1、1007-4619 (2012) 01-176-18 Journal of Remote Sensing 遥感学报Pure water absorption coefficient measurement after eliminating the impact of suspended substance in spectrum from 400 nm to 900 nmDENG Ruru, HE Yingqing, QIN Yan, CHEN Qidong, CHEN LeiSchool of Geography and Planning, Centre for Remote Sensin
2、g and Geographical Information Sciences, Sun Yat-sen University, Guangdong Guangzhou 510275, ChinaAbstract: In this paper, we designed a new device which can be used to measure the absorption coefficient of pure water di-rectly. First, we measured the irradiance of direct light penetrating through d
3、ifferent thicknesses of water layers. Then we used the ratio method to eliminate the impact of measuring instruments on experimental results. Finally, we acquired the extinction coefficient. The extinction coefficient experiment proved that extinction coefficient of suspended substance in water can
4、be calcu-lated by formula of ngstrom. We therefore put forward a method on how to eliminate the impact of suspended substance in pure waters. Eventually, we determined the absorption coefficient of pure water from 400 nm to 900 nm. Comparing with previous studies, overall results are consistent, but
5、 with better precision in the long wave length. The absorption coefficient of pure water can be used as the normal data for water quality remote sensing.Key words: absorption coefficient, pure water, suspended substance, extinction coefficientCLC number: TP701 Document code: AReceived: 2011-07-27; A
6、ccepted: 2011-10-22Foundation: The National Natural Science Foundation of China (No. 40671144); The National High Technology Research and Development Program of China(863 program) (No. 2006AA06Z416); The Introduction of International Advanced Agricultural Science and Technology Projects of China (94
7、8 Program) (No. 200820)First author biography: DENG Ruru (1963 ) male, Ph.D., professor his research interests are remote sensing for water quality and atmospheric envi-ronment. He has published more than 40 papers. E-mail: 1 INTRODUCTIONSince water plays an important role in many scientific disci-p
8、lines, e.g., chemistry, biology, meteorology and absorption spec-trum of pure water has a wide use in many areas. Large amount of existing studies (Morel, 1974; Querry Patel Tam Querry Buiteveld Lee Sydor Ruddick Gallegos Kirkpatrick Doxarean Tws, Tgare the transmissions ratio of water surface and t
9、he bottom of Glass vat respectively; cdis the distance attenuation factor,which is a factor of the irradiance decreasing with the increase of the distance,and it depends on the distance between the light source and the standard board; Rbis the reflectance of standard board. e-is the transmis-sions r
10、atio of water layer, is the optical thickness of water layer, it can be expressed as: = hk (2)where h is the thickness of tested water layer, k is the extinction coefficient of water.(2) Scattered lightAs shown in Fig. 2, most of downward scattered light is blocked by Scattered light filter, and onl
11、y a part of scattered light can pass through the hole of Scattered light filter and reach the standard board. However, a large part falls out of the detected range of Spec-trometer because refractive index of water is obviously larger than the air, and the scattering angle is enlarged by the water-a
12、ir surface. Only a small part of scattered light with scattering angle smaller than 10falls in the detected range of Spectrometer. The character-istics of this part of light are almost the same with the direct pen-etrating light. As shown by the works of Gordon (1993), measure-ments of the absorptio
