Comparing suspended sediment concentrations derived from a model and collected in a tidally dominated area

Transcription

1 Comparing suspended sediment concentrations derived from a model and collected in a tidally dominated area Maryam Rahbani, Department of oceanic and atmospheric science University of Hormozgan, maryamrahbani@yahoo.com Paper Reference Number: Name of the Presenter: Maryam Rahbani Abstract Measuring suspended sediment concentration (SSC) in field is still a challenging procedure for oceanographers. Therefore the models which are capable to produce accurate hydrodynamics of a site are relatively inaccurate to simulate sediment dynamic. This investigation is a survey to verify the deficiencies incorporate both model results and field data in a tidal dominated area in regards with SSC data. A flow model from Delf3d in which it s hydrodynamic was validated for a channel located in the southeaster part of the North Sea, has been used to simulate the sediment dynamic of the field. The investigation was carried out over two cross sections in the channel. The profile of velocity and SSC derived from the model at each cross section were compared with those from the field. Unlike velocities, SSC derived from the model was not in good agreement with those from the field. The ratio of observed (field data) to predicted (model results) SSC along the depth were prepared to find out the depth at which the dissimilarity between the model results and the field data is at most. The dissimilarity in the shallow water area was more evidence. It was suggested to be due to the uncertainty in the measuring device in shallow waters. The ratio of predicted to observed SSC were also plotted against velocity derived from the model. The results showed that the highest dissimilarity was accrued during low velocities, which was suggested to be due to constant settling velocity consideration in the model simulation. Key words: tide, model, sediment, concentration, transmissometer,

2 1. Introduction This investigation is a survey on accuracy of the suspended sediment concentrations (SSC) collected using transmissometer, as well as simulated by a model developed using Delft3D package. The area under investigation is a tidal channel system located in the southeaster part of the North Sea. Even though the use of optical device is an appropriate way to estimate the SSC of a field, it associates with some deficiencies. On the other hand the models with high accuracy in the field of hydrodynamic, represent poor results in regard with sediment dynamics. As a support, it is the statement by Van Rijn (2001) who suggested the factor of two of the field SSC to be considered as high quality of model prediction. In this investigation on the basis of some available measured data a model was developed to simulate sediment dynamic of the area. SSC data were collected from the surface to the depth using transmissometer device. The model results were compared with the field data using different methodologies. The deviation between the model results and the filed data in each method are presented and presumable reasons are discussed. 2. Data and Material The area under this investigation is central Dithmarschen Bight. It is located in the southeastern part of the North Sea and is confined from the north by the Eider estuary and from the south by the River Elbe. The most dominant morphological features of the area are tidal flats, tidal channels and sand banks over the outer region. Under moderate conditions the maximum mean water depth in the tidal channels is about 18m, and approximately 50% of the domain falls dry at low tide. The Norderpiep channel in the northwest and the Süderpiep channel in the southwest are the two main branches that drive out from the North Sea into the Dithmarschen Bight. Crossing through tidal flats eastward, the two channels merge to form the Piep channel (Fig. 1). The three channels together form the Piep tidal channel system, which has the shape of a lying Y. The width of the channels and their rivulets varies spatially and temporally from a few meters to about 4 km. The water depths of the main channels vary from 5m to 25m. This channel system was specifically selected for the modelling investigations because of the availability of measured data. The source of the required field data for this study was those collected under Prediction of Medium Term Coastal Morphodynamics, known as the PROMORPH project. It was executed during the period May 1999 to June The data used in this study cover two cross sections in the Piep tidal channel system: T1 in the Süderpiep channel, and T2 in the Piep channel (Fig. 1). The width of the channel at cross section T1 is about 2040m and the water depth varies from 7.3 to 15.6m and, at cross section T2 the width is about 1200m and the depth is from 6.2 to 17.9m.

3 Fig. 1: Area under investigation Acoustic Doppler Current Profiler (ADCP) had been used to measure current velocities. The instrument was mounted at the bow of the vessel pointing downward. Measurements covered the water column from about 1.6m below the free surface, due to transducer draught and blanking distance, down to the seabed. The vessel moved forward and backward along each transect during a full tidal cycle collecting ADCP data along the route. Vertical profiles of the current velocity thus were collected for the whole period of the tidal cycle. Fig. 2 shows the procedure schematically. According to Jimenez-Gonzalez et. al, (2005) the accuracy of the ADCPs for measurements in the tidal channels of the central Dithmarschen Bight on the basis of the standard deviations for point measurements are approximately constant. They evaluated the averaged accuracy of the device with value of about 0.15 m/s. An optical beam transmissometer device has been employed during the cruises of PROMORPH project to collect SSC. For collecting the data at different levels along the depth, the transmissometer together with one CTD (Conductivity, Temperature, and Depth) device were mounted on a frame. In each cruise the frame was lowered at specified positions from the surface to near the bottom across each cross section (Fig. 2). The CTD device in the frame provided the height at which the beam scatter data are collected. Optical transmission data collected in this way were converted to SSC, using the equation proposed by Poerbandono and Mayerle (2005). (1) where, c is concentration of sediment, and A = L 1 ln(i) is the attenuation coefficient, in which L is the transmissometer path length in cm, and I is the optical transmission as a decimal fraction.

