Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Citizens and scientists collect comparable oceanographic data: measurements of ocean transparency from the Secchi Disk study and science programmes


Marine phytoplankton accounts for approximately 50% of all photosynthesis on Earth, underpins the marine food chain and plays a central role in the Earth’s biogeochemical cycles and climate. In situ measurements of ocean transparency can be used to estimate phytoplankton biomass. The scale and challenging conditions of the ocean make it a difficult environment for in situ studies, however. Here, we show that citizen scientists (seafarers) using a simple white Secchi Disk can collect ocean transparency data to complement formal scientific efforts using similar equipment. Citizen scientist data can therefore help understand current climate-driven changes in phytoplankton biomass at a global scale.


The ocean is a difficult environment to access for in situ study due to its scale, remoteness and challenging conditions. Although ocean science research is vital for our sustainable future1, scientific research lags current, climate-driven ocean changes2. The ocean’s phytoplankton account for approximately 50% of all photosynthesis on Earth3,4 and temporal and spatial changes in the phytoplankton can influence marine productivity5, weather6 and climate7,8. Monitoring the phytoplankton is therefore essential as an early indicator of regional and global ecosystem change9,10,11,12.

Around 44% of the human population lives within 150 km of the coast13 and a number go to sea for work and recreation. The seafaring public often visits the same locations, whether as sailors on short day trips, commercial fishermen accessing fishing grounds, or offshore yachtsmen/women whose passages follow common routes dictated by the season, prevailing winds and currents14. Therefore, the seafaring public provides an opportunity to collect oceanographic data over varying spatial and temporal scales to contribute to scientific efforts and many now participate in marine citizen science15,16.

The global Secchi Disk study ( engages seafarers to use a Secchi Disk18 to collect in situ data on ocean transparency that can be used to estimate phytoplankton biomass19. The Secchi Disk study uses a simple, 30 cm diameter white Secchi Disk that is weighted and attached to a tape-measure, and lowered vertically into the water from a boat’s side. The depth (m) below the surface when the Secchi Disk disappears from sight is the Secchi depth (ZSD), which measures ocean transparency. When the bathymetry is > 25 m depth and the distance > 1 km from shore, the primary influence upon ocean transparency is phytoplankton pigments and their breakdown products and therefore, ZSD estimates phytoplankton biomass in the water column; re-suspended sediments and dissolved organic matter from rivers further reduce transparency and introduce optical errors in shallower water and closer inshore12.

Marine scientists have used Secchi Disks to measure ocean transparency since 186520 and archives of ZSD represent one of the longest-running, spatially extensive global marine datasets21. Recently, the Secchi Disk has fallen from widespread use among marine scientists22 due to spectrophotometric determination of chlorophyll a12 and satellite measures of ocean colour23. To re-establish the database of ZSD measurements and facilitate contemporary ZSD comparisons to historical data, the citizen science Secchi Disk study began in 201317.

Measuring in situ data at sea from a small boat is challenging. For citizen science efforts of any type to be useful it is essential to ensure and assess data quality24,25, and this was outlined recently in the ninth of the ten principles of citizen science outlined by the European Citizen Science Association (ECSA)26. Marine and terrestrial, observational, citizen science studies have assessed data quality and shown that citizens can collect reliable observational data when given good instructions27,28,29. We have already shown that ZSD data collected by citizens agrees with satellite measures of chlorophyll a and satellite ZSD estimates17. But, does ZSD data collected by citizens of varying backgrounds using oceanographic equipment compare to ZSD data collected by trained scientists using the same methods? Unfortunately, it is impractical to take citizens and marine scientists to sea together to measure ZSD at the same time and place to see how they compare. To overcome this impracticality, we compare ZSD measurements collected by citizens and scientists using Secchi Disks to independent satellite estimates of ZSD, i.e. the comparison Secchi Disk measurements were not used to generate the satellite algorithm.


