Aquatic Ecology 36: 153–163, 2002. © 2002 Kluwer Academic Publishers. Printed in the Netherlands.
153
Application of satellite data for investigation of dynamic processes in inland water bodies: Lake Shira (Khakasia, Siberia), a case study Anatoly P. Shevyrnogov1, Alexei V. Kartushinsky1 and Galina S. Vysotskaya2 1 Institute
of Biophysics of SB RAS, Akademgorodok, Krasnoyarsk, 660036, Russia (E-mail:
[email protected]); 2 Institute of Computational Modelling (Russian Academy of Sciences, Siberian Branch) Akademgorodok, Krasnoyarsk, 660036, Russia (E-mail:
[email protected])
Accepted 21 September 2001
Key words: modelling, phytopigments, satellite data, satellite equipment, software, temperature
Abstract This work describes avenues to use satellite information to analyse dynamic processes in aquatic ecosystems. Information for this analysis, was retrieved from AVHRR satellite sensor data. This information consisteds of time series of images of radiation temperature and turbidity. We expect this information will be of great value in analysing inland water bodies. Methods to process satellite information using original software and data processing techniques are proposed. For the investigation of the process and analyses of satellite information Shira Lake (Khakasia, Siberia) was used as a case study. To study the variability of the surface temperature and turbidity of the Lake in summer, the satellite and ground-truth data of the lake was applied. This study represents the first evaluation of the dynamic processes for Lake Shira based on satellite, ground-truth and modelling data. We developed algorithms and software to process satellite images to enable the reconstruction of time dependence of temperature and spectral reflectance of water bodies in the visible range, and to make computer-animated films visualising the spatial and temporal dynamics of the study parameters. The analyses of morphometric, meteorological and hydrological characteristics of Lake Shira have provided a realistic opportunity for processing the satellite information and to develop numerical models of variability of the hydrological regime of the lake. The results obtained demonstrate the feasibility of systematically retrieving the spatial information from the satellite data on the dynamics of the surface water temperature and of the suspended matter in the lake.
Introduction Using satellite data of aquatic ecosystems enables one to gain a basic knowledge of the seasonal and annual dynamics and spatial distribution of phyto-pigments and of the seasonal variability of primary production processes. To gain a thorough insight into the basic processes requires data on the dynamics of spatial inhomogeneity of the parameters both vertically and horizontally (Mamayev, 1995; Ozmidov, 1998). The conventional biological methods using single sampling techniques are, however, unable to reveal the dynamic activity of the processes or to adequately analyse the phenomena occurring in the lake (Gitelson et al., 1988; Piggott, 1999). Remote sensing methods, on the other hand, allow monitoring of the
dynamic processes in aquatic ecological systems and the evaluation of changes in the indicators of the aquatic environment, such as external impacts that may change or endanger the environment in a short period of time and over a very large area. The retrieval of remote space-borne data makes rapid monitoring of the water bodies possible. The high rapidity of the technique, however, imposes restrictions on analysing the additional information used to interpret the remote sensing measurements (Wittich, 1997). Notwithstanding this limitation, the spectral optical characteristics enable an initial analysis of space images to reveal the basic processes by indicating the variability of the parameters of the aquatic ecosystem under study. Currently, the methods for measuring the backscattered light of water bodies from space give the best
154 information. The temperature of the water and the concentrations of substances in the water, both the dissolved and particulate, mainly determine the value of backscattered light in the infrared and visible ranges. The best sources of information for the evaluation of dynamics and biological regimes of water bodies are the radiation temperature and the backscattered light in the visible range (Pozdnyakov & Kondratyev, 1997). Satellite monitoring of lake ecosystems allows the retrieval and analysis of information about dynamic processes due to variations in the radiation temperature and in the backscattered light. Time series of images provide information about the seasonal and annual dynamics of the hydrological and the biological parameters of water bodies (Shevyrnogov et al., 1996; Shevyrnogov & Sid’ko, 1998; Shevyrnogov & Vysotskaya, 1998). AVHRR and MSS-6 are productive orbiting space instruments that carry out scientific programmes. The data thus retrieved ensure rapid recording of parameters such as concentration of materials in suspension and radiation temperature. The scientific program SeaWiFS is promising for receiving new satellite data, which can be used to monitor the spatial distribution of chlorophyll. Further improvements in aerospace-born techniques will not only facilitate the investigation of the variability of lake ecosystems, but will also help us to gain a deeper insight into the relative importance of the parameters named. The main goal of the study was to demonstrate the dynamics of the biological and hydrological characteristics of inland water bodies using Lake Shira (Khakasia, East Siberia, Russia) as a