Influence of atmospheric and sea-surface corrections on retrieval of bottom depth and reflectance using a semi-analytical model: a case study in Kaneohe Bay, Hawaii

James A. Goodman, ZhongPing Lee, and Susan L. Ustin

James A. Goodman,^{1,}^{*} ZhongPing Lee,^{2} and Susan L. Ustin^{3}

^{1}Bernard M. Gordon Center for Subsurface Sensing and Imaging Systems, University of Puerto Rico at Mayagüez, P.O. Box 9048, Mayagüez, Puerto Rico 00681, USA

^{2}U.S. Naval Research Laboratory, Code 7333, Stennis Space Center, Mississippi 39529, USA

^{3}Department of Land, Air and Water Resources, Center for Spatial Technologies and Remote Sensing, University of California, Davis, One Shields Avenue, Davis, California 95616, USA

James A. Goodman, ZhongPing Lee, and Susan L. Ustin, "Influence of atmospheric and sea-surface corrections on retrieval of bottom depth and reflectance using a semi-analytical model: a case study in Kaneohe Bay, Hawaii," Appl. Opt. 47, F1-F11 (2008)

Hyperspectral instruments provide the spectral detail necessary for extracting multiple layers of information from inherently complex coastal environments. We evaluate the performance of a semi-analytical optimization model for deriving bathymetry, benthic reflectance, and water optical properties using hyperspectral AVIRIS imagery of Kaneohe Bay, Hawaii. We examine the relative impacts on model performance using two different atmospheric correction algorithms and two different methods for reducing the effects of sunglint. We also examine the impact of varying view and illumination geometry, changing the default bottom reflectance, and using a kernel processing scheme to normalize water properties over small areas. Results indicate robust model performance for most model formulations, with the most significant impact on model output being generated by differences in the atmospheric and deglint algorithms used for preprocessing.

Curtis D. Mobley, Lydia K. Sundman, Curtiss O. Davis, Jeffrey H. Bowles, Trijntje Valerie Downes, Robert A. Leathers, Marcos J. Montes, William Paul Bissett, David D. R. Kohler, Ruth Pamela Reid, Eric M. Louchard, and Arthur Gleason Appl. Opt. 44(17) 3576-3592 (2005)

Luis Guanter, Rudolf Richter, and José Moreno Appl. Opt. 45(10) 2360-2370 (2006)

References

You do not have subscription access to this journal. Citation lists with outbound citation links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

You do not have subscription access to this journal. Article level metrics are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

SHOALS Data Versus Estimated Bathymetry (Correlation Coefficient, r; Slope of Best Fit Curve, m; absolute error, a) for $0\u201320\text{\hspace{0.17em}}\mathrm{m}$ Water Depth Using Different Preprocessing Algorithms for Atmospheric Correction and Deglint^{
a
}

r

m

a

NO Deglint

ACORN (no artifact suppression)

$-0.104$

$-0.00$

$6.29\text{\hspace{0.17em}}\mathrm{m}$

ACORN (artifact suppression)

$-0.131$

$-0.00$

$6.29\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (deep water subset)

0.682

0.80

$3.20\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (full geometry)

0.894

1.03

$1.70\text{\hspace{0.17em}}\mathrm{m}$

Hochberg et al. Deglint

ACORN (no artifact suppr.)

$-0.234$

$-0.01$

$6.17\text{\hspace{0.17em}}\mathrm{m}$

ACORN (artifact suppr.)

$-0.253$

−0.01

$6.19\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (deep water subset)

0.486

0.81

$4.58\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (full geometry)

0.780

1.27

$3.74\text{\hspace{0.17em}}\mathrm{m}$

750 Normalizing Deglint

ACORN (no artifact suppr.)

0.912

0.80

$1.26\text{\hspace{0.17em}}\mathrm{m}$

ACORN (artifact suppr.)

0.795

0.74

$1.76\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (deep water subset)

0.907

1.13

$2.00\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (full geometry)

0.902

1.03

$1.64\text{\hspace{0.17em}}\mathrm{m}$

All model runs were performed using full geometry for view and illumination angles and an average sand spectrum as the default benthic input.

Table 2

SHOALS Data Versus Estimated Bathymetry (Correlation Coefficient, r; Slope of Best Fit Curve, m; Absolute Error, a) for $0\u201320\text{\hspace{0.17em}}\mathrm{m}$ Water Depth Using Different Subsurface View and Illumination Geometries^{
a
}

r

m

a

Nadir

0.902

1.04

$1.66\text{\hspace{0.17em}}\mathrm{m}$

Full Geometry

0.902

1.03

$1.64\text{\hspace{0.17em}}\mathrm{m}$

Both model runs were performed using full geometry Tafkaa, $750\text{\hspace{0.17em}}\mathrm{nm}$ normalizing deglint, and an average sand spectrum as the default benthic input.

Table 3

SHOALS Data Versus Estimated Bathymetry (Correlation Coefficient, r; Slope of Best Fit Curve, m; Absolute Error, a) for $0\u201320\text{\hspace{0.17em}}\mathrm{m}$ Water Depth Using Different Default Bottom Spectra as Model Input^{
a
}

r

m

a

Sand

0.902

1.03

$1.64\text{\hspace{0.17em}}\mathrm{m}$

Coral

0.902

0.99

$1.48\text{\hspace{0.17em}}\mathrm{m}$

Algae

0.902

1.00

$1.50\text{\hspace{0.17em}}\mathrm{m}$

Flat

0.901

0.96

$1.40\text{\hspace{0.17em}}\mathrm{m}$

All model runs were performed using full geometry Tafkaa, $750\text{\hspace{0.17em}}\mathrm{nm}$ normalizing deglint, and full geometry for view and illumination angles.

