RELATION OF ENVIRONMENTAL AND WATER QUALITY PARAMETERS USING REMOTELY SENSED DATA

Abstract

In the last few decades man's understanding of the environment and his awareness of the dangers of environmental pollution have increased enormously. The sources of pollution in any water body are natural as well as man made. Chemical, bacterial, organic, and thermal pollution have been given a lot of attention. Water quality of rivers, reservoirs, lakes or estuaries undergoes significant changes •owing to seasonal variations, confluence with tributaries, groundwater inflows and entry of wastewaters at the various urban centres all along their cources. The erosion of land surface, transportation of eroded materials, channels, harbors, lakes, and reservoirs are some of the important problems associated with water resources development. Identification of these sources and quantifying their effects is a major task for environmental technologists and water resource planners. Because of todays changing pollution problems, stream water quality monitoring must be extended to techniques which can be used to get rapid, reliable, and repetitive data for a better quality interpretation and management of our water resources. In this study the possibilities and applicability of using remotely sensed data in water quality monitoring is explored. Laboratory based studies related to the reflectance response of suspensions carrying different soil types and particle sizes of varying concentrations indicate dependence of the reflectance measurements on the soil type characteristics, particle size and concentration. (iii) Clarity is an extremely important aspect of water quality. Clarity level of any surface water body vary considerably in quality and quantity owing to the various sources of entry of suspended sediment materials (Bhargava and Mariam, 1990e,f). Water clarity in terms of suspended sediments concentration (mg/1), turbidity (NTU) and modified secci depth (cm) were measured. The largely varied suspended sediment concentration of the test solution ranged from 20 mg/1 to 1280 mg/1, turbidity level between 18 and 255 NTU and modified secci depth 4 cm to 55 cm, Zenith angle (©J 20° to 55° , salinity 50-3200 mg/1. Different sediment concentrations were obtained by adding weighed amounts of soils into a 50 cm by 50 cm by 75 cm high tank. The tank was painted biack in all its inside to minimize side reflectance. the samples were kept in suspension by continous agitation with a stirrer. A 1000 W tungsten lamp was used to illuminate the tank uniformly. Using spectroradiometr measurements of spectral reflectance were taken at wavelength ranging from 500 to 1000 nm at a 50 nm interval. To make this study more useful over a wide range of soil conditions sixteen soil types were used. These include 1) A local alluvial soil of the Gangetic plane (Roorkee soil), 2) Black cotton soil (montmorillonite clay), 3) Bentonite, which occurs in partially weathered volcanic deposits and in the more arid regions ,of the world, 4) Grey soil (known in India as "Dhauri clay" and 5) Kaolines (soils of humid-temperate and humid tropical regions). The dry soil samples were sieved and the geometric mean diameter of the soils particles used for this studies include 0.032, 0.046, 0.704, 0.817, 0.0962 and 0.1253 mm. The physical and chemical properties of the various soil types used in the study were determined. These include: specific gravity, .PH, volatile (iv) matter, colour, cation exchange capacity, carbonate content, A12U3, Si02, MgO, MnO, Fe203, ZnO, CaO, K20, Na20 and clay mineralogy. For many of the existing remote sensing settelites Sun is the source of energy. The actual position of the Sun will vary according to the time of the year and the reflectance value is observed to depend on light Zenith Angle (ez)- The effect of changes in 0^ values on the spectral reflectance measurements to predict the suspended sediment level was studied. The experimental results also show that reflectance value depend on salinity level of the water body. The models developed by previous investigations did not fully account for spatio-temporal variations in the type, size and concentration of the sediments, the variation in the &7 values, mineral composition and salinity level. Because of this, all previously developed predictive models are applicable only to the study area and extrapolating the models to predict water quality parameters outside the immediate study area cannot be done. This shows the need to study and isolate the effect of these variations on reflected energy and quantify some of the ,major interferring properties and incorporate them in the predictive models. Then only a model developed based on one time observation can be used for comparison of spatio-temporal variation of any parameter based on subsequent remotely sensed data will be possible. THE EFFECT OF SEDIMENT CONCENTRATION ON REFLECTANCE Based on the observed data the relationships of reflectance with wavelength (500-1000 nm range) in respect of the various soil types for different suspended sediment concentration, turbidity (v) levels and modified secci depth of the various particle sizes were observed. A significant relationship exists between a wide range of suspended material concentration and reflectance (Bhargava and Mariam, 1988). The reflectance for any soil type and size increases with suspended sediment concentration and turbidity level but decreases with an increase in the modified secci depth of the suspension. Distinct and more pronounced differences in reflectance values corresponding to the various sediment concentration were observed in the 700 to 900 nm range (Bhargava and Mariam, 1990a). This is an ideal region for prediction of a wide variation of sediment concentration (C ), turbidity (T) and £30 modified secci depth (SD) from reflectance measurements. The above functional relationship have been modelled for the observed data and are presented in Eq. la-c. Css = al R*2 -..