case study, employing remote sensing methods as well as numerical modelling. Lake Shira is a complex saline water system with pronounced seasonal hydrological and biological dynamic processes. A rapid multiparametric monitoring of the processes in the lake and development of an information model will ease the investigations of the dynamic processes that are now in progress (see Degermendzhy et al., 2002). To study the horizontal dynamic structure of the lake we used the hardware system to produce, process and analyse the satellite-sensed data. The application of remote sensing methods to investigate the dynamic processes, employed satellite data from Lake Shira, which is unique both in view of its characteristic physicochemical conditions (high salinity, turbidity, variability of the water thermal structure and high hydrogen sulphide concentrations) and biology (simplified food-chain, great variability of phytoplankton
concentrations with depth). The importance of turbidity due to the concentration of suspended matter and the annual temperature cycle are the least studied aspects in Lake Shira. The two main aims of the present study were: first, to work out a basic approach to analyse satellite information, using Lake Shira as a case study; and second, to study the variability of the surface temperature and turbidity of the Lake in summer on the basis of joint application of satellite and ground-truth data. These data were used in algorithms of a numerical model of information-prognostic complex for satellite information processing. Two lakes located near Lake Shira, namely, Lake Belio and Lake Itkul, were used as the test areas for comparison with Lake Shira, based on the surface temperature satellite data and the meteorological ground-truth data. The comparisons of the surface temperature dynamics show a similarity both in the pattern of changes and in the amplitude of the temperature due to the close geographic position and similar meteorological conditions of these three lakes.
Materials and methods NOAA satellite information receiving station The AVHRR data received from NOAA satellites were used to monitor dynamics of hydrological and biological processes in Lake Shira and to monitor the hydrothermal status of the land surface around the lake. The HRPT station is equipped with an up-todate computer system enabling the building of the first part of a geo-information system to monitor inland water bodies. After HRPT was recorded, the system automatically executed the programme to produce a calibrated file corresponding to the on-line file. The calibration created an average random basis and produced 1024 temperature levels for 10-bit data from the infrared sensors and albedo from the visible spectrum sensors. The calibrated file was used by the image processing programmes to analyse the data. A navigation updating function was required to interactively correct the errors in recording the AVHRR data. The World Database (WDB) was imposed on the images and compared to the actual land surface reference points visible in the image. Navigation/calibration allowed us to superimpose the world database (WDB) on the longitude/latitude lines image and to calibrate the temperatures and the longitudinal/latitudinal positioning of the cursor. It
155 also indicated the distance between the two points chosen by the cursor in the image. The image editing function prevented images from the errors and noise. The HRPT data with flaws were reconstructed by replacing the distorted image line with an undistorted one. Composite image function helped us to temporarily combine the different areas into a single composite image in a cartographic projection. Therefore, the vegetation index covered 100% of the time sequence of one image during a week (or any time window). The observations can cover up to one-year periods. The geographic database function showed the geographic database in every projection and in every region. Classification method The proposed automatic classification algorithm was realised within the framework of the image recognition problem by the iteration procedure of successive reconstructions of nonparametric estimates by interface equation between classes corresponding to onemode fragments of probability density. The number of classes was not assigned a priori. This additional stage (reconstructing the probability density) essentially reduced the labour intensity of classification. To define the class centres, the previously calculated probability density values did not allow us to classify the entire sampling, but only the parts that had a probability density value exceeding the assigned level. Another contribution of the developed algorithm’s efficiency is the integral nonparametric estimate of the probability density, which, compared with the classical Rosenblatt-Parzen procedure, facilitated a better approximation, owing to the smoothing operator. The proposed classification system consisted of subsystems performing the following functions: – preliminary data processing; – calculation of nonparametric estimate of probability density; – distinguishing compact groups of points (classes) corresponding to single-mode fragments of the probability density; – statistical analyses of the classes. The probability density calculated by the integral estimate combined good quality of reconstruction and was relatively less labour intensive. The next stage of realising the nonparametric automatic classification algorithm was to distinguish the groups of points corresponding to single-mode fragments of the probability density.
