Home Articles Water quality mapping using multi-date images from digital camera

Water quality mapping using multi-date images from digital camera

H. S. Lim, K. Abdullah, M. Z. MatJafri
School of Physics, Universiti Sains Malaysia,
11800 Pulau Pinang, Malaysia
Tel: 604-6577888 Fax: 604-6579150
Email : [email protected], [email protected]

Introduction
Remote sensing techniques have been widely used for water quality studies in coastal regions and in inland lakes (Ekstrand 1992, Ritchie et al.1990, Baban 1993, Dekker and Peters 1993, Forster et al. 1993, Allee and Johnson 1999, Koponen et al. 2002). To utilize the multi-spectral radiance responses detected by the satellite sensors as a means of water quality monitoring, a model and an algorithm were required to relate the remotely sensed signals to the scattering and absorption phenomena occurring within the sea. Sea-truth data that coincide with satellite over passing the study area is very important in remote sensing studies.

However, quantitative measurements from satellite remote sensing through analysis with coincident sea-truth data are rarely conducted in the equatorial region. The main reason for this drawback is the difficulty in obtaining cloud free scenes. In this study we attempted to introduce a new method to overcome the above problem. In this study we used digital images captured by a conventional digital camera from light aircraft.

In this work, the algorithms developed for use in remote sensing applications were tested with the airborne digital images. Multi-date data were used in the multi-band algorithm calibration and validation analysis. In this work we used the digital numbers as independent variables in the calibration of the algorithm. The newly developed algorithm was used to determine the distribution of TSS in the coastal water and to generate the water quality map.

Study area
The study area is located in the vicinity of the Prai river estuary, Penang Malaysia (within latitudes 5° 22′ N to 5 ° 24′ N and longitude100o 21′ E to 100o 23′ E) as shown in Figure 1. The images were captured from a light aircraft flown at the altitudes of 3000 ft on the 28 October 2001 and 8000 ft on 9 March 2002. A digital camera (Kodak DC 290) was used as our airborne remote sensor. During the image acquisitions, the water samples were collected from a small boat within the study area.


Optical model of water
A physical model relating radiance from the water column and the concentrations of the water quality constituents provides the most effective way of analyzing remotely sensed data for water quality studies. Reflectance is particularly dependent on inherent optical properties: the absorption coefficient and the backscattering coefficient. The irradiance reflectance just below the water surface, R(l), is given by Kirk (1984) as

R(l) = 0.33b(l)/a(l) —————————-(1)
where
l = the spectral wavelength
b = the backscattering coefficient
a = the absorption coefficient

The inherent optical properties are determined by the contents of the water. The contributions of the individual components to the overall properties are strictly additive (Gallegos and Correl, 1990). For a case involving two water quality components, i.e. chlorophyll, C, and suspended sediment, P, the simultaneous equations for the two channels given by Gallie and Murtha (1992) can be expressed as

where
bbw(i) = backscattering coefficient
bbc* = chlorophyll coefficient
bbp = sediment coefficient
aw(i) = absorption coefficient
ac* = chlorophyll specific absorption coefficient
ap* = sediment specific absorption coefficient
C = chlorophyll
P = suspended sediment

Regression Algorithm
TSS concentration can be obtained by solving the two simultaneous equations to get the series of terms R1 and R2 that is given as

P = ao+a1R1+a2R2+a3R1R2+ a4R12+ a5R22+ a6R1 2R2+ a7R1 2R22+a8R12R22+… (3)
where aj, j = 0, 1, 2, … are the coefficient for equation (3) that can be solved empirically using multiple regression analysis. This equation can also be extended to the three-band method given as

P = eo+e1R1+e2R2+e3R3+ e4R1R2+ e5R1R3+ e6R2 R3+ e7R12+e8R22+e9R32 ——————(4)
where the coefficient ej, j = 0, 1, 2, … can also be solve empirically.

Data analysis and results
The colour digital images of the study area captured by the digital camera contained of 1792 x 1200 pixels. The images were separated into three bands assigned as red, green, and blue bands. The separated bands were stored in raw data format in order to facilitate further analysis using in-house programs and PCI software. Ground control points (GCP) were determined by using a geometrically corrected SPOT satellite map as the reference geocoded image. Figure 2 below shows the image of 9 March 2002 used in this study.

