4SM TM tmnov-tutorial-prepare-your-image-2

Output of the AutoCal process
Glint regressions and mask mSE

Once the shapefiles are have been prepared
and the command line is specified:
it takes five seconds to get the following result:
Run the script

True color composite shapefiles are shown	Special effect Mask mSE
Glint regressions	Glint regressions There is hardly any swell-modulated glint in this image Still, we want to wipe out a few lumps of haze that surround scanty clouds Just a very small sample of cloud/haze is enough to work out a very strong set of glint correlations Deep water radiance is sampled under a -very small- cloud shadow This then allows 4SM to offer its pick for the -Lsw... argument

Output of the AutoCal process
KIKJregression

AutoCalibration diagram K1/K2 by regression in AutoCal -AutoCAL/R2_0.95/Land_-5.0_1.18_4s/GlintM25 tmnov_AutoCAL_4_3_2_1_m0_K1K2regressed This AutoCalibration is near-perfect: Just Knir=4.5 m-1 should be decreased in order to achieves a nice fit of the BPL for the Red/NIR pair of bands.	AutoCalibration diagram K1/K2=0.5 enforced by 4SM -AutoCAL/R2_1.00/Land_-5.0_1.18_4s/GlintM25 tmnov_AutoCAL_4_3_2_1_m0_K1K2=0.50 This AutoCalibration needs attention
?Regressed or Default? -AutoCAL/R2_0.95/Land_-5.0_1.18_4s/GlintM25 This is controlled by the R2KKmin variable in the -AutoCAL... argument.	?Regressed or Default? -AutoCAL/R2_1.00/Land_-5.0_1.18_4s/GlintM25 Try .../R2_1.00/... see what happens.

Output of the AutoCal process
AutoCalibration diagram

AutoCalibration diagram K1K2 regressed This AutoCalibration is outstanding just Knir=4.5 should be decreased in order to achieve a nice fit of the BPLfor the Red/NIR pair of bands. and Lw should be checked Excellent fit of the BPL model has been achieved K1/K2=0.5141 WaterType OIB+0.68 Kblue =0.107 m-1 Kgreen=0.208 m-1 Kred =0.649 m-1 CoefK=1.0 Knir =4.500 m-1 shoud be decreased Apart from a "hump" at ~2.5m BPL pixels exhibit neutral spectral signature, as revealed by white squares (ghost pixels) Exceptional Soil Line by NIR (small white circles) thanks to abundant very bright exposed mud/sand flats subtidal very dark dessicated microbial mats marshes see a FCC of the area DPL pixels (small crosses) reveal that green subtidal substrates are common/abundant Healthy vegetation on land provide abundant evidence to help estimate spectral La and therefore Lw, as evidenced by the bi-dimensional histograms	AutoCalibration diagram K1K2=0.50 This AutoCalibration is very good but many details need to be improved The BPL model just happens to be correct here K1/K2=0.50 WaterType OIB+0.6 Kblue =0.103 m-1 Kgreen=0.206 m-1 Kred =0.646 m-1 CoefK=1.0 Knir =4.500 m-1 shoud be decreased Apart from a "hump" at ~2.5m BPL pixels exhibit neutral spectral signature, as revealed by white squares (ghost pixels) Exceptional Soil Line by NIR (small white circles) thanks to abundant very bright exposed mud/sand flats subtidal very dark dessicated microbial mats marshes see a FCC of the area DPL pixels (small crosses) reveal that green subtidal substrates are common/abundant Healthy vegetation on land provide abundant evidence to help estimate spectral La and therefore Lw, as evidenced by the bi-dimensional histograms
DPL pixels Syntax_Calibrate Small crosses are DPL pixels They represent the shallowest pixels over the whole range of bottom substrate brightness It would make our life easy if they all would represent spectrally neutral reflectance (shades of gray, along the Soil Line) In contrast, what we see here is a series of very green DPL pixels with radiances 35-90 in the Green band: this is a feeding ground for turtles and dugongs, which extends over very shallow bottoms this shall be the cause of severe depth under-estimation by the Green solution	Ghost BPL pixels Syntax_Calibrate Small white squares are "ghost" BPL pixels They denote the fact that the BPL pixels which have been extracted are not spectrally contrasted, as would be expected for clean coral or oolithic sand
-Calibrate/voAI/bDsNpzg/..... b...........do not show..............BPL pixels D......................show..............DPL pixels s............do not show..............Soil Line z............do not show..............ZZ insets g............do not show..............Ghost pixels	-Calibrate/voAI/BdsNpzG/... B......................show..............BPL pixels d............do not show..............DPL pixels s............do not show..............Soil Line z............do not show..............ZZ insets G......................show..............Ghost pixels
howto_read_calibration_textfile gedit tmnov_m0.cal find the location of these BPL pixels in textfile tmnov_m0.cal Row 938 is in column 4 Line 341 is in column 5 This is a real BPL pixel for the pair of bands 1/2 bandI............1 bandJ............2 mSE..............2 Lsblue..........203.0 Lsgreen........110.0 Lsred............121.0 Lsnir..............27 .0	howto_read calibration_textfile M_002 1 2 925 334 209.0 116.0 128.0 85.0 M_002 1 2 927 333 208.0 115.0 126.0 84.0 M_002 1 2 937 345 209.0 114.0 126.0 75.0 M_002 1 2 931 337 207.0 113.0 125.0 80.0 M_002 1 2 932 337 204.0 112.0 124.0 76.0 M_002 1 2 938 341 203.0 111.0 121.0 66.0 M_002 1 2 678 31 207.0 110.0 121.0 87.0 M_002 1 2 679 31 208.0 109.0 120.0 85.0 M_002 1 2 680 32 207.0 108.0 118.0 80.0 M_002 1 2 684 31 205.0 107.0 118.0 81.0 M_002 1 2 998 548 204.0 106.0 100.0 27.0 ....

**Hump at ~2 m**
Remember: "Apart from a "hump" at ~2m": see below that the culprits sit on a sand spit! Note the haze and the small clouds: they shall need deglinting

**Mapping BPL pixels**
BPL pixels are mapped in channels 16, 17 and 18 by the extraction process apart from the sand spit, most BPL pixels are conveniently scattered all over the image subset

.m0.

Mask
m0 or m3

dir *m0*

m0 in those file names means that no mask was used to produce them
m3 would mean that the action was confined under mask_3
mask_3 to mask_20 may be used

Mask
m0 or m3

dir *m0*

tmnov_4_3_2_1_m0.eps
tmnov_AutoCal_4_3_2_1_m0.eps
tmnov_m0_255.gli
tmnov_m0.bdh
tmnov_m0.cal