We found positive relationships between satellite ZSD and both the scientist and citizen ZSD (Fig. 1a–l). Removing 20 samples did not strongly alter the regression lines (Fig. 1b,f,j). The red and blue polygons delineating the scientist and citizen data, respectively, as well as the regression lines, overlapped. Although a Kruskal–Wallis test revealed the regression lines were significantly different between scientist and citizen programmes (p < 0.01), these differences in slope were small and so there was only a small slope alteration when the data were merged. Correlation analysis revealed that scientist correlations were slightly higher when ocean + shelf and shelf only were considered (Fig. 1d,h,l). The difference in correlation was higher when open-ocean data only were analysed (Fig. 1d,h,l). In general, scientific data had less variance in correlation except for the shelf where a higher variance may be explained by outliers (red crosses below the regression line within the red, dashed polygons) (Fig. 1b,f).

Figure 1
figure 1

Relationships between scientist (red) and citizen (blue) log10(ZSD) and satellite log10(ZSD). (a,e,i) Maps showing the positions of the scientist and citizen ZSD data for shelf + open-ocean, shelf, and open-ocean, respectively. Maps were created using MATLAB version 2021a, (b,f,j) Scatterplots, polygons and regression lines showing the relationship between scientist, citizen, and satellite log10(ZSD) for shelf + open-ocean, shelf, and open-ocean data, respectively. Red and blue polygons delineate the scientist and citizen data, respectively, and regression lines in black represent combined scientist and citizen data. (c,g,k) Histograms of regression coefficients between scientist, citizen and scientist/citizen (black) and satellite log10(ZSD) data for shelf + open-ocean, shelf, and open-ocean, respectively. (d,h,l) Histograms of linear correlations between scientist, citizen, combined scientist and citizen (black), and satellite log10(ZSD) for shelf + open-ocean, shelf, and open-ocean data, respectively.

Next, we analysed the data from WOOD, WOD, ICES and the citizens, separately (Fig. 2a–l). This analysis revealed that outliers could also be detected among scientific programmes (red crosses on the left below the regression line inside the red polygon, Fig. 2b,f). As before, all the regression lines were significantly different (Kruskal–Wallis test, p < 0.01). We noticed large variability in the regression lines, especially for WOOD when open-ocean data was considered; due to a low number of degrees of freedom. The histogram of the category of slopes showed a more significant separation and less overlap for all programmes in shelf seas than the open-ocean, which could reflect regional peculiarities (Fig. 2c,g,k). The histograms for each programme and for open-ocean + shelf, shelf and open-ocean (Fig. 3a–c) showed that part of the difference in regression coefficients may be due to location rather than the reliability of either citizen or scientist ZSD. We also found a difference in the correlations between open-ocean and shelf. With the exception of the open-ocean where citizen correlations were slightly lower than WOD, correlations involving citizen data were higher than other programmes (Fig. 2d,h,l).

Figure 2
figure 2

Relationships between WOD (magenta), WOOD (green), ICES (red) and citizen (blue) log10(ZSD) and satellite log10(ZSD). (a,e,i). Maps showing the positions of the WOD, WOOD, ICES, and citizen ZSD data for shelf + open-ocean, shelf, and open-ocean, respectively. Maps were created using MATLAB version 2021a, (b,f,j) Scatterplots, with red and blue polygons delineating the scientist and citizen data respectively, and regression lines showing the relationship between WOD, WOOD, ICES, citizen and satellite log10(ZSD) for shelf + open-ocean, shelf, and open-ocean, respectively. (c,g,k) Histograms of regression coefficients between WOD, WOOD, ICES, citizen and satellite log10(ZSD) for shelf + open-ocean, shelf, and open-ocean, respectively. (d,h,l) Histograms of the linear correlations between WOD, WOOD, ICES, citizen and satellite log10(ZSD) for shelf + open-ocean, shelf, and open-ocean, respectively.

Figure 3
figure 3

Histograms of ZSD for the three scientific programmes and for citizens. (a–c) Histograms for WOD (magenta), WOOD (green), ICES (red) and citizen (blue) ZSD data for shelf + open-ocean, shelf, and open-ocean, respectively.

