Conclusion - GERS Laboratory at UPRM
1
Abstract Methods Results Conclusion Literature Cited Acknowledgements This study presents descriptive and empirical results of optical parameters associated to Total Suspended Sediments (TSS) in a tropical open bay located at the west coast of Puerto Rico. Spatial analysis indicated that absolute values of TSS, remote sensing reflectance (R rs ) and total backscattering (b b ) increases with proximity to the shoreline. A good relationship between b bp and TSS was established (R 2 =0.7) in all wavelengths. The regression analyses between R rs and TSS indicate that best wavelengths to estimate TSS are between 589 to 645 nm. It was found that a red to green ratio improved the correlation results. R rs at 645 nm measured with the field spectoradiometer showed the highest correlation coefficient with MODIS reflectance. Sensors with better spatial and spectral resolution are needed in order to generate operational products of TSS in these highly variable optical waters. Fiigure 2. Magnitude and variability of TSS values increases with proximity to the shoreline (Fig. 2). Overall mean values from inshore, middle and offshore stations were 7.5, 3.6 and 2.7 mg/l, respectively. This spatial association is not only explained with river discharge, but also with the effect of re-suspension in shallow waters. Spatial variability of TSS R rs parameter (X) Regression relationship; n=72; TSS = R 2 R rs (442) 1357.9 * (X) - 1.512 0.17 R rs (470) 1331.6 * (X) - 3.4123 0.24 R rs (510) 1153.2 * (X) - 4.2096 0.45 R rs (589) 579.58 * (X) + 1.6079 0.73 R rs (620) 599.3 * (X) + 2.7821 0.73 R rs (645) 602.63 * (X) + 3.1481 0.73 R rs (665) 634.34 * (X) + 3.5357 0.71 R rs (675) 641.76 * (X) + 3.6363 0.71 R rs (859) 3675.4 * (X) + 5.9588 0.46 R rs (560)/R rs (620) 16.941* (X) -1.232 0.75 R rs (412)/R rs (670) 7.7701 * (X) -0.5628 0.77 R rs (665)/R rs (555) 23.943 * (X) - 0.7366 0.81 R rs (655)/R rs (545) 20.353 * (X) - 0.3937 0.84 Field Work Figure 1. (a) Study area showing the areas of river discharge and the location of monitored stations. (b) An optical package was used in all the stations. (c) Water samples were collected in each station and in river mouths. (d) Spectoradiometer GER- 1500- Gives measurements of R rs within the range of ~370-1000 nm. (a) (b) (d) (c) Dates Season No. of stations Fiel Data missing Good quality images available January 12-14, 2004 Dry 12 2 February 12, 2004 Dry 9 TSS Depth; R rs 1 August 19, 2004 Dry 10 TSS Depth 0 July 19, 2005 Rainy 6 TSS Depth 0 August 17, 2005 Rainy 5 TSS Depth 1 September 20, 2005 Rainy 6 0 October 19, 2005 Rainy 6 1 December 6, 2005 Dry 6 1 March 8, 2006 Dry 5 R rs 1 April 21, 2006 Dry 5 0 September 26, 2006 Rainy 6 0 October 26, 2006 Rainy 6 0 Cruise Details Total suspended sediments (mg/l) Remote sensing reflectance (sr^-1) Particle Backscattering (m^-1) 442, 470, 510, 589, 620, 675 nm Correlation Analyses Spatial variability of b bp and R rs R rs vs. b bp Vilmaliz Rodríguez-Guzmán and Fernando Gilbes-Santaella Geological and Environmental Remote Sensing Laboratory, Department of Geology, University of Puerto Rico at Mayagüez Empirical relationships between in situ optical measurements and TSS TSS Spectral Response This Project is Sponsored by NOAA-CREST Grant no. NA06OAR4810162 Special thanks to Patrick Reyes. Ramón López and Marrcos Rosado for helping in getting and processing part the data and giving helpful recommendations. Our appreciation to all the people that collaborate in the field and laboratory work. Results of this study provided a baseline for better understanding spatial and spectral variability of remote sensing parameters and their relationship with TSS. Regression analyses suggest that this quality parameter dominates the water-leaving signal between 589- 645 nm. The high spatial variability in optical parameters of the study site in combination with the relative low spatial resolution of MODIS demonstrated that a better ocean color sensor is required for coastal studies in tropical open bays. Results also show that use of band ratios will improve the development of algorithms and their application in these waters. Coastal Remote Sensing is one of NOAA-CREST trust research areas. In this study we derived empirical relationships between TSS, optical backscattering (b b ), in situ remote sensing reflectance (R rs ) and MODIS data in order to provide a baseline for the development of site specific algorithms to estimate TSS in tropical and highly variable coastal environments. General Objective MODIS Reflectance vs. in situ measurements Figure 5. Relationship between R rs and b b suggests that in long wavelengths (>510 nm) backscattering dominates the signal, while in shorter wavelengths (< 510 nm) absorption appears to dominate over backscattering (Fig. 4) The single band regression analysis supports the use of wavelengths at 589 nm or 620 to