# Assessing the Effect of Tandem Phase Sentinel-3 OLCI Sensor Uncertainty on the Estimation of Potential Ocean Chlorophyll-a Trends

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Materials and Methods

^{−1}[5,7,18,19,20]), were imposed on the time series. The range of trend values allow us to assess the interaction between the noise and trend magnitude/direction.

## 3. Results

^{−1}is shown in Figure 3. From this it can be seen that the peak of the distribution, i.e., the modal value, is ~−0.5%·yr

^{−1}, the same as the input trend for all the comparisons. Some variability can be seen around this peak; all distributions have a standard deviation of ~0.27%·yr

^{−1}. All distributions show some asymmetry, which can be attributed to the trends being estimated using log transformed chl-a data before being converted for plotting. It can be seen that for the three datasets OLCI-A-Aligned and OLCI-B are the most similar (expected from how OLCI-A-Aligned is created [14]) and that OLCI-A-Vicarious has a slightly, but negligibly, larger standard deviation of trend estimates (0.266%·yr

^{−1}for OLCI-A-Aligned, 0.272%·yr

^{−1}for OLCI-A-Vicarious, and 0.268%·yr

^{−1}for OLCI-B). These differences would not be expected to lead to any appreciable difference in trend detection when comparing the datasets. A realistic range of chl-a trends was tested, with very similar results in terms of consistency between datasets, although the standard deviation of trend estimates does increase with increasing values of input trend, and non-normal distributions indicate interactions with the underlying dataset (see Appendix A).

^{−1}). In Figure 4, for each grid cell the difference between (a) OLCI-A-Aligned and OLCI-A-Vicarious, (b) OLCI-A-Aligned and OLCI-B, and (c) OLCI-A-Vicarious and OLCI-B was calculated, with a positive value indicating that the latter had a greater standard deviation of chl-a trend estimates than the former. Appendix B shows similar figures at all input trend levels. The overall observation is the same as for the global histograms in that there is a relatively small difference between the datasets. OLCI-A-Vicarious shows more variance of chl-a trend estimates than OLCI-A-Aligned. OLCI-A-Aligned shows more similarity to OLCI-B than OLCI-A-Vicarious, which is expected from how the OLCI-A-Aligned dataset is produced. However, differences are typically small, with 90% of the differences in chl-a trend estimates between OLCI-A-Aligned and OLCI-A-Vicarious in the range ±0.11%·yr

^{−1}, ±0.027%·yr

^{−1}for OLCI-A-Aligned and OLCI-B, and ±0.11%·yr

^{−1}for OLCI-A-Vicarious and OLCI-B. By means of the radiometric alignment of OLCI-A to OLCI-B (OLCI-A-Aligned being understood as a “duplicate” of OLCI-B) it is completely consistent to see similar results between OLCI-A-Vicarious against either OLCI-A-Aligned or OLCI-B. Moreover, the differences are expected to be much smaller between OLCI-A-Aligned and OLCI-B.

## 4. Discussion

## 5. Conclusions

^{−1}for OLCI-B versus 0.266%·yr

^{−1}for OLCI-A-Aligned). OLCI-A-Vicarious has a slightly greater standard deviation, expected from how it is calculated (0.272%·yr

^{−1}), but this can still be considered a negligible difference from OLCI-B and OLCI-A-Aligned. There is some spatial variation in the differences, with open ocean regions being the most similar and high northern latitudes and coastal regions being the most different, albeit still with an almost negligible difference. This analysis shows that the effect of differences in sensor uncertainty will only have a negligible impact on future chl-a trend estimation using the Sentinel-3 platforms, where each satellite carries the same sensor design. This analysis does not, however, and cannot, estimate the differences from future drift in sensors, which may well have a significant impact and will need to be monitored. Likewise, differences between past, present, and future non-Sentinel sensors are not considered, even though these differences may have a significant impact on chl-a trend detection, which typically relies on records merged from multiple sensors.

## Author Contributions

## Funding

## Acknowledgments

## Conflicts of Interest

## Appendix A

**Figure A1.**Histograms of all chl-a trends estimated in all 1° grid cells and all bootstraps for OLCI-A-Aligned and OLCI-A-Vicarious. This shows that OLCI-A-Aligned and OLCI-A-Vicarious both show very similar distributions of chl-a trends, except that OLCI-A-Aligned has slightly smaller variance, which applies to all input trends.