Table 4

SHOALS Data Versus Estimated Bathymetry (Correlation Coefficient, r; Slope of Best Fit Curve, m; Absolute Error, a) for $0\u201320\text{\hspace{0.17em}}\mathrm{m}$ Water Depth Using Different Spatial Kernels for Averaging Water Properties^{
a
}

r

m

a

$1\times 1$

0.902

1.03

$1.64\text{\hspace{0.17em}}\mathrm{m}$

$3\times 3$

0.801

0.85

$1.89\text{\hspace{0.17em}}\mathrm{m}$

$5\times 5$

0.801

0.81

$1.88\text{\hspace{0.17em}}\mathrm{m}$

All model runs performed using full geometry Tafkaa, $750\text{\hspace{0.17em}}\mathrm{nm}$ normalizing deglint, full geometry for view and illumination angles, and an average sand spectrum as the default benthic input.

Tables (4)

Table 1

SHOALS Data Versus Estimated Bathymetry (Correlation Coefficient, r; Slope of Best Fit Curve, m; absolute error, a) for $0\u201320\text{\hspace{0.17em}}\mathrm{m}$ Water Depth Using Different Preprocessing Algorithms for Atmospheric Correction and Deglint^{
a
}

r

m

a

NO Deglint

ACORN (no artifact suppression)

$-0.104$

$-0.00$

$6.29\text{\hspace{0.17em}}\mathrm{m}$

ACORN (artifact suppression)

$-0.131$

$-0.00$

$6.29\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (deep water subset)

0.682

0.80

$3.20\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (full geometry)

0.894

1.03

$1.70\text{\hspace{0.17em}}\mathrm{m}$

Hochberg et al. Deglint

ACORN (no artifact suppr.)

$-0.234$

$-0.01$

$6.17\text{\hspace{0.17em}}\mathrm{m}$

ACORN (artifact suppr.)

$-0.253$

−0.01

$6.19\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (deep water subset)

0.486

0.81

$4.58\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (full geometry)

0.780

1.27

$3.74\text{\hspace{0.17em}}\mathrm{m}$

750 Normalizing Deglint

ACORN (no artifact suppr.)

0.912

0.80

$1.26\text{\hspace{0.17em}}\mathrm{m}$

ACORN (artifact suppr.)

0.795

0.74

$1.76\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (deep water subset)

0.907

1.13

$2.00\text{\hspace{0.17em}}\mathrm{m}$

Tafkaa (full geometry)

0.902

1.03

$1.64\text{\hspace{0.17em}}\mathrm{m}$

All model runs were performed using full geometry for view and illumination angles and an average sand spectrum as the default benthic input.

Table 2

SHOALS Data Versus Estimated Bathymetry (Correlation Coefficient, r; Slope of Best Fit Curve, m; Absolute Error, a) for $0\u201320\text{\hspace{0.17em}}\mathrm{m}$ Water Depth Using Different Subsurface View and Illumination Geometries^{
a
}

r

m

a

Nadir

0.902

1.04

$1.66\text{\hspace{0.17em}}\mathrm{m}$

Full Geometry

0.902

1.03

$1.64\text{\hspace{0.17em}}\mathrm{m}$

Both model runs were performed using full geometry Tafkaa, $750\text{\hspace{0.17em}}\mathrm{nm}$ normalizing deglint, and an average sand spectrum as the default benthic input.

Table 3

SHOALS Data Versus Estimated Bathymetry (Correlation Coefficient, r; Slope of Best Fit Curve, m; Absolute Error, a) for $0\u201320\text{\hspace{0.17em}}\mathrm{m}$ Water Depth Using Different Default Bottom Spectra as Model Input^{
a
}

r

m

a

Sand

0.902

1.03

$1.64\text{\hspace{0.17em}}\mathrm{m}$

Coral

0.902

0.99

$1.48\text{\hspace{0.17em}}\mathrm{m}$

Algae

0.902

1.00

$1.50\text{\hspace{0.17em}}\mathrm{m}$

Flat

0.901

0.96

$1.40\text{\hspace{0.17em}}\mathrm{m}$

All model runs were performed using full geometry Tafkaa, $750\text{\hspace{0.17em}}\mathrm{nm}$ normalizing deglint, and full geometry for view and illumination angles.

Table 4

SHOALS Data Versus Estimated Bathymetry (Correlation Coefficient, r; Slope of Best Fit Curve, m; Absolute Error, a) for $0\u201320\text{\hspace{0.17em}}\mathrm{m}$ Water Depth Using Different Spatial Kernels for Averaging Water Properties^{
a
}

r

m

a

$1\times 1$

0.902

1.03

$1.64\text{\hspace{0.17em}}\mathrm{m}$

$3\times 3$

0.801

0.85

$1.89\text{\hspace{0.17em}}\mathrm{m}$

$5\times 5$

0.801

0.81

$1.88\text{\hspace{0.17em}}\mathrm{m}$

All model runs performed using full geometry Tafkaa, $750\text{\hspace{0.17em}}\mathrm{nm}$ normalizing deglint, full geometry for view and illumination angles, and an average sand spectrum as the default benthic input.