13、n coefficient and volume-scattering function 178 Journal of Remote Sensing 遥感学报 2012,16(1)for scattering angles 15of water are sufficient for predicting the transport of irradiance in ocean, and it is not necessary to measure small-angle scattering in many applications. Thus, the influence of scatte
14、ring light in this apparatus system is negligible.Fig. 2 Diagram of light pathTghScattered lightE0Standard boardScattered light filterProbe ofspectrometerGlass vatdiverging lightfilterLight sourceLensesScattered lightwaterRbETws2.3 CalibrationThe detector of the FieldSpec3 ASD spectrometer is compos
15、ed of three groups of detect elements. The first one is the group of vis-ible spectrum in the range of 3501000 nm, the second is the near infrared spectrum of 10011700 nm, and the third is the infrared spectrum of 17012500 nm. The gain of visible spectrum group often appears a deviation for the grea
16、t contrast between light passed through water and that illuminated directly. As shown in Fig. 3(a), some curves measured from radiance of water layers with different depths appear a jump at 1000 nm. One of the important character-istics of the deviation is that the gain of deviation for the whole vi
17、sible spectrum is a constant, so it can be rectified according to the adjacent values of near infrared spectrum. Radiance of visible spec-trum got by ASD spectrometer can be calculated as:(3)where a is the constant of error of gain, L is the original measured radiance in visible spectrum. Constant a
18、 can be calculated as Eq. (4):(4)where Lvand Liare radiances of visible and near infrared spectrum, respectively. =1000 nm is wavelength, =1 nm is increasing step of wavelength, L()is the average increment of measured radiances.(5)Rectified radiance spectrums of water layer with different depths are
19、 show in Fig. 3(b), without the jumps of measured radi-ance at 1000 nm.2.4 Calculation of extinction coefficientAfter calibration, radiance measured by ASD spectrometer can be written as Eq. (1). To avoid measurements of Tws, Tgand cd, we take ratios of two radiance L1, L2of different water depths h
20、1, h2to compute the optical thickness. Optical thickness between depth h1 and h2of water is:(6)The E0, TWS, Tgand cdin Eq. (1) are removed by taking the ratio. Absorption coefficient of pure water in the 400 nm to 900 nm wavelength region is not large. Therefore, there is no need for such a thin wat
21、er layer. A total of 14 groups of experimental data were measured, with the thinnest and thickest thickness of the water layer of 0.5 cm and 15 cm, respectively. Radiances of water layer with the thicknesses larger than 1cm are divided by the radiance of 0.5 cm to acquire the optical thickness for t
22、he spectrum between 3501100 nm. Then extinction coefficient of water is: (7)0.120.100.080.060.040.020350 600 850 1100 1350 1600 1850 2100 2350radiance(W/m2.nm.sr)radiance(W/m2.nm.sr)(b) After rectify0.120.100.080.060.040.020350 600 850 1100 1350 1600 1850 2100 2350Wavelength /nm Wavelength /nm(a) Be
23、fore rectifyStandard board 0.5 cm 5 cm 10 cm 15 cmFig. 3. Diagram of original measured radiance of water layers with different depths before (a) and after (b) rectify179DENG Ruru, et al.: Pure water absorption coefficient measurement after eliminating the impact of suspended substance in spectrum fr
24、om 400 nm to 900 nmSeveral results of extinction coefficient of pure water with dif-ferent thicknesses and from independent experiments are shown as in Fig. 4. Three characteristics are observed on this figure:(1) The extinction coefficients of water we measured with different depths from independen
25、t experiments have very good coherence as a whole, but still have obvious errors at different regions.(2) The er-rors between different extinction coefficient spectrums in optical re-gion present obvious systematic displacements. As stated above, it is due to the suspending substance in the water. (