When we compared the citizen and scientist data for systematic under- or overestimation using satellite ZSD as a reference, scientists always highly overestimate satellite ZSD as shown by the percentage of overestimates that are consistently well above 50% (Table 1, Fig. 4). We found that citizens tend to slightly underestimate satellite ZSD when all data are considered (44.5%) and to a greater extent when ZSD is greater than or equal to 25 m (36.4%). Both citizens and scientists overestimate ZSD compared to satellite ZSD when the measurement is less than 25 m. In all cases however, the under- or over estimation is less for citizens than for scientists. In contrast to the percentage of overestimates, the mean deviation (Table 1) is a more quantitative measurement of the degree of under- or overestimation and can be influenced by individual deviation (the individual differences between in situ and satellite ZSD measurements). The mean deviation shows nearly the same results as the percentage of overestimates although when we consider all ZSD together the underestimation by citizens (− 1.6 m) is slightly higher than the overestimation made by scientists (1.5 m) (Table 1). (The discrepancy observed between the percentage of overestimates and the mean deviation is likely due to a few large, individual deviations that influence the latter estimator). When we separated measurements of ZSD below and above 25 m the mean deviation is smaller in absolute value for citizens in agreement with what is shown by the percentage of overestimates.

Table 1 Indirect comparison of citizen and scientist ZSD data with respect to under and overestimation and closeness.
Figure 4
figure 4

Relationships between citizen and scientist in situ ZSD, and satellite ZSD. (a–c) Scatterplots showing the relationship between scientist (red crosses), citizen (blue circles), and satellite ZSD for all the whole dataset, where ZSD < 25 m and where ZSD ≥ 25 m, respectively. Black dashes represent the line y = x. (d–f) Histograms of the mean square deviation (m2) for the whole dataset, where ZSD < 25 m and where ZSD ≥ 25 m, respectively. Orange, blue and magenta bars represent scientist, citizen and where scientist and citizen data overlap, respectively.

The mean square deviation (here, a measure of the level of the closeness between in situ and satellite ZSD) revealed that ZSD measured by scientists are consistently closer than ZSD measured by citizens. Histograms prepared to examine the mean square deviation (Fig. 4d–f) suggest that a few observations made by citizens might cause the higher mean square deviation (Table 1). When we remove categories above 100 m2 the mean square deviation for citizens reduced by more than 50%, from 39.2 to 18.9 m2, and became very close to the scientist value (16.6 m2). Nevertheless, the degree of closeness between satellite and in situ ZSD is higher for scientists than for citizens. When the first 3 categories of the histogram are considered together (Fig. 4d) they comprise 87% of the scientific data and 70% of the citizen data, which is a difference of 17%.


The Secchi Disk is among the simplest pieces of oceanographic equipment with its method of use, optical properties, limitations and variability well understood18,30; these features are taken into account in the instructions provided to citizens in the Smartphone application called Secchi that accompanies the citizen science study17. While our method of assessing citizen and scientist ZSD data by comparison to satellite ZSD is indirect and so imperfect (ideally citizens and scientists would go to sea together to make simultaneous measurements for such a comparison), our results show that scientists and citizens take reliable in situ measurements of ZSD at sea.

Although we found that citizens tend to underestimate satellite ZSD, scientists tend to overestimate to a greater extent (Table 1) when compared to satellite ZSD. We do not know the cause of this under- and overestimation and cannot exclude this is a chance result, and because citizens and scientists under- and overestimate respectively, it is not possible at present to determine a correction factor; we nevertheless hope this can be resolved when more data is evaluated from both scientist and citizen programmes. We found higher variability in the mean square deviation for citizens and a 17% difference between citizen ZSD and scientist ZSD (when scientist data was combined from the three scientist datasets we used) (Table 1). When we performed the analysis on log10 transformed data the differences between citizens and scientists were much less apparent and to see them we need to look at the third digit of the mantissa (Supplementary Table S1).

The difference in the mean square deviation between citizens and scientists could be explained by several factors. For example, the three science programmes we used likely involve measurements collected by just a few individual scientists (for example on a single cruise), which will reduce the amount of individual observer variability that is inherent in the in situ measurement of ZSD30 when compared to the citizen data where many individual participants each collect a few measurements. Interestingly, when we excluded categories of mean square deviation above 100 m2 the closeness between citizens and scientists improved greatly. Consequently, as the citizen data set increases in size, we hope to be able to conduct analyses at a participant level to better understand individual variability and to remove outliers31. (The comparison between scientists and citizens may also change when data from many scientific programmes are considered and over longer time periods and larger spatial scales). Other influences on the data may include the fact that scientific measurements may be focused on a particular water body and so spatial heterogeneity will be different compared to the citizen data (Fig. 1a,e,i) and the fact that research vessels are likely to be more stable platforms compared to small boats and so measurements are likely to be easier to take. Finally, and importantly, scientists interact and train each other, which is different to a global citizen science study where similar training is impractical. Consequently, even though instructions are given to citizens in the Secchi smartphone application upon how to take ZSD measurements, scientists are likely to show less variability because of their more formal training.

