/166.9/083.4/117.9/096.7_AutoCAL
-Lsw/066.0/016.9/011.0/004.0
-dLsw000.0/000.0/000.0/000.0
-Lw/0007.5/001.5/000.0/000.0
-Lm/0001.0/001.0/001.0/255.0
-KK1_2_0.8076/Knir4.500/mask_3
-Z/MSL0.00/n_1/RLBgb1.000_1.00/mask_3
-B/tclNe/Bmin0/cLM1.00
@deglint/vRbaD/GlintM25.0
-extract/v/mBPL1/mask_3
-Calibrate/v/BdSNpZG/BDh_12_13_34
-Model/mask_3
-Smooth/0/d/smart-
@-E1/500/1/500    @-E1/500/1/500
@-DN300/100/3/comment @-DD550/125/4/comment

calibration1_1
straight from AutoCAL

	One step at a time First get LsM straight -Calibrate/v/BdSNpzg/BDh_12_13_34 z to get rid if the ZZ inserts g to get rid if the Ghost pixels -LsM/166.9/083.4/117.9/156.7 to get LsM_nir straight -LsM/166.9/091.4/117.9/156.7 to get LsM_green straight -LsM/156.9/091.4/117.9/156.7 to get LsM_blue straight
	One step at a time Then increase LsM -B/tclNe/Bmin0/cLM1.36 then -LsM/192.3/118.8/156.4/211.7_cLM=1.360 and -B/tclNe/Bmin0/cLM1 .36
	Then adjust the BPL model to the BPL pixels -KK1_2_0.5900/Knir4.500/mask_3
	All done Some other aspects of the calibration shall be dealt with later on Ready for first modeling -calibrate/v/BdSNpzg/BDh_12_13_34 -Smooth/5/d/Smart-

Most BPL pixels
are located along the beach
that means a lot...

BPL pixels are mapped in channels_16_17_18
NoSmoothing

	Notice that five fairly deep BPL pixels are precisely aligned with a slope of K1/K2=0.59 This is quite quite remarquable This is likely to denote a uniform and bright bottom strip ~15 m deep or at least the brightest bottom type being present at scattered pixels in the 10-15 m depth range Two more BPL pixels are on the same straight line at a depth of ~5-6 m ================== These five pixels appear in negril85_m0.cal HowTo read the calibration textfile ... M_002 1 2 269 245 90.0 23.0 10.0 6.0 M_002 1 2 289 152 87.0 22.0 11.0 6.0 M_002 1 2 271 259 85.0 21.0 10.0 5.0 M_002 1 2 264 248 82.0 20.0 11.0 6.0 M_002 1 2 260 295 79.0 19.0 10.0 4.0 ....
==> we have very reason to trust the -KK1_2_0.5900... specification the -LsM/192.2/118.7... specification	We can make them Calibration Pixels in the commandline Syntax CP .... -M/@000001/00002/00003/00004 -CP00090.0/023.0/011.0/005.0 -CP00087.0/022.0/011.0/005.0 -CP00085.0/021.0/011.0/005.0 -CP00082.0/020.0/011.0/005.0 -CP00079.0/019.0/011.0/005.0 -LsM/163.1/080.4/113.6/096.7_AutoCAL .... This result is obtained in the 10-14 m depth range. K_blue/K_green=0.59 is a nice statement, knowing that these pixels are genuine NotSmoothed DNs system noise is high in TM data waters in this heavily developped residential area might not be quite as homogeneous as assumed

-smooth/5/Smart+	-Smooth/5/Smart+

Image_B without any smoothing : @-Smooth/5/Smart+
White circles show the location of the BPL pixels in the Blue/Green pair
The 4SM Extraction process picks up isolated bright bottomed BPL pixels,
which are located in small sized clearances in vegetated bottoms.
No hand-picking of arguably bright bottoms at various depths ,
Lyzenga style,
could have achieved that remarkable result!