4 Fig. 2: Measuring technique along a cross section 3. Results and Analysis As a first analogy the variations of the current velocity and the SSC along the depth obtained from the model are compared with those collected from the field for all monitoring points. Data collection at various points relate to certain time. The model results, therefore had been extracted in such a way that their times and locations was matched with the times and the locations of the field data. The time difference between the field data and the model results for comparison never exceeded 5 minutes, and the spatial difference of the points in the field data and the model did not exceed 50 meters. This was found reasonable in view of the grid length being 90m. Typical profiles of the velocity and SSC for all monitoring points in cross-sections T1, and T2 are presented in Fig (3) for one ebb condition. The sets of data are those collected on 21 to 23 of March 2000, covering a sequence of spring tides with an average tidal range of about 4m. It can be seen that the current velocity profiles derived from the model are in good agreement with those from the field which also approves the results obtained by Jimenez-Gonzalez et. al, (2005). For the SSC profile however, some dissimilarity between the model results and field data can be observed. In cross section T1, the SSC profiles derived from the model are generally in good agreement with the field data in monitoring points 1, 2 and 4. Marked disagreement is evident between the model results and field data in profiles 3 and 5 to 9, especially from the near bed layer to the middle of the depth. In cross section T2 underprediction by the model is evident in all of the monitoring points except for profile 2. Comparisons between the SSC profiles from the model and from the field during a full tidal cycle revealed certain dissimilarities at shallow water regions.

5 Fig. 3: Current Velocity and Suspended Sediment Concentration Profile derived from the model and in situ measurements during the ebb condition for cross sections T1 and T2. It was found necessary to detect the relevance between the predicted (model results) and observed (field data) SSC with respect to the depth. Therefore, the ratios of the observed to the predicted SSC along the depth were calculated at each cross section. Fig. (4) and Fig. (5) show results for cross sections T1 and T2 respectively. At each cross section two monitoring points, one in shallow water region and the other in deep water region were considered. In each figure, the plots on the left show the ratios during a whole ebb phase and the ones on the right show the ratios for a flood phase. The monitoring point in shallow water region and its corresponding results are shown in blue and those for the deep water are presented in red.

6 Fig. 4: The ratio of observed to predicted SSC along the depth for monitoring points in deep water (red) and in shallow water (blue) at the cross section T1 during one ebb phase (left plots) and one flood phase (right plots). Fig. 5: The ratio of observed to predicted SSC along the depth for monitoring points in deep water (red) and in shallow water (blue) at the cross section T2 during one ebb phase (left plots) and one flood phase (right plots). At first glance it can be seen that predicted SSC values in shallow water regions are appreciably higher than observed ones. It can also be seen that the ratio during the ebb condition are much higher than that during flood phase especially in near bed layers. The dissimilarities observed specifically in the shallow regions can be related to the existence of some error in measuring devices. Variation in grain size distribution, existence of

7 biological matter, and generation of air bubbles in such a regions can be count as the reason for measuring error. As is stated Dithmarschen Bight is a tidally dominated area, thus currents in the area are essentially generated by the tide. Hence it is appropriate to test the correlation of SSC computed from the model and those from the field during a tidal range by means of current velocity. As the current velocity of the model has been qualified with an error of less than 0.15 m/s, this variable was found proper representative factor of the tidal range. To minimize the errors regarding hydrodynamic results only those current velocities from the model were considered in which their differences with those of field data were less than 0.15 m/s. For each selected current velocity the corresponding temporal and spatial values of the SSC were obtained from the model and the field data. The ratio of predicted to observed SSC was calculated for each current velocity. Using semi-logarithmic coordinates the scatter plots of these ratios versus current velocities are shown in Fig. (6) for cross section T1 and T2. Red lines in the figures represent the factor of two of the field data. Fig. 6: The ratio of predicted to observed suspended sediment concentration versus current velocity at cross section T1, and T2. Referring to the figure, a relatively similar trend in both plots can be observed for the variation of the ratios of SSC (predicted to observed SSC) with current velocities. That is, the lowest scatters can be seen during slack waters when the current velocity decreases to nearly zero, the ratios are approximately between 0.5 and 1. The scatter tends to increase with increasing current velocities during both ebb and flood conditions. The scatter of the SSC ratios reaches its maximum at the current velocity of about 0.5 m/s (0.1<ratio of predicted to observed SSC <4). The scatter again decreases with increasing current velocities, during both ebb and flood conditions. At maximum current velocities during both ebb and flood conditions the predicted SSC values are once again in quite good agreement with the observed