When we considered spatial heterogeneity, the ZSD collected by citizens and scientists showed a better relationship to satellite ZSD in the open-ocean (Fig. 2j,k). The poorer relationship found in shelf seas between in situ and satellite ZSD (Figs. 1a,e,I, 2a,e,i) may therefore be due to the inherent properties of measurements made in specific oceanographic environments (Fig. 3) rather than the reliability of the Secchi Disk measurement in the hands of either citizens or scientists. Most likely, it is due to the greater spatial and temporal heterogeneity of shelf seas (where there may also be abiotic factors), which may all lead to an imperfect match between in situ and satellite ZSD data due to violations in the assumptions underlying the satellite algorithm. The small overlap between volunteers (citizens) and professionals (scientists) (Fig. 1) when all the scientist data are combined together, compared to the clear overlap when we split the different science data sets (Fig. 2), may consequently, be explained by data originating from different ocean basins and environments, which is likely to alter in future when the data is more widespread.

Satellite estimates of ZSD are influenced by the in situ ZSD datasets used to generate the satellite algorithm and so will have inherent errors and biases. The Morel et al.32 algorithm we used is based upon a relationship between chlorophyll a and ZSD for clear, open-ocean waters and a specific, scientist-collected, in situ ZSD dataset where the majority of measurements were collected in the summertime in the northern hemisphere. Validation using MODIS Level-3 composite data and in situ ZSD resulted in adjusted coefficients as the modelled values tended to exceed the measured values32, and it is these adjusted coefficients that have been used in the implementation, in other words, the algorithm was tuned to match the specific in situ ZSD measurements.

In the open-ocean, satellite ZSD can overestimate in situ ZSD because the former assumes the layer extending down to ZSD is homogenous. In contrast, chlorophyll a always varies with depth32; chlorophyll levels are typically lower at the surface than at depth under stratified open-ocean conditions and so satellite chlorophyll underestimates the depth-averaged value. Although a recent algorithm to calculate ZSD from current satellite sensors has improved the relationship between satellite and in situ ZSD data going forwards33, satellite remote sensing still only provides an integrated value that simplifies the biological and environmental vertical and horizontal heterogeneity34,35; as our results in shelf seas may indicate in the latter case in particular. Consequently, even in the modern era of remote sensing of ocean colour, in situ measurements of ZSD provide an essential insight into ocean transparency and phytoplankton biomass.

Despite the growth in marine science36, due to the constraints of achieving widespread coverage by the limited scope of professional oceanographic cruises, the participation of the seafaring public provides an opportunity to extend oceanographic data collection. Citizen science has proven to contribute to gaps in our scientific knowledge in many fields15,37. By collecting ocean transparency data to contribute to historical and current in situ scientific ZSD efforts, citizen scientists can help better understand the effects of current global-scale, climate-driven changes in the phytoplankton.

In order to assess long-term, global, ocean phytoplankton change in situ from the beginning of the 20th Century, without combining different types of data, it is necessary to increase the coverage of present-day Secchi Disk ocean transparency measurements to enable a direct comparison with historical Secchi Disk data12,18. Seafarers acting as citizen scientists provide a unique opportunity to collect in situ ocean transparency data. This study addressed the ninth principle of citizen science outlined by the ECSA, which recommends that citizen science should be evaluated for scientific data quality26. We found that when citizen and scientist ZSD are compared to satellite ZSD calculated using the algorithm of Morel et al.32 citizen ZSD slightly underestimates the satellite value and scientist ZSD deeply overestimates the value (Table 1). The percentage of estimates and the mean deviation suggest that citizens perform as well as scientists. Although citizen data is more variable, it remains reliable and consequently, with this understanding, it is possible to combine citizen-collected ocean transparency data together with scientific studies (Figs. 1d,h,l, 2d,h,l). We would recommend the type of analysis we have conducted when ZSD data from different studies are combined so that long-term changes in phytoplankton biomass are estimated correctly.