Image conversion program The images were retrieved in a MAC format from the HRPT station. To compare the images and to facilitate their processing by conventional programmes, we have developed a conversion of the images of this format to the IMG format of the IDRISI package. This programme geometrically updated the image to the rectangular coordinate system, which is convenient for bodies such as Lake Shira and allows comparing the ground-truth and the space-borne data. The programme has been developed for Microsoft Windows 95, 98, and 2000. It enables one to: – choose every combination of input files; – navigate in the Windows file system; – facilitate the orientation in a large array of similar image files; – produce rapid certification information about a space image file. A programme was developed to choose images from the time series made to view and reveal dynamics of the received satellite information. A multiplayer image was made from the images chosen. With the cursor moving over the image the screen displayed parameter values at a current point and its coordinates. Additionally, the programme enabled us to zoom in the image and to study dynamics of the chosen parameter for the entire time series taken. Remote sensing methods used for investigation of turbidity in inland water bodies High turbidity waters are waters with large amounts of suspended particles. Both the high absorption and high scattering reduce visibility in natural waters. Strong scattering by turbid water reflects much light, while waters with a low albedo – of black lake water type – are very dark. The scattering particles that determine the turbidity of water may differ in nature: phytoplankton, suspended inorganic material, detritus, etc. Such water turbidity can be adequately evaluated from a satellite by measuring the reflection in the visible spectrum. AVHRR employs the first channel with the wavelengths from 580 to 680 ηm. The results derived by processing the data produced in the different time periods and in the different geographic coordinates required a correction for atmospheric factors. It was also necessary to take the Rayleigh scattering and atmospheric spray scattering into account. The degree of Rayleigh scattering was obtained by using the satellite tracking angle and the sun’s zenith angle
156 to subtract the produced value from the backscattered light recorded by the first channel of the AVHRR satellite equipment. The aerosol component was measured by the second AVHRR channel in the near infrared spectrum. The Rayleigh scattering calculated in the second channel was subtracted from the Rayleigh scattering calculated in the first channel. The emission corrected by the Rayleigh scattering in the second AVHRR channel was assumed to be determined by the aerosol component, because in the near infrared spectrum the radiation is not expected to have an underwater component due to light absorption in this spectral range (maximum water transparency is about 483 ηm). As channels 1 and 2 are relatively close to one another in the spectrum of electromagnetic radiation, the aerosol component in these ranges was assumed to be equal. If needed, an additional correction was carried out to enhance the accuracy, taking into account the specific features of the spectral radiation in the first channel of AVHRR. Numerical modelling based on ground-truth data To study the vertical pattern of Lake Shira we used the ground-truth data of summer 1999 on the distribution of temperature, concentrations of phytoplankton, zooplankton and of the dissolved nutrients (N, P). These parameters were used for the numerical modelling. Weather data on air temperature, intensity of solar radiation and wind speed were also used. Meteorological ground-truth data were obtained from the meteorological station. Our model is adapted to Lake Shira to calculate vertical profiles of water temperature and concentrations of nutrients and phytoplankton (biomass values, mg cm−3 ). A mathematical model of an aquatic ecosystem was developed to solve differential equations by the ‘sweep method’ using finite-difference approximation made by an implicit numerical scheme (Samarski, 1977). The layout of the model is given in the section on results. The satellite data on surface temperature are relatively more exact than the data on turbidity. Therefore, at this stage we use in our model the satellite data on the temperature and those measured in the water. Because the evaluated data on the distribution of turbidity are not quite exact, as mentioned, we use them as the qualitative results that can potentially be used in the model. It is reported, for example, that in the Baltic Sea in spring, the surface distribution of phy-
toplankton can be roughly estimated by the turbidity distribution (Semovski, 1999). Within some limited time, the turbidity distribution can roughly match the information on phytoplankton distribution in the surface layer. The major factors at the water surface that influence the temperature and the concentration of different compounds are the wind and solar radiation. We found this true for the water surface of Lake Shira and for the other lakes located nearby (Lake Belio and Lake Itkul).
