The use of a pocket camera (Nikon Coolpix) to estimate chlorophyll concentration for both water quality and plant research has been demonstrated (L. M. Goddjin-Murphy, 2009, http://www.geos.ed.ac.uk/homes/lwingate/coolpix.pdf). HydroColor has taken this a step further by using a more robust method to measure reflectance and by providing an estimate of water turbidity.
How HydroColor WorksThe methodology used used by HydroColor is simliar to that used with precision radiometers (Mobley, 1999). HydroColor uses three images collected by the user to: (1) measure how much light is emanating from the waters surface, (2) correct that value for sun glint off the surface, and (3) normalize it by the total incoming radiation. This provides a nearly illumination independent measure of the amount light reflected out of the water. This measure of the waters reflectance is known as remote sensing reflectance (Rrs). Using the camera on a smartphone, HydroColor can calculate remote sensing reflectance in the red, green, and blue (RGB) color channels of the camera (figure 1; figure 4). The absolute or relative magnitude of remote sensing reflectance in the RGB channels can then be used to measure chlorophyll and/or turbidity of a water body.
Figure 1. A Bayer filter (left) is used to cover the detector array in a color camera. The filter passes three broad bands of color in the red, green, and blue portions of the visible spectrum. The three resulting spectral sensitivity curves are plotted on the right.
The three images required to calculate the remote sensing reflectance are: an image of the water surface, an image of the sky, and an image of a photographers gray card. The gray card is a gray sheet of paper that has a known 18% reflectance. Gray cards are convenient because they are cheap and easy to find. If you cannot find a gray card locally, they are available online (one example). If you have purchased HydroColor and are unable to obtain a gray card, please contact us. Click to send us an email.
Figure 2. Example of the three images collected by a HydroColor user. These three images are used to calculate the remote sensing reflectance.
HydroColor takes advantage of the internal GPS, compass, gyroscope, and clock to compute the position of the sun in the sky. Using the suns location, HydroColor directs the user to the correct angles where sun glint off the surface is minimal. The North arrow of the compass in figure 3 should be aligned with one of the green tick marks before capturing an image. This insures the user takes both the sky and water images 135° from the azimuth angle of the sun. The green bar of the inclinometer display should be aligned with the two green triangles. This ensures the image is either taken at 40° (for the water image) or 130° (for the sky image) from nadir.
After collection of three satisfactory images, the user can select to analyze and save the data. HydroColor will save each of the three images and display the remote sensing reflectance values along with turbidity, suspended particulate matter, and the backscattering coefficient. (chlorophyll algorithm is still being developed, however, proof of concept has already been shown: L. M. Goddjin-Murphy, 2009).
Figure 3. HydroColor's user interface. The user is prompted for three images (left): Grey Card, Sky, and Water. Upon selecting a prompt, the inclinometer and compass are displayed in order to guide the user to the correct angle to capture the image (left). Once the green elements of the compass and inclinometer are aligned the capture button turns green, indicating the angles are correct.
Table 1. List of parameters derived by HydroColor along with the estimated uncertainty for each method.
|Remote Sensing Reflectance||Mobley 1999||±15% (mean absolute relative error from figure 4, for all channels)|
|Turbidity||Figure 5||±36% (mean absolute relative error from figure 5)|
|Suspended Particulate Matter||Neukermans et al. 2012||±38% (propagation of error in turbidity and the relationship between turbidity and SPM)|
|Backscatter Coefficient||Solved for bb assuming constant ap*||Gordon et al. 1998||±41% (propagation of error in SPM and Rrs)|
Figure 4. Comparison of HydroColor Rrs with Rrs measured using the Water Insight Spectrometer (WISP). The WISP spectra (Lt, Ld, & Ed) were averaged using the spectral sensitivities in figure 1 as weights. The solid line is the least squares linear regression, the dashed line shows the one to one line. Error bars display the standard error. Boxed data points are outliers identified by falling more than 3.5 standard deviations from the regression model. Outliers were likely caused by changing sky conditions during a HydroColor measurement.