8 SSC (within the factor of two). The simplifications associated with the modelling, specifically consideration of constant settling velocity during a whole tidal range, can be resulted in the poor correlation between the predicted and observed SSC. 4. Conclusions As mentioned dissimilarities between the modelled and measured SSC in the shallow water area, can be due to measurement errors in these areas. Considerable variation that exists in the particle size distribution in shallow water areas could be a source of errors in the measurements. Gordon and Clark (1980), Bishop (1986), Moody et.al. (1987) and Bunt er.al. (1999) reported that the variation in particle size distribution is the most influential physical characteristic of the sediments on the response of optical devices. Bunt et.al. (1999) suggested that variations in floc size could double the variation in instrument response for similar mass concentrations. Existence of biological matter in shallow water area can also affect the recorded data by transmissometer. As pointed out by Walker (1981), biological matters such as chlorophyll-a and phytoplankton even though relatively insignificant by mass, their effect on the response of optical instruments is significant. These organisms are known to be active in the shallow water area where light is sufficient. The sticky nature of these particles causes flocculation between the fine particles. Ebb conditions are favourable for their activities because of the decrease in the water level and increase in the transmitted light. Discrepancy in transmissometer results could also be due to air bubbles originated by water organisms. Bunt et.al. (1999), and Campbell et.al. (2005) reported the significance of air bubbles to the response of the optical backscatter devices. They reported that air bubbles can double the response of the device. In addition to the errors resulted from the measuring device, some model related factors can be considered for discrepancies observed between the field and modeled SSC, specifically under low current velocity conditions. The use of a constant settling velocity for the whole area and for the whole tidal cycle can be counted as a model limitation. This is the limit associated with the Delft3D modeling and does not allow the use of variable values of settling velocities over the area. According to Winterwerp (2001) there are large variations in the value of the settling velocity having the higher values around the slack water due to mainly flocculation of sediment. His conclusion is that flocculation is a factor that explains why it is not possible to simulate the observed features in suspended sediment concentrations properly using constant settling velocity. On the basis of the suggestion by Winterwerp and van Kesteren (2004), the value of the settling velocity around slack water is controlled by several factors most importantly flocculation. Talke & Swart (2006) also emphasized the necessity of considering variation of the settling velocity during a tidal cycle in order to simulate the behavior of the suspended sediment. In their investigations they showed that biological matters and turbulence processes play important role in the variation of the settling velocity during a tidal cycle. Considering constant settling velocity for the tidal channel and the tidal flat can also affect the results. The model might not be able to properly simulate the amount of sediment washed out from the land and the tidal flat areas through the channel during the ebb conditions because of the insufficient supply of sediments. This is applicable specifically to the cross section T2 due to its proximity to tidal flats and the water-land interactions (see Fig.3). The SSC values obtained from the model during ebb condition show mostly underprediction for this cross section.

9 References Bishop, J.K.B. (1986). The correction and suspended particulate matter calibration of Sea- Tech transmissometer data, Deep-Sea Research, vol. 33, pp Bunt, J.A.C., Larcombe, P., and Jago, C.F. (1999). Quantifying the response of optical backscatter devices and transmissometers to variations in suspended particulate matter, Continental shelf research, vol. 19(9), pp Campbell, C.G., Laycak, D.T., Hoppes,W., Tran, N.T., Shi, F.G. (2005). High concentration suspended sediment measurements using a continuous fiber optic in-stream transmissometer, Journal of Hydrology, vol. 311, pp Gordon, H.R. and Clark, D.K. (1980). Initial Coastal Zone Color Imagery, Proceedings of the 14th Symposium on Remote Sensing of Environment, San Jose, Costa Rica, April Jiménez Gonzalez, S., Mayerle, R., and Egozcue, J.J. (2005). A Proposed Approach for Determination of the Accuracy of Acoustic Profilers in the Field, Die Küste, Heft 69, pp Moody, J. A., Blrrman, B., and Bothner, M.H. (1987). Near-bottom suspended matter concentration on the continental shelf during storms: estimates based on in situ observations of light transmission and a particle size dependent transmissometer calibration, Continental Shelf Research, vol. 7, pp Poerbandono and Mayerle, R. (2005). Effectiveness of Acoustic Profiling for Estimating the Concentration of Suspended Material, Die Küste, Heft 69, pp Talke, S.A., Swart, H.E. (2006). Hydrodynamics and morphology in the Ems / Dollard estuary: review of models, measurements, scientific literature and the effects of changing conditions, Technical Report, IMAU Reports, Issue: R06-01, University of Utrecht, 78 pp. Walker, T.A. (1981). Dependence of phytoplankton chlorophyll on bottom resuspension in Cleveland Bay, Northern Queensland, Australian Journal of Marine and Freshwater Research, vol. 32, pp Van Rijn, L.C. (2001). Simulation of nearshore hydrodynamics and morphodynamics on the time scale of storm and seasons using profiles and area models; a comparison of field data and model results; Coast3D experiments R3 Report Z2394, Delft Hydraulics, Delft, the Netherlands. Winterwerp, J.C. (2001). Stratification effects by cohesive and noncohesive sediment, Journal of Geophysical Research, vol. 106 (C10), pp Winterwerp, J.C., Van Kesteren, W.G.M. (2004). Introduction to the physics of cohesive sediments in the marine environment, In: Developments in Sedimentology, vol. 56. Elsevier, Amsterdam, 576 pp.