Results A software to study the dynamics of water bodies’ parameters by space images Software has been developed to analyse the dynamics of the surface water temperature and turbidity using satellite images in the visible and infrared ranges. The programme was written in a Windows format and enabled one to distinguish bodies of water; to smooth the data; visualise the temperature dynamics in a chosen point in space; to show differences in radiation over time at any given place; make an animated computer film about the variability of a chosen characteristic in space and time; and to improve the images with smoothing functions. Examples of the graphic display of information are given in Figures 1 and 2. In our case study, one of the most important stages was the information establishing relationships between data from the satellite, ground-truth, and modelling. The software developed initially enabled us to analyse satellite data of Lake Shira using the distribution of the radiation temperature and turbidity retrieved in the spring to summer periods from 1998 to 2000. The variability of the hydrological conditions was studied by satellite and measured data. The satellite data were processed to derive changes in the water temperature and turbidity in the surface layer of Lake Shira. During 1999–2000, maps of the spatial distribution of radiation temperature of Lake Shira, Lake Belio and Lake Itkul were made. In addition, using the software developed it became possible to load interpolated maps of temperature with a coordinate grid. For example, Figures 3a–d show the distribution of radiation temperature during the summer season of 1999. The satellite image obtained on 23 July (Figure 3c) showed a sharp increase in temperature between 30 May and 7 July 1999 in the central and shore areas
157
Figure 1. Radiation temperature variabillity.
of the lake. The temperature dynamics in the summer period was pronounced and showed different temperature zones. The remote sensing data of the temperature were used for developing the mathematical models. We developed the input data to realise a numerical model of the vertical distribution of temperature (T ), concentration of phytoplankton (F ) and the nutrients (N) (Equations (1–6)). A suitable algorithm has been developed to calculate the hydrological and biological parameters of an aquatic ecosystem (Kartushinsky, 1997): ∂w T 1 ∂Q ∂T ∂T +w =− + , ∂t ∂z ∂z Cp · ρ ∂z
(1)
∂N ∂w N ∂N +w =− − R, ∂t ∂z ∂z
(2)
∂w F ∂F ∂F + (w − v) =− + M, ∂t ∂z ∂z
(3)
∂T , ∂z
(4)
−w N = (Kz + αz )
∂N , ∂z
(5)
−w F = (Kz + αz )
∂F , ∂z
(6)
−w T = (KT + αT )
where R is the function of the nutrient associated with biochemical processes, M is the function of
the biomass associated with phytoplankton functioning. Hydrodynamic parameters: w, ν, vertical speed; Kz, αz, diffusion coefficients. The effects of certain abiotic factors (thermodynamic processes, turbulent mixing, vertical currents; and distribution of nutrients and light) on the vital activity and vertical distribution of phytoplankton over several days are shown by our model (Figure 4). The description of the behaviour of objects is determined by various conditions, but at this stage of our investigation we consider the wind speed and solar radiation as the main determining factors. The functional characteristics of phytoplankton used in the model are presumably closely connected with the turbidity in the surface layer of the water body, but this investigation has not yielded a good result for Lake Shira. Analytical dependencies we derived to evaluate the spatial variability of biological inhomogeneities caused by the hydro physical processes, on the basis of the numerical experiments and field measurement data for Lake Baikal (Kartushinsky, 1997), and employing the similarity theory (Equations (7) and (8)). Here the scale of inhomogeneity of phytoplankton (Lp ) is associated with the specific scale of water mass movement (V ) (Equation (7)) and the coefficient of turbulent exchange by momentum (Km ) (Equation (8)) in time t: V · t · Pmax , (7) Lp = µ0
158
Figure 2. Time dynamics of radiation temperature and turbidity in an isolated point Lake Shira.