estimate TSS, because both showed the higher square correlation coefficients comparing wavelengths of HS-6 channels. Better relationships between red to green reflectance ratios suggest that suspended sediments dominate water optical properties, over the effect of phytoplankton and CDOM substances (Binding et al., 2003). This presents an advantage in the development of algorithms to estimate TSS from satellite derived data in Mayagüez Bay because the effect of other water constituents in the water-leaving signal is reduced. Figure 3. Measurements of b b in offshore stations presented typical behavior of clear waters where higher values were observed in the blue region continuously decreasing with longer wavelengths (Morel et al., 2007). High backscattering values at 589 nm and 620 nm observed in inshore stations correspond to target wavelengths to estimate TSS because they reach ideal conditions where there is a dominance of TSS in the backscattering signal and the chlorophyll fluorescence effect is reduced. Figure 4. Main variations associated to proximity to the shoreline are: (1) an increase in magnitude, which is attributed to higher scattering by water constituents, and (2) a shifting from 490 nm to 575 nm of the higher peak due to higher absorption by chlorophyll and CDOM combined with higher scattering of TSS in the red region (Froidefond et al., 2002). Figure 3 Figure 4 Figure 5 Figure 8 shows the variations in magnitude and spectral shape as affected by the concentration of TSS. The highest difference in R rs magnitude due to TSS was observed from 550 nm to 700 nm. Figure 7 Figure 6 Figure 9. The best regression with MODIS data was found with R rs , with a fairly good correlation coefficient of 0.69. However this analysis indicated that it is a challenge to define any relationship between MODIS reflectance and in situ measurements of optically active components in this area. This difficulty could be diminished by the incorporation of more observations in the dataset. Figure 2 Morel, A., Gentili, B., Claustre, H., Babin, M., Bricaud, A., Ras, J., & Tièche, F. (2007). Optical properties of the “clearest” natural waters. Limnoogy and Oceanograghy, 52 (1), 217-229. Binding, C.E., Bowers, D. G. & Mithelson-Jacob, E. G. (2003). An algorithm for the retrieval of suspended sediment concentrations in the Irish Sea from SeaWiFS ocean colour satellite imagery. International Journal of Remote Sensing, 24 (19), 3791–3806. Froidefond J.M, Gardelb L., Guiralb D., Parrab M., Ternonb J.F. (2002). Spectral remote sensing reflectances of coastal waters in French Guiana under the Amazon influence. Remote Sensing of Environment, 80, 225-232.
Transcript of Conclusion - GERS Laboratory at UPRM
Abstract
Methods
Results
Conclusion
Literature Cited
Acknowledgements
This study presents descriptive and empirical results of optical parameters associated to TotalSuspended Sediments (TSS) in a tropical open bay located at the west coast of Puerto Rico.Spatial analysis indicated that absolute values of TSS, remote sensing reflectance (Rrs) and totalbackscattering (bb) increases with proximity to the shoreline. A good relationship between bbpand TSS was established (R2=0.7) in all wavelengths. The regression analyses between Rrs andTSS indicate that best wavelengths to estimate TSS are between 589 to 645 nm. It was foundthat a red to green ratio improved the correlation results. Rrs at 645 nm measured with thefield spectoradiometer showed the highest correlation coefficient with MODIS reflectance.Sensors with better spatial and spectral resolution are needed in order to generate operationalproducts of TSS in these highly variable optical waters.
Fiigure 2. Magnitude and variability of TSS valuesincreases with proximity to the shoreline (Fig. 2).Overall mean values from inshore, middle andoffshore stations were 7.5, 3.6 and 2.7 mg/l,respectively. This spatial association is not onlyexplained with river discharge, but also with theeffect of re-suspension in shallow waters.
Spatial variability of TSS
Rrs parameter (X) Regression relationship; n=72; TSS = R2
Rrs(442) 1357.9 * (X) - 1.512 0.17
Rrs(470) 1331.6 * (X) - 3.4123 0.24
Rrs(510) 1153.2 * (X) - 4.2096 0.45
Rrs(589) 579.58 * (X) + 1.6079 0.73
Rrs(620) 599.3 * (X) + 2.7821 0.73
Rrs(645) 602.63 * (X) + 3.1481 0.73
Rrs(665) 634.34 * (X) + 3.5357 0.71
Rrs(675) 641.76 * (X) + 3.6363 0.71
Rrs(859) 3675.4 * (X) + 5.9588 0.46
Rrs(560)/Rrs(620) 16.941* (X) -1.232 0.75
Rrs(412)/Rrs(670) 7.7701 * (X) -0.5628 0.77
Rrs(665)/Rrs(555) 23.943 * (X) - 0.7366 0.81
Rrs(655)/Rrs(545) 20.353 * (X) - 0.3937 0.84
Field Work
Figure 1. (a) Study area showing the areas of riverdischarge and the location of monitored stations.(b) An optical package was used in all the stations.(c) Water samples were collected in each stationand in river mouths. (d) Spectoradiometer GER-1500- Gives measurements of Rrs within the rangeof ~370-1000 nm.
