GISdevelopment.net ---> AARS ---> ACRS 1999 ---> Environment

Spectral Model of Water Quality parameters

Yin Qui, Shu Xiaozhou, Kuang Dingbo
Shanghai Institue of Technology Physics, Chinese Academy of sciences

Lake Taihu, in Changjiang Delta Area, is the third largest plain fresh lake of China. In recent years, the water quality of Taibu descends year by year. Especially, the north area of Taihu is in obvious nutritive state. In this paper, according to the field measurement and the Landsat TM Data, the relations between water surface reflected spectra and water pollutant concentration and the change of water quality from 1986 to 1998 about Taihu are studied.

Two field experiments are made. The first is in Nov. 1997 with 21 measurement points distributed in whole Taihu, and the second is in Aug. 1998 with 13 measurement points distributed in the north areea of Taihu. Two water quality parameters, chlorophyll-a (Chl) and suspended substance (SS), are analyzed. According, to the ratio of reflected spectra from water surface and that from a standard white plate measured by GER-1500, an ultra-spectra instrument with 346 channels from 350nm wavelength to 900nm wavelength, the reflectivity (diffusion reflectivity) spectra of water surface are determined. BY statistical regression, the ultra and multi-spectral models about water quality parameters are established for different pollutant concentration ranges. The relation model between reflectivity and Chl and SS is established for every GER -1500 channel. The relation models between the reflectivity of one channel or the reflectivity differences/ratio of two channels and the water quality parameters are established for remote sensing channels of several existing satellite and a set of hypothetical water quality remote sensing channels.

The ultra-Spectral Model of Water Quality Parameters for GER-1500 Channels. The scattering effect of chlorophyll can be expressed by the linear item Chl. The absorption effect of chlorophyll can be expressed by the linear item In(Chl+1). The linear item SS(for SS < 100mg/L) or the log-linear item In(SS+1) can express the effect of suspended substance. The scattering effect of chlorophyll is mainly reflected at wavelengths larger than 740nm. The absorption effect of chlorophyll is mainly reflected at 400~520nm and 575~690nm wavelengths. The scattering effect of suspended substance is reflected at 680 820nm wavelengths most obviously. If S<100mg/L, the scattering of S will affect of Chlorophyll will also reflected at wavelengths in the vicinity of 350nm and 550nm.

The Multi-Spectral Model of Water Quality Parameters for TM Channels. For samples of Chl=24~500ug/L and S=0~100mh/L, Scan not be retrieved by the reflectivity, reflectivity difference or reflectivity ratio of any TM channels. If the single channel model is applied, TM4 is the most suitable channel for Chl retrieving and the corresponding model, TM4 =C0+C1*Chl with a related coefficient 0.86. if the double channel model is applied, every combination of TM channels except the combination of TM1 and TM3 (which are the absorption channels of Ch1) can reflect the effect of Ch1, in which the combinations of TM4 and TM3, TM2 or TM1 have relatively good correlation with Ch1. The related coefficients of TM4 - (TM1, TM2 or TM3)= C0+C1*Ch1 are not smaller than 0.85 and the related coefficients of TM4/ (TM1, TM2 or TM3)= C0+C1*Chl are not smaller than 0.80. For samples of Ch1=0~24ug/L and SS=0~200nm/L, Ch1 can not be retrieved by any TM channels. TM4 and TM3 have log-linear correlations with S to a certain extent, a and the related coefficients are 0.57 and 0.49 respectively. Some double channel combinations can reflect the effect Ch1. the combination of the absorption channels of Ch1, TM1 and TM3, can retrieve SS fairly good. The related coefficients of TM3-TM1=Co+C1*In(SS+1) is 0.85 and the related coefficient of TM3/TM1=Co+C1*in(Ch1+1)+C2*In(SS+1) is 0.90 with FCh1=10.45 and FSS=47.20. For samples of Ch1=0~500ug/L and SS=0~200mg/L, the reflectivity of every TM channels are related with Ch1 and SS, in which the relation of reflectivity with S has a long -linear form. The combination of the near infrared channel TM4 and the visible channel, TM1 TM2 or TM3, can filter the effect of SS and retrieve Ch1 fairly good with a related coefficients not smaller than 0.84. The combination of the green channel TM2 and the absorption channel of Ch1, TM1 or TM3 is related with Ch1. In addition, the combination of TM2 and TM3 is related with SS. the difference of the absorption channels of Ch1, TM3 and TM1, can filter the effect of Ch1 and reflect the effect of SS.

The data retrieving steps from Landsat TM data to water quality parameters are (1). The change of grey frequency (f) with grey (N) is analyzed for every TM channels. If d2f/d2N attains maximum and when the grey of the reflectivity of atmosphere layer to the direct sunlight at the top of atmosphere (Rsun) is set to 1;(2). The relations between the direct transmittance, the diffusion transmittance and the diffuse reflectivity of atmosphere layer at different directions (up and down) and Rsun are established by an atmospheric Radiative transfer theory, which is used to determine the reflectivity of different TM channels at water surface from TM data; (3). The concentration of Ch1 and that of S are retrieved by the reflectivity about channel TM1 TM3 and TM4 at water surface and the multi-spectral model of water quality with TM4/TM3=C0+C1*Ch1 and TM3/TM1=C0+C1*S; (4). The change of concentration frequency (f) with concentration (N) for Vh1 and that for S are analyzed respectively. If d2f/d2N attains maximum and åNO "1% when N=J<0, the concentration is adjusted from N to N+J. As an example the change of water quality about Taihu is determined by the Landsat TM data on 1986.07.25, 1997.06.29, 1997.05.04 and 1998.08.11.