Figure 3. Spatial distribution of temperature on Lake Shira in 1999 (black – maximum temperatures, white – minimum temperatures).
159
Figure 4. Mathematical model layout.
√ Km · t · (Pmax )3/2 Lp = , µ0
(8)
where Pmax is the maximum growth rate for conditional phytoplankton groups, µ0 is the extinction rate. Table 1 shows the magnitude of vertical inhomogeneity for the distribution of diatoms and blue-green algae under the effect of vertical movement of water with specific scales of transfer rate according to Equation (7). If the turbulent diffusion is considered as the basic mechanism governing the phytoplankton inhomogeneities, Equation (8) must be used. We think that these equations can be used for Lake Shira and other aquatic ecosystems, but the coefficient of turbulent exchange as well as the vertical and horizontal water velocities must be exactly determined. Space images from the spring and summer periods of 1998, 1999 and 2000 were analysed by the software developed. The missing satellite data were reconstructed by linear interpolation and Bezier spline curve. The analysis yielded data on the turbidity patterns in Lake Shira. Based on the satellite measurements in recent years, clear cloudless days in Khakasia in the spring period are relatively few. Nevertheless, appropriate processing of the satellite information provided opportunities to analyze the dynamics of suspended materials in the lake’s surface water layer and confirm its relationship with the hydrodynamic and biological models. The output information can be pre-
sented both as maps of turbidity (Figure 5a–c) and as time profiles for specific points of the lake (Figure 2). These pictures show both temporal and spatial differences in turbidity and temperature in the lake. However, we did not obtain statistically significant estimates of the correlationships between the temperature and turbidity for the different points of the lake. The rapidity of changes in these parameters depends on the meteorological factors: wind velocity, air temperature, and solar radiation. The maximal changes in the water temperature occurred in July, as in this period the differences between the air temperature and the water temperature are minimal (about 1–2 ◦ C), and the wind velocity varies between ‘calm’ and 6–8 m s−1 (Figure 6). Model experiments also confirm these empirical observations, that the major parameters governing the changes in the turbidity are a result of the wind conditions and the solar radiation prevailing over the lake surface.
Discussion The remote sensing data about any water body must be decoded and interpreted employing a priori information about the water body under study (Goetz et al., 1999). Information on several parameters is needed: for example, on the hydrological features of the lake, on the bottom topography, seasonal phenomena (temperature cycle, ice formation, period of circulation
160 Table 1. Vertical variabillity of biological inhomogeneity of phytoplankton based on numerical modelling Lp , m
V , m s−1
T,h
Phytoplankton
Lp , m
Km , m s−1
t, h
Phytoplankton
4.3 8.6 2.4 4.9 43 86 24 49
10−5
12 24 12 24 12 24 12 24
Diatoms
43.2 86.4 18.6 37.2 13.6 27.3 5.9 11.8
10−3
12 24 12 24 12 24 12 24
Diatoms
10−5 10−4 10−4
Blue-greens Diatoms Blue-greens
10−3 10−4 10−4
Blue-greens Diatoms Blue-greens
Figure 5. Spatial distribution of turbidity in Lake Shira 1999. (black – minimum turbidity, white – maximum turbidity).
of the lake) and on the eutrophication level. Knowledge and evidence of the human impacts on the lake and optical characteristics of water masses, etc., are indispensable. Data on meteorological conditions during the survey period and on the days preceding the remote sensing measurements are important prerequisites. Biological measurements needed for ground truth verification will include information on an array of parameters: the concentrations of materials in suspension (seston) and chlorophyll or phytopigments (as measures of phytoplankton concentration), on the concentrations of organic and mineral suspended matter, relative transparency of the water, etc. Although we used these data in our research, they are, so far, not quite satisfactory for good statistical estimates and for calibration with the satellite data. The key parameters for investigation by spaceborne methods of the dynamic processes in Lake Shira are the radiation temperature and spectral reflectance in the visible range. We used these parameters to study the variability of the thermodynamic structure of the water, the dynamics of biological processes, and the concentrations of organic and mineral suspended elements.