**Figure A2.**Histograms of all chl-a trends estimated in all 1° grid cells and all bootstraps for OLCI-A-Aligned and OLCI-B. This shows that OLCI-A-Aligned and OLCI-B both show very similar distributions of chl-a trends, except that OLCI-A-Aligned has slightly smaller variance, which applies to all input trends.

**Figure A3.**Histogram of all chl-a trends estimated in all 1° grid cells and all bootstraps for OLCI-A-Vicarious and OLCI-B. This shows that OLCI-A-Vicarious and OLCI-B both show very similar distributions of chl-a trends, except that OLCI-A has slightly smaller variance, which applies to all input trends.

## Appendix B

**Figure A4.**Global 1° maps of the differences between the standard deviation of all chl-a trend estimates from all bootstraps for OLCI-A-Aligned and OLCI-A-Vicarious in each grid cell. The value expressed in percent per year above each subplot indicates the trend imposed on the synthetic time series. Red indicates that OLCI-A-Vicarious has higher standard deviation of chl-a trend estimates than OLCI-A-Aligned. Black grid cells indicate where missing data prevents analysis.

**Figure A5.**Global 1° maps of the differences between the standard deviation of all chl-a trend estimates from all bootstraps for OLCI-A-Aligned and OLCI-B in each grid cell. The value expressed in percent per year above each subplot indicates the trend imposed on the synthetic time series. Red indicates that OLCI-B has higher standard deviation of chl-a trend estimates than OLCI-A-Aligned. Black grid cells indicate where missing data prevents analysis.

**Figure A6.**Global 1° maps of the differences between the standard deviation of all chl-a trend estimates from all bootstraps for OLCI-A-Vicarious and OLCI-B in each grid cell. The value expressed in percent per year above each subplot indicates the trend imposed on the synthetic time series. Blue indicates that OLCI-A-Vicarious has higher standard deviation of chl-a trend estimates than OLCI-B. Black grid cells indicate where missing data prevents analysis.

## Appendix C

**Figure A7.**Global maps at 1° resolution of p-values from Levene’s test for significant difference in the variance of chl-a trend estimates from all bootstraps for an input trend of −0.5%·yr

^{−1}for (

**a**) OLCI-A-Aligned and OLCI-A-Vicarious, (

**b**) OLCI-A-Aligned and OLCI-B, and (

**c**) OLCI-A-Vicarious and OLCI-B in each grid cell. Values of less than 0.05 indicate evidence that the two datasets have a significant difference in their variance of chl-a trend estimates. Black grid cells indicate missing data.