Such small patches are smoothed out by the Smoothing process,
even though a "smart" one.

nice -20 4SM.4.08 -Process -Origin/UofUpsala
-DB/negril85/31_1_0_0/4_1/500_500/98.362_192.214/1_1
-Mis/Jamaica/Negril/Landsat/TM/UpsalaU/UTM_19_008/0.030_0.030/22_NOV_1990
-LS/0255.0/255.0/255.0/255.0
-cWL/0.500/0.599/0.500/0.500_AutoCAL
-M/@000001/00002/00003/00004
-CP00090.0/023.0/011.0/005.0
-CP00087.0/022.0/011.0/005.0
-CP00085.0/021.0/011.0/005.0
-CP00082.0/020.0/011.0/005.0
-CP00079.0/019.0/011.0/005.0
-LsM/163.1/080.4/113.6/096.7_AutoCAL
-LsM/156.9/091.4/113.6/156.7
-LsM/192.2/118.7/150.4/211.0_cLM=1.360
-Lsw/066.5/017.1/011.3/005.9_AutoCAL
-dLsw000.0/000.0/000.0/000.0
-Lw/0007.7/001.5/000.0/000.0_AutoCAL
-Lm/0001.0/001.0/001.0/255.0
-KK1_2_0.8271/Knir4.500/mask_3
-KK1_2_0.5900/Knir4.500/mask_3
-Z/MSL0.00/n_1/RLBgb1.000_1.000/mask_3
-B/tclNe/Bmin0/cLM1 .36
@deglint/vRbaD/GlintM25.0
-extract/v/mBPL1/mask_3
-calibrate/v/BdSNpzG/BDh_12_13_34
-Model/mask_3
@-Smooth/5/Smart+

Healthy vegetation
and
Water Volume Reflectance

vegetation.shp over a backdrop of FCC

Vegetation.shp shapefile

The red, green and blue dots in the calibration diagram represent the average healthy vegetation which is delineated by the vegetation.shp shapefile

Healthy vegetation is expected to be high in Green, and low in Blue and Red

We use this to help estimate Lw_blue

	Lw_blue See that Lw_blue=7.5 appears to be convenient as this value ensures that the SoilsLine model strikes right through the blue dot which represents the vegetation Lw_blue=7.5 over a range of 100 DNs BOA makes 7.5% of the reflectance of the local brand of bright bottoms La_blue = 66 - 7.5 = 58.5 is the atmospheric path radiance
	The vegetation is plentyfull in this image, this is quite remarquable in this bidimensional histogram Blue and Red bands are distinctly lower than Green for a vegetation pixel Making the SoilsLine model fit the vegetation pixels in this bidim histogram is a sensible thing to do This provides a means to ballpark the atmospheric path radiance, and therefore also the water volume reflectance It takes of Lw_blue >>> Lw_green to achieve such a marked curvature Should Lw_blue be ~= Lw_green, then the display of land pixels, vegetated or not, would be in a diagonal position

	Lw_green As for the green band, it is fitting that the green dot, which represents the vegetation, is located well above the SoilsLine model, i.e. on the green side
	The vegetation is plentyfull in this image, this is quite remarquable in this bidimensional histogram Red band is distinctly lower than Green band for a vegetation pixel Making the SoilsLine model strike well below the vegetation pixels in this bidim histogram is a sensible thing to do This provides a means to ballpark the atmospheric path radiance, and therefore also the water volume reflectance It takes of Lw_green > Lw_green to achieve just a slight curvature See that most bland pixels are located above the diagonal, i.e. on the green side Should Lw_green be ~= Lw_red, then the display of land pixels, vegetated or not, would be in a diagonal position

	Lw_red Lw_red and Lw_nir are assumed to be null See that the vegetation is clearly a case of negative bottom contrast in the blue band
	Summary for Blue waters Lw_nir=0 Lw_red=0 Lw_green just one or two DNs Lw_blue much higher For an extreme case of Blue waters, see see Kauai in the Hawaiian Islands where Lw_blue is as high I've ever seen where Lw_green is as high I've ever seen

LwN in calibration diagram

At Kauai Kauai is a remote volcanic island in the open ocean Pristine Pacific waters there K₁/K₂~=0.37 Even Lw_green is very high LwN_blue.........79.5 LwNgreen......18.4 =============== At Negril Negril is a popular resort area in Jamaica Affected by landmass and human activities Waters not so clear K₁/K₂~=0.59 LwN_blue.........11.8 LwN_green.......2.9 ========== Although the Kauai figures are not finalized, and there probably are no coral sands there, they show the trend as regards water volume reflectance	Water volume reflectance is reported to reach up to in excess of 12% In a calibrated model in 4SM, radiances are converted to BOA Ls=Ls-La then scaled in the range 0-200 Ls=(Ls-La) * 200/(LsM-La) This amounts to a process of normalization LwN = Lw * 200/(LsM-La) this allows the practioner to compare LwN values among various images except that LsM-La values themselves are not normalized (this would require conversion to units of reflectance)

We cannot afford to just ignore Lw ,
even in a "simplified" radiative transfer model

More research must be devoted to this subject,
as there are certainly many factors which determine Lw

Afficher la suite de cette page