Dynamics of water temperature Water temperature distribution appears to be determined not only by the climate peculiarities of the region, but also by temperature zones, even for small water bodies as Lake Shira. Such zones appear to be formed by wind-induced surge, in particular, which can change the structure of turbulent movements and advective currents. The information on changes in radiation temperature has enabled us to show the thermodynamic activity zones associated with the hydrological status of the water body using thee satellite data. One of the important characteristics of such zones is their spatial scale and time. Anomalous changes in the surface water temperature reflecting the effect of physical factors (primarily water movement and turbulent mixing) on the ecosystem are important for the investigation (Xu et al., 1999; Scardi & Harding, 1999). To employ the satellite information to reconstruct the vertical structure of a water body status, we had to find the key factors responsible for the changes in the general regime of the aquatic ecosystem. For this purpose we used the numerical model as described in the Results section. The initial data for calculations in the model are the surface temperature and turbidity based on satellite data.
161
Figure 6. Variations in parameters of Lake Shira for the period 15.07.99–23.07.99. (a) comparison of satellite (Tn) and model data (Tmod) on surface temperature; (b) comparison of satellite data on surface temperature (Tn), air temperature (Ta), and wind velocity (W).
Dynamics of turbidity fields Serious difficulties in evaluating the turbidity were encontered. This is because armant of the backscattered light from water bodies received by the satellite is very small. The probability for a photon to be reflected and leave the surface water decreases exponentially with increase of the path in the water, because the water absorbs it. The absorbing and scattering characteristics of water masses define the vertical attenuation of light and limits recording of the light signal from a satellite. It is assumed that the most useful information obtained comes from the water layer that has thickness equal to the depth of a standard Secchi disc transparency plus 30% of this depth (in metres). This depth depends on the composition and quantity of the suspended particulate materials in the water. For the first channel the penetration depth is generally from 1 to 10 m, depending on the composition of suspended and dissolved compounds. The turbidity currents found at 10 m depth can hardly be recorded from a satellite. However, we used the measured data on the concentration of phytoplankton and nutrients in Lake Shira for these conditions (Gaevsky et al., 2002; Kalacheva et al., 2002). The analysis of satellite information in the case of Lake Shira mainly concerns the optical inhomogeneities of the surface layer of water determined by variations in the content of suspended matter of organic and mineral origin (Karlin et al., 1992). The processing and interpretation of the satellite data within the optical range facilitates solving several problems. These include the location of hydrological
fronts; the human impact zones, including their spatial and temporal variability; the turbidity fields and evaluation of the intensity of the phytoplankton development and its seasonal and annual variability. To obtain all such information we need to further improve the software and collect more satellite data for Lake Shira and other lakes in the vicinity. Our study will make use of high and medium resolution satellite images of Lake Shira to meet the following requirements: – the execution fragment of the image has no obvious radiometric noise; – cloudiness is absent or clouds cover insignificant parts of the water area; – the image is made beyond the flash area; – the wave height with foaming is not strong (generally wind velocity is less than 7 m s−1 ). Inhomogeneous biological zones differ in their origin. Spatial distribution of phytoplankton is primarily dependent on water dynamics and the geographical latitude of a locality. On the basis of the spatial and temporal variability of chlorophyll concentration, particularly its seasonal changes, temperature gradients and phytopigment dynamics, we may divide the Lake Shira into two main zones: coastal and central. In the central zone, at a depth of 21–22 m the dynamic regime differs from that in the coastal zone. For this zone we have more biological and temperature field data. Therefore, we shall use in the ongoing study the central zone for the calibration of the Lake Shira model.