  1. Atmospheric Correction
  2. The image of 28 October 2001 was taken obliquely. The view angle correction was first performed to correct the angular dependence of image brightness. Then the multi-date correction technique was carried out. The vertical image captured on 9 March 2002 (Figure 2) was used as the reference image. Different types of materials visible in the reference image and the image to be corrected (image on 28 October 2001) were selected as correction targets. We assumed the reflectance of these targets did not change with time. The DNs of the targets in the second image were related to the reference image. The relationship was then used to adjust the DN values of the second image. This correction forced the image to have the same atmospheric conditions and the multi-date analyses were then performed using these data sets.

  3. Calibration and Validation
  4. The DN values corresponding to the water sample locations were extracted from both images. The DNs for window size of 7 by 7 were used for the calibration of the algorithm. This window size was selected because the data set produced higher correlation coefficient and lower RMS value. Half of the data were chosen randomly for the calibration data set. The remaining data were used in the validation analysis. The relationship between TSS and DN for the calibration data set is shown in Figure 3. Superior result was produced by the proposed model, which achieved the correlation coefficient of about 0.9988 as shown in Figure 4. The Accuracy of the proposed calibrated algorithm was then assessed using the validation data set. The analysis produced high correlation coefficient (R=0.95) between the predicted and the sea-truth values as shown in Figure 5.

    Various water quality algorithms were also tested and their accuracies (R and RMS values) were noted (Keiner and Yan 1998, Pulliainen et al. 2001). Table 1 shows comparison for various algorithms. The proposed algorithm produces higher correlation coefficient (R=0.9988) between the predicted and the measured TSS values and lower RMS value (1.5233 mg/l) compared to other algorithms.

  5. TSS Map
  6. The TSS map was generated using the proposed calibrated algorithm. The map was then geometrically corrected by using the cubic convolution method. This method was used because it produced a smoother geocoded image. The generated map was then filtered by using 5 by 5 pixels average for removing the random noise. Then, the colour-code was produced for visual interpretation as shown in Figure 6.

    Table 1. Regression results using different forms of algorithms for TSS

    Algorithm R RMS (mg/l)
    TSS=a0+a1B1+a2B12 0.978929 3.921557
    TSS=a0+a1B2+a2B22 0.946784 6.217749
    TSS=a0+a1B3+a2B32 0.626099 14.48568
    TSS=a0+a1lnB1+a2lnB12 0.978315 3.392354
    TSS=a0+a1lnB2+a2lnB22 0.947629 6.322768
    TSS=a0+a1lnB3+a2lnB32 0.626179 14.49076
    TSS=a0+a1(B1/B3)+a2(B1/B3)2 0.969742 4.614237
    TSS=a0+a1(B1/B2)+a2(B1/B2)2 0.954358 5.745689
    TSS=a0+a1(B2/B3)+a2(B2/B3)2 0.587367 15.05422
    TSS=a0+a1ln(B1/B3)+a2ln(B1/B3)2 0.968607 4.725158
    TSS=a0+a1ln(B1/B2)+a2ln(B1/B2)2 0.959635 5.343059
    TSS=a0+a1ln(B2/B3)+a2ln(B2/B3)2 0.536563 16.03402
    TSS=ao+a1(B1-B3)/B2+ a2((B1-B3)/B2)2 0.936483 6.50917
    TSS=ao+a1(B2-B3)/B1+ a2((B2-B3)/B1)2 0.714773 12.97994
    TSS=ao+a1(B1-B2)/B3+ a2((B1-B2)/B3)2 0.945939 6.020821
    TSS=ao+a1(B1+B3)/B2+ a2((B1+B3)/B2)2 0.954358 5.546343
    TSS=ao+a1(B2+B3)/B1+ a2((B2+B3)/B1)2 0.960573 5.159105
    TSS=ao+a1(B1+B2)/B3+ a2((B1+B2)/B3)2 0.950737 5.7523
    TSS=ao+a1(B2-B1)/(B1+B2)+ a2(B2-B1)/(B1+B2)2 (Waldron M. C. et al.) 0.952628 5.64638
    TSS=ao+a1(B2-B3)/(B2+B3)+ a2(B2-B3)/(B2+B3)2 (Waldron M. C. et al.) 0.587537 15.01895
    TSS=ao+a1(B1-B3)/(B1+B3)+ a2(B1-B3)/(B1+B3)2 0.968039 4.653643
    TSS=ao+a1ln(B2-B1)/(B1+B2)+a2ln(B2-B1)/(B1+B2)2 (Waldron M. C. et al.) 0.960365 5.17469
    TSS=a0+a1(B2-B1) +a2(B2-B1)2 0.889400 6.173401
    TSS=a0+a1(B2-B3) +a2(B2-B3)2 0.549636 19.74538
    TSS=a0+a1(B1-B3) +a2(B1-B3)2 0.964572 4.897456
    TSS=ao+a1(B1+B2)+ a2((B1+B2)/2)2 (Waldron M. C. et al.) 0.981580 3.576295
    TSS=ao+a1(B1+B3)/2+ a2((B1+B3)/2)2 (Waldron M. C. et al.) 0.959792 5.238677
    TSS=ao+a1(B2+B3)/2+ a2((B2+B3)/2)2 (Waldron M. C. et al.) 0.831625 10.33726
    TSS=ao+a1B1+a2B2+a3B3+a4B1B2+a5B1B3+a6B2B3+a7B12+a8B22+a9B32 (Proposed) 0.996633 1.52333