(a)
(b)
(d)
(c)
Dates Season No. of stations
Fiel Data missing
Good quality images available
January 12-14, 2004 Dry 12 2
February 12, 2004 Dry 9 TSS Depth; Rrs 1
August 19, 2004 Dry 10 TSS Depth 0
July 19, 2005 Rainy 6 TSS Depth 0
August 17, 2005 Rainy 5 TSS Depth 1
September 20, 2005 Rainy 6 0
October 19, 2005 Rainy 6 1
December 6, 2005 Dry 6 1
March 8, 2006 Dry 5 Rrs 1
April 21, 2006 Dry 5 0
September 26, 2006 Rainy 6 0
October 26, 2006 Rainy 6 0
Cruise Details
Total suspended sediments (mg/l)
Remote sensing reflectance (sr^-1)
Particle Backscattering (m^-1)
442, 470, 510, 589, 620, 675 nm
Correlation Analyses
Spatial variability of bbp and Rrs Rrs vs. bbp
Vilmaliz Rodríguez-Guzmán and Fernando Gilbes-SantaellaGeological and Environmental Remote Sensing Laboratory, Department of Geology, University of Puerto Rico at Mayagüez
Empirical relationships between in situ optical measurements and TSS
TSS Spectral Response
This Project is Sponsored by NOAA-CREST Grant no. NA06OAR4810162
Special thanks to Patrick Reyes. Ramón López and Marrcos Rosado for helping in getting andprocessing part the data and giving helpful recommendations. Our appreciation to all thepeople that collaborate in the field and laboratory work.
Results of this study provided a baseline for better understanding spatial and spectralvariability of remote sensing parameters and their relationship with TSS. Regressionanalyses suggest that this quality parameter dominates the water-leaving signal between589- 645 nm. The high spatial variability in optical parameters of the study site incombination with the relative low spatial resolution of MODIS demonstrated that a betterocean color sensor is required for coastal studies in tropical open bays. Results also showthat use of band ratios will improve the development of algorithms and their application inthese waters.
Coastal Remote Sensing is one of NOAA-CREST trust research areas. In this study we derived empirical relationships between TSS, optical backscattering (bb), in situ remote sensing reflectance (Rrs) and MODIS data in order to provide a
baseline for the development of site specific algorithms to estimate TSS in tropical and highly variable coastal environments.
General Objective
MODIS Reflectance vs. in situ measurements
Figure 5. Relationship between Rrs and bb suggests that in long wavelengths (>510 nm) backscattering dominates the signal, while in shorter wavelengths (< 510 nm) absorption appears to dominate over backscattering (Fig. 4)
The single band regression analysis supports the use of wavelengthsat 589 nm or 620 to estimate TSS, because both showed the highersquare correlation coefficients comparing wavelengths of HS-6channels. Better relationships between red to green reflectanceratios suggest that suspended sediments dominate water opticalproperties, over the effect of phytoplankton and CDOM substances(Binding et al., 2003). This presents an advantage in thedevelopment of algorithms to estimate TSS from satellite deriveddata in Mayagüez Bay because the effect of other water constituentsin the water-leaving signal is reduced.
Figure 3. Measurements of bb in offshore stations presented typical behavior of clear waterswhere higher values were observed in the blue region continuously decreasing with longerwavelengths (Morel et al., 2007). High backscattering values at 589 nm and 620 nm observedin inshore stations correspond to target wavelengths to estimate TSS because they reach idealconditions where there is a dominance of TSS in the backscattering signal and the chlorophyllfluorescence effect is reduced.
Figure 4. Main variations associated to proximity to the shoreline are: (1) an increase inmagnitude, which is attributed to higher scattering by water constituents, and (2) a shiftingfrom 490 nm to 575 nm of the higher peak due to higher absorption by chlorophyll andCDOM combined with higher scattering of TSS in the red region (Froidefond et al., 2002).
Figure 3 Figure 4 Figure 5
Figure 8 shows the variations inmagnitude and spectral shape asaffected by the concentration ofTSS. The highest difference in Rrsmagnitude due to TSS was observedfrom 550 nm to 700 nm.
Figure 7
Figure 6
Figure 9. The best regressionwith MODIS data was foundwith Rrs, with a fairly goodcorrelation coefficient of 0.69.However this analysisindicated that it is a challengeto define any relationshipbetween MODIS reflectanceand in situ measurements ofoptically active componentsin this area. This difficultycould be diminished by theincorporation of moreobservations in the dataset.
Figure 2
Morel, A., Gentili, B., Claustre, H., Babin, M., Bricaud, A., Ras, J., & Tièche, F. (2007). Optical properties ofthe “clearest” natural waters. Limnoogy and Oceanograghy, 52 (1), 217-229.
Binding, C.E., Bowers, D. G. & Mithelson-Jacob, E. G. (2003). An algorithm for the retrieval of suspendedsediment concentrations in the Irish Sea from SeaWiFS ocean colour satellite imagery. International Journalof Remote Sensing, 24 (19), 3791–3806.
Froidefond J.M, Gardelb L., Guiralb D., Parrab M., Ternonb J.F. (2002). Spectral remote sensing reflectances ofcoastal waters in French Guiana under the Amazon influence. Remote Sensing of Environment, 80, 225-232.