## Appendix D

**Figure A8.**Global maps at 1° resolution of the OLCI-A-Aligned chl-a (

**a**) mean and (

**b**) standard deviation over the tandem period. Black grid cells indicate missing data.

**Figure A9.**Global maps at 1° resolution of the OLCI-A-Vicarious chl-a (

**a**) mean and (

**b**) standard deviation over the tandem period. Black grid cells indicate missing data.

## References

- Donlon, C.; Berruti, B.; Buongiorno, A.; Ferreira, M.; Féménias, P.; Frerick, J.; Goryl, P.; Klein, U.; Laur, H.; Mavrocordatos, C.; et al. The Global Monitoring for Environment and Security (GMES) Sentinel-3 mission. Remote Sens. Environ.
**2012**, 120, 37–57. [Google Scholar] [CrossRef] - GCOS. Systematic observation requirements for satellite-based data products for climate. 2011 Update. In Implementation Plan for the Global Observing System for Climate in Support of the UNFCCC; GCOS-154; World Meteorological Organization: Geneva, Switzerland, 2011. [Google Scholar]
- Clerc, S.; Donlon, C.; Borde, F.; Lamquin, N.; Hunt, S.; Smith, D.; McMillan, M.; Mittaz, J.; Wooliams, E.; Hammond, M.; et al. Benefits and lessons learned from the Sentinel-3 Tandem phase. Remote Sens. in preparation.
- McClain, C.R. A Decade of Satellite Ocean Color Observations. Ann. Rev. Mar. Sci.
**2009**, 1, 19–42. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Hammond, M.L.; Beaulieu, C.; Sahu, S.K.; Henson, S.A. Assessing trends and uncertainties in satellite-era ocean chlorophyll using space-time modeling. Global Biogeochem. Cycles
**2017**, 31, 1103–1117. [Google Scholar] [CrossRef] [Green Version] - Henson, S.A.; Beaulieu, C.; Lampitt, R. Observing climate change trends in ocean biogeochemistry: When and where. Glob. Chang. Biol.
**2016**, 22, 1561–1571. [Google Scholar] [CrossRef] [Green Version] - Beaulieu, C.; Henson, S.A.; Sarmiento, J.L.; Dunne, J.P.; Doney, S.C.; Rykaczewski, R.R.; Bopp, L. Factors challenging our ability to detect long-term trends in ocean chlorophyll. Biogeosciences
**2013**, 10, 2711–2724. [Google Scholar] [CrossRef] [Green Version] - Boyce, D.G.; Worm, B. Patterns and ecological implications of historical marine phytoplankton change. Mar. Ecol. Prog. Ser.
**2015**, 534, 251–272. [Google Scholar] [CrossRef] - Henson, S.A.; Sarmiento, J.L.; Dunne, J.P.; Bopp, L.; Lima, I.; Doney, S.C.; John, J.; Beaulieu, C. Detection of anthropogenic climate change in satellite records of ocean chlorophyll and productivity. Biogeosciences
**2010**, 7, 621–640. [Google Scholar] [CrossRef] [Green Version] - Garnesson, P.; Mangin, A.; D’Andon, O.F.; Demaria, J.; Bretagnon, M. The CMEMS GlobColour chlorophyll a product based on satellite observation: Multi-sensor merging and flagging strategies. Ocean Sci.
**2019**, 15, 819–830. [Google Scholar] [CrossRef] [Green Version] - Lavender, S.; Jackson, T.; Sathyendranath, S. The Ocean Colour Climate Change Initiative: Merging ocean colour observations seamlessly. Ocean Chall.
**2015**, 21, 29–31. [Google Scholar] - Maritorena, S.; Siegel, D.A. Consistent merging of satellite ocean color data sets using a bio-optical model. Remote Sens. Environ.
**2005**, 94, 429–440. [Google Scholar] [CrossRef] - Djavidnia, S. Comparison of global ocean colour data records. Ocean Sci.
**2010**, 61–76. [Google Scholar] [CrossRef] [Green Version] - Lamquin, N.; Clerc, S.; Bourg, L.; Donlon, C. OLCI A/B tandem phase analysis, part 1: Level 1 homogenisation and harmonisation. Remote Sens.
**2020**, 12, 1804. [Google Scholar] [CrossRef] - Lamquin, N.; Déru, A.; Clerc, S.; Bourg, L.; Donlon, C. OLCI A/B tandem phase analysis, part 2: Benefits of sensors harmonisation for Level 2 products. Remote Sens. in preparation.
- Antoine, D.; Fanton d’Andon, O. Sentinel-3 optical products and algorithm definition. In OLCI Level 2 Algorithm Theoretical Basis Document: Ocean Color Products in Case 1 Waters; European Space Agency: Paris, France, 2010. [Google Scholar]
- Antoine, D.; Morel, A. A multiple scattering algorithm for atmospheric correction of remotely-sensed ocean colour (MERIS instrument): Principle and implementation for atmospheres carrying various aerosols including absorbing ones. Int. J. Remote Sens.
**1999**, 20, 1875–1916. [Google Scholar] [CrossRef] - Gregg, W.W.; Rousseaux, C.S. Decadal trends in global pelagic ocean chlorophyll: A new assessment integrating multiple satellites, in situ data, and models. J. Geophys. Res. Ocean.
**2014**, 119, 5921–5933. [Google Scholar] [CrossRef] [Green Version] - Gregg, W.W.; Rousseaux, C.S.; Franz, B.A. Global trends in ocean phytoplankton: A new assessment using revised ocean colour data. Remote Sens. Lett.
**2017**, 8, 1102–1111. [Google Scholar] [CrossRef] - Vantrepotte, V.; Mélin, F. Inter-annual variations in the SeaWiFS global chlorophyll a concentration (1997–2007). Deep. Res. Part I Oceanogr. Res. Pap.
**2011**, 58, 429–441. [Google Scholar] [CrossRef] - Levene, H. Robust Tests for Equality of Variances. In Contributions to Probability and Statistics: Essays in Honor of Harold Hotelling; Stanford University Press: Palo Alto, CA, USA, 1960; pp. 278–292. [Google Scholar]
- Bélanger, S.; Babin, M.; Tremblay, J.-É. Increasing cloudiness in Arctic damps the increase in phytoplankton primary production due to sea ice receding. Biogeosciences
**2013**, 10, 4087–4101. [Google Scholar] [CrossRef] [Green Version] - Mélin, F. Impact of inter-mission differences and drifts on chlorophyll-a trend estimates. Int. J. Remote Sens.
**2016**, 37, 2233–2251. [Google Scholar] [CrossRef] [Green Version] - Lamquin, N.; Clerc, S.; Bourg, L.; Donlon, C. OLCI A/B tandem phase analysis, part 3: Post-tandem monitoring of cross-calibration from statistics of Deep Convective Clouds observations. Remote Sens. in preparation.