162 Numerical modelling based on satellite information and ground-truth data Analysis of the literature over the last 5–8 years shows that satellite remote sensing data are mostly used for comparative evaluation of model experiments and are seldom used as a basis for the calibration of models and prediction (Van der Wel et al., 1998; Jacobs et al., 2000). This is largely due to the permanent updating of the satellite data and calibration of the satellite data on the spot (Shih et al., 1999; Goetz, et al., 1999). There is no universal approach to evaluating the formation mechanisms of the biological structure and dynamic processes associated with physical processes in water. But nowadays the assimilation of satellite data in the mathematical models and the realisation of models for prognosis problems are at the starting stage of their development (Semovski, 1999; Semovski et al., 1999; Kartushinsky, 2000; Yang et al., 2000). Unfortunately, we lack the series of ground-truth data of Lake Shira for several consecutive hours that are essential for the comparison of the model data on phytoplankton concentration with the satellite data on turbidity. We have elaborated the structure of the numerical model and algorithms for Lake Shira and other lakes to decide prognostic tasks and to demonstrate different physical and biological regimes (Kartushinsky, 1997). Hydrophysical processes in the lake change the temperature regime (Figures 1–3), and turbidity changes also cause changes in the biological components (Figures 2 and 5). Determination of the regimes that change with time is necessary for a precise prediction of seasonal and daily variations of the ecosystem dynamics. These regimes will be investigated after collecting dditional data from space and from the field of Lake Shira, and after properly synchronizing these in time.
flectance to analyse the dynamics of hydrological characteristics of Lake Shira and other lakes. The algorithms and software allow us to make computer-animated films visualising the space and time dynamics of the parameters under study. – New, efficient, nonparametric statistical methods to analyse satellite images have been developed. – Morphometric, meteorological and hydrological characteristics of Lake Shira have been analysed, providing a realistic opportunity to apply the data in prognostic complex for satellite information processing and to develop numerical models of variability of the hydrological regime of the lake. – The basis to produce quantitative indices required to evaluate advective and diffusive effects on space and time variability of phytoplankton inhomogeneity is formed by numerical modelling and processing of satellite and ground-truth data. The results obtained demonstrate the feasibility of systematically retrieving spatial information from satellite data regarding the dynamics of surface temperature and suspension. This information can be further employed in hydrodynamic and ecological models, to both supply them with data and verify their prognostic properties.
Acknowledgements The research described in this publication was supported by grants REC-002 (Ministry of Education RF, US-CRDF), INTAS-97-OPEN-519 and RFBR No. 99-05-64338. The comments by the reviewers were a great help in improving earlier versions of this manuscripts and the linguistics.
References Conclusions This study represents the first evaluation of the dynamic processes for Lake Shira, based on satellite data, ground-truth data and modelling data. The main achievements are: – the development of algorithms and software to process satellite images and to enable one to reconstruct time dependence of temperature and spectral reflectance of water bodies in the visible range. Using these algorithms and software we can now plot dependencies of temperature and spectral re-