26、3) Signal-to-Noise (SNR) of water extinction coefficient measured with the depth greater than 13 cm on spectrum 960 to 1010 nm is very low, be-cause the values of extinction coefficient are too great in this region and the energy of light penetrating through the water is so small and out of the dete
27、cting range of spectrometer (Fig. 4). Therefore, we only used the data whose wavelengths are shorter than 900 nm to represent the final results.50403020100350 450 550Extinctioncoefficient/m-1650 750 850 950 1050Wavelength/nmk_2 cm k_3 cm k_4 cm k_5 cm k_6 cmk_7 cm k_8 cm k_9 cm k_10 cm k_11 cmk_12 c
28、m k_13 cm k_14 cm k_15 cmFig. 4 Extinction coefficient of water measured with different thicknesses of water layer 3 SEPARATING ABSORPTION SPECTRUM OF PURE WATER FROM THE EXTINCTION COEFFCIENT OF SUSPENDING SUBSTANCE3.1 Composition of extinction coefficient of waterThe extinction coefficients of wat
29、er are composed of absorption coefficient of water aw, scattering coefficient of molecular of water aw and extinction coefficient of suspending substance ks. Suspend-ing substance is the particles that mainly come from the air during the course of preparing and the experiment. Thus the values of ex-
30、tinction coefficients of water we acquired are obviously larger than its actual values. It can be written as:k=aw+bw+ks(8)Molecular scattering coefficient of water bwis a stable and rela-tive simple parameter. Based on the result measured by Smith and Baker (1981), it can be calculated as:bw=0.00014
31、54.32 (9)The extinction coefficient of suspending substance ks varies ac-cording to the concentration and the granularity of suspending sub-stance in the water. 3.2 Experiment on extinction coefficient of water suspending substanceThe optical thickness of particles in the air can be written with the
32、 formula of ngstrom (Cachorro at each wavelength, values are given for the absorption coefficient, and its statistical standard deviation. The absorption coefficient is also shown in Fig. 9. Fig. 8 Absorption coefficient curve of water (350 nm900 nm)10864201086420350 350450 450550550650 650750 75085
33、0 850k_NO5k_NO4k_NO3k_NO2k_NO1 w_NO5w_NO4w_NO3w_NO2w_NO1Wavelength/nmExtinctioncoefficient/m-1Absorptioncoefficient/m-1Wavelength/nm(a) Extinction coefficient(b) absorption coefficients of pure water after eliminate extinction coefficient of suspending substance Wavelength/nmWavelength/nm8642010.001
34、0.010.110Absorption coefficient/m-1Absorption coefficient/m-1400 450 500 550 600 650 700 750 800 850 900400 450 500 550 600 650 700 750 800 850 900(a) Conventional coordinates (b) Logarithmic coordinatesFig. 9 Absorption coefficients of pure water between 400 nm and 900 nm 182 Journal of Remote Sens
35、ing 遥感学报 2012,16(1)Table 2 Absorption coefficients wand standard deviations , for pure water as a function of wavelength /nm w/m-1/m-1/nm w/m-1/m-1/nm w/m-1/m-1400 0.0133 0.0049 570 0.0742 0.0031 740 2.6494 0.0224405 0.0112 0.0028 575 0.0871 0.0033 745 2.7595 0.0131410 0.0096 0.0035 580 0.1058 0.003
36、6 750 2.7746 0.0122415 0.0084 0.0040 585 0.1276 0.0042 755 2.7709 0.0140420 0.0076 0.0020 590 0.1502 0.0025 760 2.7569 0.0162425 0.0073 0.0025 595 0.1861 0.0033 765 2.7301 0.0177430 0.0075 0.0020 600 0.2374 0.0042 770 2.6868 0.0198435 0.0079 0.0023 605 0.2732 0.0043 775 2.6232 0.0218440 0.0080 0.001
37、8 610 0.2849 0.0031 780 2.5390 0.0234445 0.0082 0.0016 615 0.2913 0.0031 785 2.4392 0.0245450 0.0085 0.0015 620 0.2974 0.0026 790 2.3285 0.0259455 0.0087 0.0015 625 0.3039 0.0025 795 2.2157 0.0271460 0.0090 0.0019 630 0.3104 0.0020 800 2.1151 0.0276465 0.0094 0.0013 635 0.3179 0.0019 805 2.0441 0.02