**Figure 1.**Comparison between Sentinel-3A Ocean and Land Colour Imager (OLCI-A) system vicarious calibration (SVC) gains and the mean inter-calibration coefficients used to align OLCI-A to Sentinel-3B OLCI (OLCI-B) per camera. Adapted from [15].

**Figure 2.**(

**a**) Diagrammatic example of a monthly chlorophyll-a (chl-a) time series of 10 years with seasonal and interannual variability but no trend. (

**b**) adds random observational noise to this series which induces an apparent small non-zero positive trend, a negative trend would be just as likely with another sampling of random noise.

**Figure 3.**Histograms of −0.5%·yr

^{−1}input chl-a trends estimated in all 1° grid cells and all bootstraps. From (

**a**) OLCI-A-Aligned and OLCI-A-Vicarious, (

**b**) OLCI-A-Aligned and OLCI-B, and (

**c**) OLCI-A-Vicarious and OLCI-B. All datasets show very similar distributions of chl-a trends. See Appendix A for corresponding histograms for all input trends.

**Figure 4.**Global maps at 1° resolution of the differences between the standard deviation of all chl-a trend estimates from all bootstraps for an input trend of −0.5%·yr

^{−1}for (

**a**) OLCI-A-Aligned and OLCI-A-Vicarious, (

**b**) OLCI-A-Aligned and OLCI-B, and (

**c**) OLCI-A-Vicarious and OLCI-B in each grid cell. Red indicates that the former dataset in each pair has higher standard deviation of chl-a trend estimates than the latter dataset in each pair. See Appendix B for similar maps for all input trends. If statistical testing determined that the difference in the variance of chl-a trend estimates between the two datasets was not significant (Appendix C) the cell is considered as having a difference of zero (i.e., white). Black grid cells indicate where missing data prevents analysis.

**Figure 5.**Global maps at 1° resolution of the OLCI-B chl-a (

**a**) mean and (

**b**) standard deviation over the tandem period. See Appendix D for similar maps produced for OLCI-A-Aligned and OLCI-A-Vicarious. Black grid cells indicate missing data.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Hammond, M.L.; Henson, S.A.; Lamquin, N.; Clerc, S.; Donlon, C.
Assessing the Effect of Tandem Phase Sentinel-3 OLCI Sensor Uncertainty on the Estimation of Potential Ocean Chlorophyll-*a* Trends. *Remote Sens.* **2020**, *12*, 2522.
https://doi.org/10.3390/rs12162522

**AMA Style**

Hammond ML, Henson SA, Lamquin N, Clerc S, Donlon C.
Assessing the Effect of Tandem Phase Sentinel-3 OLCI Sensor Uncertainty on the Estimation of Potential Ocean Chlorophyll-*a* Trends. *Remote Sensing*. 2020; 12(16):2522.
https://doi.org/10.3390/rs12162522

**Chicago/Turabian Style**

Hammond, Matthew L., Stephanie A. Henson, Nicolas Lamquin, Sébastien Clerc, and Craig Donlon.
2020. "Assessing the Effect of Tandem Phase Sentinel-3 OLCI Sensor Uncertainty on the Estimation of Potential Ocean Chlorophyll-*a* Trends" *Remote Sensing* 12, no. 16: 2522.
https://doi.org/10.3390/rs12162522