Degermendzhy AG, Belolipetsky VM, Zotina TA and Gulati RD (2002) Formation of the vertical heterogeneity in the Shira Lake ecosystem: the biological mechanisms and mathematical model. Aquat Ecol 36: 271–297 Gaevsky NA, Zotina TA and Gorbaneva TB (2002) Vertical structure and photosynthetic activity of Shira Lake phytoplankton. Aquat Ecol 36: 165-178 Gitelson JI, Shevyrnogov AP, Molvinskikh SL and Chepilov VV (1988) Determination of Photosynthetic Pigments in Aqueous Ecosystems. Mitt. Geol. -Paleont. Inst. Univ. Hamburg, SCOPE/UNEP Sonderband, Hamburg, Heft 66: 331–340 Goetz SJ, Prince SD, Goward SG, Thawley MM and Small J (1999) Satellite remote sensing of primary production: an improved production efficiency modelling approach. Ecol Modelling 122: 239–255
163 Jacobs JM, Coulter RL and Brutsaert W (2000) Surface heat flux estimation with wind-profiler/RASS and radiosonde observations Adv Water Res 23: 339–348. Kalacheva GS, Gubanov VG, Gribovskaya IV, Gladchenko IA, Zinenko GK and Savitsky SV (2002) Chemical analysis of Lake Shira water (1997-2000) Aquat Ecol 36: 123–141 Karlin LN, Razumov EV and Shevyrnogov AP (1992) Local Unconformities in Satellite Images of Water Surfaces // Interactions of Biogeochemical Cycles in Aqueous Systems, Mitt. Geol.- Paleont. Inst., Univ. Gamburg, SCOPE/UNEP Sonderband, part 7, Hamburg, Heft 72: 281–289 Kartushinsky AV (1997) Effect of Daily Variation in Abiotic Conditions on Distribution and Production of Phytoplankton in Lake Baikal (in Russian; English Abstract). Water Res 24: 66–73 Kartushinsky AV (2000) Time-space structure and variability of surface temperature frontal zones in the ocean (based on AVHRR satellite data). Adv Space Res 25: 1107–1110 Mamayev OI (1995) About space-time scales of oceanic and atmospheric processes. Oceanology 35, 6: 805–808 (in Russian; English Abstract) Ozmidov RV (1998) Patches of phytoplankton in the ocean during various regimes of oceanic turbulence. Oceanology 38: 7–15 (in Russian; English Abstract) Piggott AR (1999) Publicly accessible software for groundwater modelling and analysis. Comput Geosci 25: 341–346 Pozdnyakov DV and Kondratyev KYa (1997) Remote sensing monitoring of natural water in the visible spectrum. II. Avenues of attack on inverse problems. Space-Borne Investigation Earth 2: 3–24 Samarski AA (1977) The Theory of Finite-Difference Schemes, Nauka, Moscow, (in Russian) Scardi M, and Harding Jr LW (1999) Developing an empirical model of phytoplankton primary production: a neural network case study. Ecol Modelling 120: 213–223
Semovski SV (1999) The Baltic Sea and Lake Baikal underwater bio-optical fields simulation using ecodynamical model. Ecol Modelling 116: 149–163 Semovski SV, Dowell MD, Hapter R, Szczucka J, BeszczynskaMoller A and Darecki M (1999) The integration of remotely sensed, seatruth and modelled data in the investigation of mesoscale features in the Baltic coastal phytoplankton field. Int J Remote Sensing 20: 1265–1287 Shevyrnogov AP and Sid’ko AF (1998) Ground truth methods as a part of space mapping of inland water phytopigment dynamics. Adv Space Res 22: 705–708 Shevyrnogov AP and Vysotskaya GS (1998) Observed trends of chlorophyll concentration in the surface layer of the northern and central Atlantic (1979–1983). Adv Space Res 22: 701–704 Shevyrnogov AP, Vysotskaya GS and Gitelzon JI (1996) Quasistationary areas of chlorophyll concentration in the world ocean as observed satellite data. Adv Space Res 18: 129–132 Shih T-Y, Hwang J-T and T-Je Tsai T-Je (1999) The fractal properties of sea surface topography derived from TOPEX/POSEIDON (1992-1996). Comput Geosci 25: 1051–1058 Van Der Wel FJM, Van Der Gaag LAC and Gorte BGH (1998) Visual exploration of uncertanly in remote sensing classification. Comp Geosci 24: 335–343 Wittich K-P (1997) Some simple relationships between land-surface emissivity, greenness and the plant cover fraction for use in satellite remote sensing. Int J Biometeorol 41: 58–64 Xu F-L, Jorgensen SE, Tao Shu and Li B-G (1999) Modelling the effects of ecological engineering on ecosystem health of a shallow eutrophic Chinese lake (Lake Chao) Ecol Modelling 117: 239–260 Yang MD, Sykes RM and Merry CJ (2000) Estimation of algal biological parameters using water quality modelling and SPOT satellite data. Ecol Modelling 125: 1–13.