Scientist measurements of ZSD

We obtained scientist ZSD from 3 datasets. (1) The World Ocean Database (WOD) ( (2) The World-wide Ocean Optics Database (WOOD) ( (3) The International Council for the Exploration of the Sea (ICES) ( We used ZSD data collected between 1997 to 2005, 1997 to 2001, and 1997 to 1998 from WOD, WOOD and ICES datasets, respectively.

Citizen measurements of ZSD

We used measurements of ZSD collected from 2013 to 2019 by seafarers participating in the on-going Secchi Disk study17.

Satellite estimates of ZSD

Satellite estimates of ZSD were the only possibility to compare citizen and scientist measurements of ZSD since we do not have in situ data collected by citizens and scientists at the same locations contemporaneously. We obtained satellite measurements from the Hermes GlobColour website ( The dataset, obtained by merging data from four sensors (SeaWiFS, MODIS, MERIS, VIIRS) with an arithmetic mean on a 4 km resolution spatial grid, provided overlapping time series of ZSD for the time period 1997–2019. Estimates of ZSD were calculated using the algorithm of Morel et al.32.

Bathymetry and distance from the nearest coast

We only used measurements of ZSD collected in water > 25 m deep and > 1 km from the coast. Bathymetry data were obtained from a global ocean bathymetry chart (0.1 degrees latitude by 0.1 degrees longitude)40. We calculated distance from the nearest coast from a global data set of distances from the nearest coastline estimated at a spatial resolution of 0.01 degrees41. We used the nearest method of interpolation42,43 to give a bathymetry or distance from the nearest coast for each ZSD.

Data pre-processing

We attributed a value of satellite-based ZSD for each in situ measurement of ZSD if a satellite estimate was available within 100 km of the in situ sample, plus three days before or after it’s measurement. To retain as many measurements as possible near to the coast we calculated the estimation using the nearest method of interpolation43,44.

Data analyses

We conducted three analyses: (i) indirectly comparing all scientist ZSD data together and citizen ZSD data to satellite ZSD (ii) comparing the three scientific programmes’ data separately and citizen data, to satellite ZSD and (iii)  determining whether any biases occurred between citizen and scientist ZSD data when compared to satellite data.

First, we examined the relationship between satellite ZSD and both scientific (all programmes combined) and citizen ZSD data using scatter plots for open-ocean and shelf systems; shelf where bathymetry ≤ 200 m and distance to coast > 1 km, and open-ocean where bathymetry > 200 m. We highlighted the datasets by drawing polygons around only-scientist, only-citizen and both scientist/citizen data. We performed three regression analyses37 (i) all data (scientist/citizen), (ii) scientist and, (iii) citizen. Each regression analysis was performed 1000 times, each time removing 20 samples to examine potential outliers’ influence on the regression. Increasing the removal of samples did not alter our conclusions; the variance increased when the removal of samples was higher and conversely. We therefore created 1000 regression lines for each of the three regression analyses. We displayed the histograms of all regression coefficients for scientist, citizen and scientist/citizen data. A Kruskal–Wallis test was applied to test whether regression coefficients for each dataset were significant.

We also calculated the linear correlations between both scientist and citizen ZSD and satellite ZSD. We performed 1000 simulations for this analysis, recalculating the correlation after randomly choosing 100 pairs of points when all areas (shelf + open-ocean) were selected, 100 for just shelf data and 90 for open-ocean data. The number of pairs of points available drove the choice of these thresholds. We used histograms to display the correlation coefficients.

Second, we performed similar tests to investigate the three scientific programmes separately (WOD, WOOD and ICES) and citizen data. However, while we used 1000 simulations to calculate linear correlations, we considered 65 pairs of points for open-ocean + shelf data and shelf data, and 29 pairs of points for open-ocean. The ICES and Pacific Ocean data were not used when we tested the relationships between scientist ZSD data and satellite ZSD for the open-ocean and shelf, respectively. We also produced histograms of ZSD for WOD, WOOD, ICES and citizen data for open-ocean + shelf, shelf, and open-ocean to determine if any changes in slopes (regression coefficients) might be explained by the sampling region.