HydroColor was calibrated to measure turbidity by collecting HydroColor data in parallel with a Hach portable turbidity meter. The magnitude of the HydroColor reflectance in the red increases with turbidity in mannor consistant with remote sensing theory (figure 5). The data set in figure 5 spans many different environments (ocean, river, estuary), conditions (sunny, cloudy), and platforms (iPhone 5, iPhone 4, iPod Touch). Therefore, HydroColor's calculation of turbidity is valid for many different environments and works across platforms.
Figure 5. HydroColor Rrs in the red channel as function of water turbidity. The right plot shows a close up of the low turbidity values. Error bars display the standard error (when available). The blue line shows the least squares fit to the data. The red line is the result of fitting the oceanic radiance model from Gordon et al. 1988 to the data. Fitting the model provided the average mass specific optical characteristics of the the particles encountered at the various study sites. The average bbp* (mass-specific backscattering) of the particles was 0.01 m2 g-1, which is consistant with Neukermans et al. 2012. The average ap* (mass-specific absorption) of the particles was 0.009 m2 g-1, which is consistant with Babin and Stramski 2004. The green line is the least squares fit to the TSM-Rrs(Red) data from Nechad et al. 2010. A 1:1 relationship was used to convert turbidity (NTU) to suspended particulate matter (g m-3) (Neukermans et al. 2012).
At the end of each measurement, additional information is written to a text file that can be downloaded from the device (found in HydroColor's documents folder). The data saved to the file contains: latitude, longitude, date, time, sun zenith, sun azimuth, device zenith, device azimuth, exposure values, remote sensing reflectance, turbidity, suspended particulate matter concentration, and the backscattering coefficient. A MatLab function to read in HydroColor data and images can be found here (note that this function is for the HydroColor v1.2 text file). In the future, we would like to provide the user with the option to share the data they collected by uploading it to an online database directly from their phone.
How Could HydroColor be Used in Education?The prevalence of cell phones among teachers and students provides for the possibility of water quality monitoring with minimal financial investment (In contrast, the Hach turbidity meter used to calibrate HydroColor costs ~$1,000.00). The only investment required (assuming an iPhone is available) is a photographers gray card, which can be purchased for a few dollars.
HydroColor could be used in classes to teach a variety of topics:
- Physics - light, absorption, scattering, reflectance, albedo
- Astrophysics - using location and time to know where the sun position is in the sky
- Geosciences/Environmental science - water, sediments, algae, remote sensing
- Technology - sensors, measurements
ReferencesBabin, M., Stramski, D. 2004. Variations in the mass-specific absorption coefficient of mineral particles suspended in water. Limnology and Oceanography, 49(3): 756-767.
Goddijn-Murphy, L.M., D. Dailloux, M. White, D. Bowers. 2009. Fundamentals of in situ digital camera methodology for water quality monitoring of coast and ocean. Sensors, 9(7): 5825-5843, doi:10.3390/s90705825.
Gordon, H.R., Brown, O.B., Evans, R.H., Brown, J.W., Smith, R.C., Baker, K.S., Clark, D.K. 1988. A semianalytic radiance model of ocean color. Journal of Geophysical Research, 93(D9): 10909-10924.
Mobley C.D. 1999. Estimation of the remote-sensing reflectance from above-surface measurements. Applied Optics, 38(36): 7442:7455.
Nechad, B., Ruddick, K.G., Park, Y. 2010. Calibration and validation of a generic multisensor algorithm for mapping of total suspended matter in turbid waters. Remote Sensing of the Environment, 114: 854-866.
Neukermans, G., Loisel, H., Meriaux, X., Astoreca, R., McKee, D. 2012. In situ variability of mass-specific beam attenuation and backscattering of marine particles with respect to particle size, density, composition. Limnology and Oceanography, 57(1): 124-144.