-da) T = a3 Ra4 (lb) SD = a5 Ra6 (lc) In Eq. la-c, C represented the suspended sediments DO concentration (mg/1), T the turbidity level (NTU), SD the modified secci depth (cm) and R the reflectance. The linearized forms of equations la-c were regressed and the function coefficients al'a2'a3'a4'a5 and a6 were determined through a linear regression of the data. The computed function coefficients values of Eq. la-c are different for different particle sizes but one soil type and different soil types but one particle size. Hence, it establishes that the soil type characteristics as well as the particle size play a dominant role in affecting the reflection characteristics of a suspension (Bhargava and Mariam, 1990b). (vi) THE EFFECT OF PARTICLE SIZE VARIATION ON REFLECTANCE Sediments of different properties and particle sizes are encountered at the different reaches of a river or in different seasons in a reservoir. Reflectance is a function of particle size. The relationship of reflectance with wavelength in respect of the different particle sizes for the five soil types of the same suspended sediment concentration was studied in the wavelength range of 500-1000 nm. It was observed that a decrease in particle size results in an increase in reflectance. This is because for any given concentration, fine grained sediments contains more number of particles and will have larger surface area and thus scatters more than would an equal weight of coarse grained sediments. The effect of particle size on reflectance measurement is almost equally appreciable in the entire 500-1000 nm range. The functional relationship of particle size variation and reflectance for the given sediment concentration and soil type has been modelled and expressed by an equation of the type shown in Eq. 2. R = a? + aQ 1/D (2) In Eq . 2, D represents the geometric mean diameter (mm), the function coefficients a7 and a8 were computed for all soil types through linear regression of the observed data at 700 nm wavelength and 1280 mg/1 suspended sediment concentration. The soil type characteristics and the particle size play significant role in affecting the reflectance characteristics of a suspension. The above noted dependence of the reflectance on the particle size shows the need to account for the effect of particle size variation while developing a model to predict the suspended » (vii) sediment concentration, the turbidity level and modified secci depth from remotely sensed data. THE INTEGRATED EFFECT OF SEDIMENT CONCENTRATION AND PARTICLE SIZE VARIATION ON REFLECTANCE: The spectral response relationship with concentration and particle sizes was observed in the wavelength range of 500-1000 nm. The wavelength range of 700-900 nm is an ideal range for prediction of suspended sediments concentration, turbidity level and modified secci depth as this wavelength range shows a more pronounced and distinct differences in reflectance values corresponding to the different sediment concentrations. Related models have been developed for the evolved functinal relationship and presented in Eq. 3a-c to respectively predict the suspended sediment concentration, turbidity level and the modified secci depth (Bhargava and Mariam, 1990c). Css = (a9 + al0 D + all R)3 <3a) lnT = ai2 + a13 D + a14 R Ob) InSD = a15 + a16 D + a1? R (3c) The function coefficients ng to &n (Eq.3a), a12 to a.. (Eq .3b) and alg to a1? (Eq .3c) were determined through mutliple linear regression in 700-900 nm range (at 50 nm interval). The predicted values using Eq . 3a-c are in good agreement with the observed values and this manifests the robustness of the models. To use Eq. 3 the function coefficient values must' be available at all the wavelengths and this restricts the flexibility in working conditions. To avoid this limitations of the knowledge of model coefficients at different wavelengths, (viii) wavelength value in nm was also incorporated in the models and the function coefficients were obtained Css (a18 + a19 D + a20 R + a21 X)3 ---(4) The function coefficients of Eq.4 were obtained with D,R,X and Css exPressed in mm, percentage, nm and mg/1 respectively. The function coefficients in Eq .4 differ only because of the soil type but not the wavelength and this equation has an additional advantage of avoiding the knowledge of model coefficients at the various wavelengths (Bhargava and Mariam, 1990d). INTEGRATED EFFECTS OF THE SEDIMENT CONCENTRATION, SIZE AND PROPERTIES ON REFLECTANCE: Environmental properties change from place to place and time to time. Reflectance is influenced by some properties of the sediment present in suspension. From the observed data pertaining to the same sediment concentration, it is seen, on qualitative comparision with the physical and chemical properties of the various soils, that the soil's organic matter (manifasted by the parameter volatile matter which alters the solids surface features)(Bhargava and Mariam,19901) bears a rough correlation with the reflectance values of the various soils in the 550 to 750 nm wavelength range (higher reflectance at lower content of the volatile matter, and vice-versa, showing that the volatile matter contributes little to the reflectance in the stated wavelength range). The reflectance value for the wavelengths between 550 and 750 nm black cotton soil with highest organic content shows the lowest reflectance values whereas grey soil with lower organic content shows the highest reflectance value. Thus, the organic (ix) content of the soil present in the water body can be predicted, and this property of the soil may help to identify the soil type present in the various zones of water stream. The soils pH bears on approximate relationship with the reflectance values of the various soils in the wavelength rane of 800 to 900 nm (higher reflectance at higher PH); and the soils specific gravity bears a rough relationship with the reflectance values of the various ooilo in tho 800 to 1000 nm wavolongth region [higher reflectance for lwoer specific gravity and vice-versa because the denser particles provide lesser volume (or surface area) and, thus, lower