Third, we determined if any biases occurred between citizen and science data when each was compared to satellite measurements independently. Specifically, we investigated whether citizen or scientist in situ measurements of ZSD underestimated or overestimated satellite measurements of ZSD that were derived from the algorithm of Morel et al.32,ZSD (i.e. without log10 transformation). Although a logarithmic transformation improves the regression analysis by stabilising the variance of the variable (homoscedasticity) (Fig. 1b), it was more straightforward to compare real values; we present the results for a logarithmic transformation in Supplementary Table S1. We assessed three estimators that are generally used to examine the performance of diversity indices45: (i) the percentage of overestimates, (ii) the mean deviation MD and (iii) the mean square deviation (MSD).

The percentage of overestimates is a measure of bias varying between 0 and 100%. An unbiased estimator should have a value of 50%, meaning that it underestimates or overestimates 50% of the time. In our context, we use this estimator to determine how frequently in situ citizen or scientist measurements were above or below satellite estimates. For example, a value of 20% for citizen ZSD would suggest that citizens deeply underestimate ZSD. The MD (expressed here in m) is the average of the differences between an estimator and its true or expected value. In the context of our study, it is the mean difference between in situ and satellite measurements and it is calculated as follows:


where O is the in situ ZSD measurement (either citizen or scientist), S is the satellite measurement of ZSD and n is the number of couples of points (for example, citizen and satellite estimates of ZSD). MD can therefore be negative if in situ values are lower than satellite estimates and inversely.

In contrast to the two previous estimators, MSD is used generally to estimate accur45. It is calculated by estimating the average of the square of the differences between in situ and satellite measurements as follows:


MSD (expressed here in m2) varies between 0 and \(\infty\) and is a measure of closeness between in situ and satellite data. A value close to 0 indicates that the accuracy of the estimator is high and inversely. In our case, it suggests that the values originating from in situ measurement (either citizen or scientist) are constantly close to the values assessed from satellites using the algorithm of Morel et al.32.

We also calculated histograms of the deviations (a deviation is the difference between O and S) between in situ and satellite measurements to compare in greater detail the degree of accuracy between citizens and scientists.

The estimators and histograms, were calculated (i) on the whole set of untransformed data (see Fig. 4a) and data (ii) where ZSD < 25 m and (iii) where ZSD ≥ 25 m. They were also calculated between in situ citizen and scientist data. Therefore, we had 6 values of percentages of overestimates, MD and MSD as well as six histograms (displayed as two per figure) at the end of the procedure (see Table 1, Fig. 4).


  1. 1.

    Visbeck, M. Ocean science research is key for a sustainable future. Nat. Commun. 9, 690. (2018).

    ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

  2. 2.

    Pendleton, L., Evans, K. & Visbeck, M. We need a global movement to transform ocean science for a better world. Proc. Natl. Acad. Sci. U.S.A. 117, 9652–9655. (2020).

    Article  PubMed  PubMed Central  Google Scholar 

  3. 3.

    Longhurst, A. R., Sathyendranath, S., Platt, T. & Caverhill, C. An estimate of global primary production in the ocean from satellite radiometer data. J. Plankton. Res. 17, 1245–1271 (1995).

    Article  Google Scholar 

  4. 4.

    Field, C. B., Behrenfeld, M. J., Randerson, J. T. & Falkowski, P. Primary production of the biosphere: Integrating terrestrial and oceanic components. Science 28, 237–240 (1998).

    ADS  Article  Google Scholar 

  5. 5.

    Stock, C. A. et al. Reconciling fisheries catch and ocean productivity. Proc. Natl. Acad. Sci. 114, E1441–E1449. (2017).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  6. 6.

    Gnanadesikan, A., Emanuel, K., Vecchi, G. A., Anderson, W. G. & Hallberg, R. How ocean color steers Pacific tropical cyclones. Geophys. Res. Lett. 37, L18802. (2010).

    ADS  Article  Google Scholar 

  7. 7.

    Holligan, P. M. Do marine phytoplankton influence global climate? In Primary Productivity and Biogeochemical Cycles in the Sea (eds Falkowski, P. G. & Woodhead, A. D.) 487–501 (Plenum Press, 1992).

    Chapter  Google Scholar 

  8. 8.

    Berner, R. A. The Phanerozoic Carbon Cycle 158 (Oxford University Press, 2004).

    Book  Google Scholar 

  9. 9.

    Roxy, M. et al. A reduction in marine primary productivity driven by rapid warming over the tropical Indian Ocean. Geophys. Res. Lett. 43, 826–833. (2016).