    Note: B1, B2 and B3 are the digital numbers for red band, green band and blue band respectively

  7. Qualitative comparison
  8. The contour map was plotted using sea-truth data with Kriging method. Quantitative comparison between contour map plotted using the sea-truth data and image generated by the algorithm gives acceptable result. Both images show almost the same pattern as can be seen in Figure 6 and 7. This shows the capability of the algorithm to generate TSS concentration map. Generally, TSS is higher in the river estuary closer to the beach as shown by both maps. However, the map generated using the algorithm shows that the concentration of TSS near to the beach is more than 250mg/l, whereas the contour map plotted based on sea-truth data gives the range of 201 to 250 mg/l. The difference is due to the limited sea-truth data used to generate the contour map by the interpolation technique.

Conclusion
A digital image captured by conventional digital camera can be used to generate water quality mapping. This technique will reduce the cost in acquiring the airborne imaging. The problem of cloud cover can also be avoided because the light aircraft, from where the image is being captured, normally flies below the cloud levels. A digital camera that can capture digital images will provide multi-band data by separating the colour images into individual components. The proposed image correction techniques produced encouraging results with high value of correlation coefficient.

A new multi-spectral algorithm has been developed for mapping the total suspended solid by using the digital images capture from the light aircraft. Multi-date sea-truth data also can be used to validate the algorithm.


Figure 7. Contour map of TSS for the study area. (Orange, TSS=150-200 mg/l; red, TSS= 201-250 mg/l; white,TSS>250 mg/l.)

References

  1. Alle, R.J., and Johnson, J.E., 1999, Use of satellite imagery to estimate surface chlorophyll-a and Secchi disc depth of Bull Shoals, Arkansas, USA. International Journal of Remote Sensing, 20, 1057-1072.
  2. Baban, S.M., 1993, Detecting water quality parameters in the Norfolk Broads, U. K., using Landsat imagery. International Journal of Remote Sensing, 14, 1247-1267.
  3. Dekker, A.G., and Peters, S.W.M., 1993, The use of Thematic Mapper for the analysis of eutrophic lakes: a case study in the Netherlands. International Journal of Remote Sensing, 14, 799-821.
  4. Ekstrand, S., 1992, Landsat TM based quantification of chlorophyll-a during algae blooms in coastal waters. International Journal of Remote Sensing, 13, 1913-1926.
  5. Forster, B.C., Xingwei, I.S., and Baide, X., 1993, Remote sensing of water quality parameters using landsat TM. International Journal of Remote Sensing, 14, 2759-2771.
  6. Gallegos, C. L., and Correl, D. L., 1990, Modeling spectral diffuse attenuation, absorption and scattering coefficients in a turbid estuary. Limnology and Oceanography, 35, 1486-1502.
  7. Gallie, E. A., and Murtha, P. A., 1992, Specific absorption and backscattering spectra for suspended minerals and chlorophyll-a in Chilko Lake, British Columbia. Remote Sensing of Environment, 39, 103-118.
  8. Keiner, L. E., and Xiao, H. Y, 1998, A neural network model for estimating sea surface chlorophyll and sediments from thematic mapper imagery. Remoter Sensing of Environment, 66, 153-165.
  9. Kirk, J. T. O., 1984, Dependence of relationship between inherent and apparent optical properties of water on solar altitude. Limnology and Oceanography, 29, 350-356.
  10. Koponen, S., Pulliainen, J., Kallio, K., and Hallikainen, M., 2002, Lake water quality classification with airborne hyperspectral spectrometer and simulated MERIS data. Remote Sensing of Environment, 79, 51-59.
  11. Ritchie, C. J., Cooper, C. M., and Yong, Q. J., 1987, Using Landsat Multispectral Scanner data to estimate suspended sediments in Moon Lake, Missisippi. Remote Sensing of Environment, 23, 65-81.