38、74470 0.0099 0.0021 640 0.3264 0.0017 810 2.0236 0.0287475 0.0105 0.0024 645 0.3360 0.0014 815 2.0626 0.0290480 0.0112 0.0033 650 0.3491 0.0012 820 2.1717 0.0273485 0.0118 0.0023 655 0.3725 0.0010 825 2.4206 0.0216490 0.0129 0.0024 660 0.4033 0.0019 830 2.8964 0.0276495 0.0145 0.0025 665 0.4223 0.00
39、05 835 3.4215 0.0462500 0.0166 0.0023 670 0.4303 0.0001 840 3.7419 0.0365505 0.0225 0.0031 675 0.4382 0.0007 845 3.9149 0.0315510 0.0299 0.0030 680 0.4502 0.0015 850 4.0587 0.0313515 0.0354 0.0033 685 0.4689 0.0014 855 4.1958 0.0320520 0.0378 0.0036 690 0.4975 0.0027 860 4.3248 0.0317525 0.0397 0.00
40、34 695 0.5413 0.0025 865 4.4585 0.0308530 0.0415 0.0040 700 0.6023 0.0028 870 4.6160 0.0290535 0.0438 0.0038 705 0.6835 0.0038 875 4.8151 0.0280540 0.0471 0.0039 710 0.7960 0.0057 880 5.0604 0.0293545 0.0517 0.0040 715 0.9607 0.0062 885 5.3365 0.0296550 0.0579 0.0044 720 1.1831 0.0054 890 5.6197 0.0
41、287555 0.0630 0.0040 725 1.4600 0.0088 895 5.9007 0.0286560 0.0670 0.0044 730 1.8413 0.0108 900 6.1884 0.0256565 0.0698 0.0040 735 2.3130 0.02004.2 Standard deviationsSince the method we used is quite stable, the standard devia-tions are generally small. But in the region from 400 nm to 500 nm, perc
42、ent errors are obvious larger. The main reason is that the intensity of source light energy is very low in the spectrum, leading to the increasing of percent error along with the decrease of wave-length. Then, water absorption in this region is much more sensitive to dissolved organics in water. Eve
43、n most slight contamination of organics will lead to the obviously increasing of absorption. Slight contamination of organics in different measurements was the other causes which lead to the increasing of percent error in this region. So the results of absorption spectrum between 350 nm to 400 nm we
44、re abandonde for too large standard deviation.4.3 Error analysis and discussionFig.10 shows the results of absorption coefficient of pure wa-ter between 400nm and 800nm plotted with those from Smith and Baker (1981) ( )Pope and Fry (1997) (+), Sogandares and Fry (1997) (), Hale and Querry (1973) (),
45、 Kou and Labrie and Chylek (1993) (x), and Palmer and Williams (1975) ( ). It is obvi-ously in the figure that our result of absorption spectrum of pure water in the region of 450-700nm is very close to that of Pope and Fry and that from Smith and Baker. In order to get the nuances of low areas of a
46、bsorption spectrum of pure water, we used logarithmic coordinate in Fig. 10 (b). From the figure it is clear that, in the 450730 nm spectral range, the results are so close to the results from Sogandares and Fry and Pope and Fry with very small difference; and the resonance absorp-tion band of water
47、 molecules in the 400700 nm has been clearly shown. There is disagreement below 450 nm with the results given by Pope and Fry. This disagreement is most likely due to: (1) in the present experiment, the temperature of the water sample was 261C, where in the Pope and Fry experiment, it was actively a
48、t a temperature of 221C. Increase in temperature may cause slight 183DENG Ruru, et al.: Pure water absorption coefficient measurement after eliminating the impact of suspended substance in spectrum from 400 nm to 900 nmincrease in results (Quickenden Trabjerg Pegau Fry, 2000; Quickenden & Free-man,
49、2000). (2) the influence of trace organics in the water. Despite we used ultrapure water in the test and have made as much effort to keep it free from being contaminated as we can, some trace or-ganics still exist in water, that leads to the slightly increasing and undulating of absorption in the region below 450 nm.Though our measured datum of absorption coefficient of wa-ter is obviously larger than that from Pope and Fry data, but is smaller than those from Smith and Baker. So the disagreement is limited.Fig. 10(a) a