    ADS  Article  Google Scholar 

  10. 10.

    Dutkiewicz, S. et al. Ocean colour signature of climate change. Nat. Commun. 10, 578. (2019).

    ADS  CAS  Article  PubMed  PubMed Central  Google Scholar 

  11. 11.

    Van de Waal, B. & Litchman, E. Multiple global change stressor effects on phytoplankton nutrient acquisition in a future ocean. Philos. Trans. R. Soc. B 375, 20190706. (2020).

    CAS  Article  Google Scholar 

  12. 12.

    Boyce, D. G., Lewis, M. R. & Worm, B. Global phytoplankton decline over the past century. Nature 446, 591–596. (2010).

    ADS  CAS  Article  Google Scholar 

  13. 13.

    UN Atlas of the Oceans. (2002).

  14. 14.

    Cornell, J. Cornell’s Ocean Atlas: Pilot Charts for All Oceans of the World (Cornell Sailing. Ltd., 2011).

    Google Scholar 

  15. 15.

    Thiel, M. et al. Citizen scientists and marine research: Volunteer participants, their contributions and projection for the future. Oceanogr. Mar. Biol. Ann. Rev. 52, 257–314. (2014).

    Article  Google Scholar 

  16. 16.

    Kelly, R., Fleming, A., Peci, G. T., von Gönner, J. & Bonn, A. Citizen science and marine conservation: A global review. Philos. Trans. R. Soc. B 375, 20190461. (2020).

    Article  Google Scholar 

  17. 17.

    Seafarers, S. D. et al. Seafarer citizen scientist ocean transparency data as a resource for phytoplankton and climate research. PLoS ONE 12, e0186092. (2017).

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  18. 18.

    Wernand, M. R. On the history of the Secchi disc. J. Eur. Opt. Soc. Rapid Publ. 5, 10013. (2010).

    Article  Google Scholar 

  19. 19.

    Fleming-Lehtinen, V. & Laamanen, M. Long-term changes in Secchi depth and the role of phytoplankton in explaining light attenuation in the Baltic Sea. Estuar. Coast. Shelf Sci. 102, 1–10 (2012).

    ADS  Article  Google Scholar 

  20. 20.

    Secchi, P. A. Relazione delle esperienze fatte a bordo della pontificia pirocorvetta Imacolata Concezione per determinare la trasparenza del mare; Memoria del P.A. Secchi. Il Nuovo Cimento Giornale de Fisica Chim. e Storia Nat. 20, 205–238 (1865).

    Google Scholar 

  21. 21

    Lee, Z. et al. Global water clarity: Continuing a century-long monitoring. Eos (2018).

    Article  Google Scholar 

  22. 22.

    Aarup, T. Transparency of the North Sea and Baltic Sea—A Secchi depth data mining study. Oceanologia 44, 323–337 (2002).

    Google Scholar 

  23. 23.

    Feldman, G. C. & McClain, C. R. Ocean color web. In MODIS Reprocessing R2013.1 (eds Kuring, N. & Bailey, S. W.) (NASA Goddard Space Flight Center, 2014).

    Google Scholar 

  24. 24.

    Wiggins, A., Newman, G., Stevenson, R. D. & Crowston, K. Mechanisms for data quality and validation in citizen science. In 7th IEEE International Conference on e-Science Workshops, 14–19. (2011).

  25. 25.

    Dickinson, J. & Bonney, R. (eds) Citizen Science: Public Participation in Environmental Research 304 (Cornell University Press, 2012).

    Google Scholar 

  26. 26.

    Robinson, L. D., Cawthray-Sims, J. L., West, S. E., Bonn, A. & Ansine, J. Ten principles of citizen science. In Innovation in Open Science, Society and Policy (eds Hecker, S. et al.) 27–40 (UCL Press, 2018).

    Chapter  Google Scholar 

  27. 27.

    Delaney, D. G., Sperling, C. D., Adams, C. S. & Leung, B. Marine invasive species: validation of citizen science and implications for national monitoring networks. Biol. Invas. 10, 117–128. (2008).

    Article  Google Scholar 

  28. 28.

    Cooper, C. B. & Bonter, D. N. Data validation in citizen science: A case study from project FeederWatch. Front. Ecol. Environ. 10, 305–307. (2012).

    Article  Google Scholar 

  29. 29.