reflectance values]. Black cotton soil exhibits the lowest reflectance value for all the wavelength ranges that were used in this study. In the blue-green and near infrared (IR) region of the spectrum, Kaoline shows the highest reflectance value whereas in the red region, grey soil has highest reflectance (Bhargava and Mariam, 1990a). Hence, some of the major interfering environmental properties and sediment characteristics need to be identfied and quantified and then included in the predictive models. Organic matter has a tendency to decrease the reflectance and Iron oxide, selectively reflects red light and absorb green light. To incorporate these properties trial correlations were attempted between the function coefficients al5 and al7 with iron oxide and organic matter respectively and Equ.4 is rewritten as shown in Eq.5. Jo 0m 3 [a14'0.0431 IQ-0 m °ss |a14'0.0431 1-0.0588" D+a16 R+91.62920 -61.4804 X m (5) (x) Where Om and lQ are organic matter (percentage) and iron oxide (mg/g) respectively. As shown in Eq .5 for a given type of suspedned sediments, the value of suspended sediment concentrations (Cgs) can be determined from Eq .5 by substituting the reflectance value (R) at a chosen wavelength (X), the appropriate values of the function coefficients, organic matter (0m), iron oxide (IQ), and geometric mean diameter (D) of the suspended sediments. There exists a good agreement between the predicted (using Eq. 5) and observed values of the suspended sediment concentration. From the above discussions we have obtained the following functinal relationships for the determinatin of suspended sediment concentrations. R = f <C08> da) R = f (C33.D) (3a) R = f (Css,D,\) (4a) i R = f (C8a,D,X ,Io,0m) (5a) Models have been evolved based on the above functinal relationships for the predict in of sediment concentration, turbidity level and modified secci depth from measured reflectance values with due involvement of the characteristics of some of the environmental and sediment properties. Depending on the availability of the data and the required accuracy, any one of the equations derived from the above functional. relationships can be used to determine suspended sediment concentration (C ) S3 Equation 5 is obviously the most accurate of the above equations because it involves the largest number of parameters. (xi) EFFECT OF LIGHT ZENITH ANGLE VARIATION ON REFLECTANCE The spectral response of the suspended materials was measured in terms of percentage reflectance. The effect of light Zenith angle (©z) variation on reflectance measured from sediment laden water was also studied at different wavelengths. Reflectance values measured from the same sediment concentration but different ©z values do not coincide and this establishes that ©z values play a role in affecting the spectral characteristics of a suspension. Based on the observed data at different e values, the reflectance increased with an increase in the e value for a Lt given sediment concentration. At lower e values, the difference in the observed reflectance values for any two sediment concentrations, is reduced. At higher 6>z value, such a difference is wider, and thus more helpful in differentiating the minor sediment concentration differences. The change in reflectance value for the same sediment concentration but different soil types is also more pronounced at higher e^ values (Bhargava and Mariam, 1990g). The above noted dependence of reflectance on the d value shows the need to account for the effect of e variation while developing a predictive models. Reflectance (R) as a function of ©z [R = f(^z)3 has been modelled to quantify the effect of changes in ez variation on reflectance measurements and is presented in a general form in Eq.6. R = a22+ a230Z (6) The function coefficients values for a22 and a23 at any given sediment concentration and soil type were computed through linear regression of the observed data at 700 nm. (xii) EFFECT OF SALINITY ON REFLECTANCE The spectral response was measured from saline waters (of concentrations ranging from 50 to 3200 mg/1) in terms of the percentage reflectance in the wavelength range of 500-900 nm. Relationships of percentage reflectance with wavelength based, on the observed data of the various salinity concentrations was studied. The spectral response increases with a decrease in salinity in the wavelength range of 500-900 nm, and a more pronounced variation of reflectance with respect to salinity is observed to be in the 600-800 nm range. This type relationship has been modelled and presented in a general form in Eq.8. ln S = a24+ a25 R (8) In Eq.8, S represents the salinity (mg/1) and R the reflectance. The function coefficients were determined through a linear regression analysis of the observed data. In nature waters of varying combinations of salinity and sediment concentrations are encountered. The interference effect of the presence of sediment in a saline water was also studied. Spectral response increases with suspended sediment concentration and decreases with an increase in salinity. In a sediment laden saline water, salinity and sediment concentrations cummulatively cause a change in the spectral response (Bhargava and Mariam,1990h). Reflectance is not only a function of the suspended sediment concentration level but is also a function of the Properties of the sediments present and some environmental influences. To develop a model which can be used in different geographic areas and various seasons and flow conditions, these (xiii) properties need to be considered. Based on extensive laboratory experiments an attempt has been made to identify and quantify the interference effects of some of the environmental parameters and sediment properties. Models have been evolved for the prediction of suspended sediment concentration, turbidity level, modified secci depth and salinity with due involvement of the interfering environmental and sediment properties like sediment sizes, mineral composition of the sediments present in suspension and B Lt variation.