    Serret, H., Deguines, N., Jang, Y., Lois, G. & Juillard, R. Data quality and participant engagement in citizen science, comparing two approaches for monitoring pollinators in France and South Korea. Citiz. Sci. Theory Pract. 4, 22. (2019).

    Article  Google Scholar 

  30. 30.

    Preisendorfer, R. W. Secchi disk science; visual optics of natural waters. Limnol. Oceanogr. 31, 909–926 (1986).

    ADS  CAS  Article  Google Scholar 

  31. 31.

    Montgomery, D. C. Introduction to Statistical Quality Control (Wiley, 1991).

    MATH  Google Scholar 

  32. 32.

    Morel, A. et al. Examining the consistency of products derived from various ocean color sensors in open ocean (Case 1) waters in the perspective of a multi-sensor approach. Remote Sens. Environ. 111, 69–88. (2007).

    ADS  Article  Google Scholar 

  33. 33.

    Lee, Z. et al. Secchi disk depth: A new theory and mechanistic model for underwater visibility. Remote Sens. Environ. 169, 139–149. (2015).

    ADS  Article  Google Scholar 

  34. 34.

    Behrenfeld, M. J. et al. Revaluating ocean warming impacts on global phytoplankton. Nat. Clim. Change 6, 323–330. (2015).

    ADS  Article  Google Scholar 

  35. 35.

    Piontkovski, S. A. et al. Subsurface algal blooms of the northwestern Arabian Sea. Mar. Ecol. Prog. Ser. 566, 76–78. (2017).

    ADS  CAS  Article  Google Scholar 

  36. 36.

    Steele, J. (ed.) Ships 409–418 (Academic Press, 2008).

    Google Scholar 

  37. 37.

    Poisson, A. C. et al. Quantifying the contribution of citizen science to broad-scale ecological databases. Front. Ecol. Environ. 18, 19–26. (2019).

    Article  Google Scholar 

  38. 38.

    Levitus, S. & Mishonov, A. World Ocean Database, NOAA Atlas NESDIS 72 209 (Silver Spring, 2013).

    Google Scholar 

  39. 39.

    Smart, J. H. World-wide ocean optics database (WOOD). Oceanography 13, 70–74. (2000).

    Article  Google Scholar 

  40. 40.

    Smith, W. H. F. & Sandwell, D. T. Global sea floor topography from satellite altimetry and ship depth soundings. Science 277, 1956–1962 (1997).

    CAS  Article  Google Scholar 

  41. 41.

    Stumpf, R. P. NASA Goddard Space Flight Center, Ocean Color Group. Distance from Nearest Coastline: 0.01-Degree Grid. (Accessed November 2019).

  42. 42.

    Lam, N. S. N. Spatial interpolation methods: A review. Am. Cartograph. 10, 129–149 (1983).

    Article  Google Scholar 

  43. 43

    Wackernagel, H. Multivariate Geostatistics. An Introduction with Applications (Springer, 1995).

    Book  Google Scholar 

  44. 44.

    Sokal, R. R. & Rohlf, F. J. Biometry (W.H. Freeman and Company, 1995).

    MATH  Google Scholar 

  45. 45.

    Palmer, M. W. The estimation of species richness by extrapolation. Ecology 71, 1195–1198 (1990).

    Article  Google Scholar 

Download references


We are grateful to all of the citizen science participants in the Secchi Disk study ( who collected ZSD data while they were sailing. We thank the Whirlwind Charitable Trust ( for financial support of the Secchi Disk study charity, Lastly, we are grateful for the comments of referees.


The funding was provided by Whirlwind Charitable Trust (1200).

Author information




R.R.K. conceived the study. S.G. is the seafaring ambassador for the Secchi Disk study and collected some of the citizen data from the open-ocean. S.L. curated the Secchi Disk data. G.B. and L.K. performed the statistical analyses. R.R.K., G.B., S.L. and L.K. wrote the main manuscript text and R.R.K. and G.B. prepared the figures. All authors reviewed the manuscript.

Corresponding author

Correspondence to Richard R. Kirby.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Kirby, R.R., Beaugrand, G., Kleparski, L. et al. Citizens and scientists collect comparable oceanographic data: measurements of ocean transparency from the Secchi Disk study and science programmes. Sci Rep 11, 15499 (2021).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing