A methodology to attain daily variability of turbidity in the Dead Sea by means of remote sensing was developed. 250 m/pixel moderate resolution imaging spectroradiometer (MODIS) surface reflectance data were used to characterize the seasonal cycle of turbidity and plume spreading generated by flood events in the lake. Fifteen minutes interval images from meteosat second generation 1.6 km/pixel high-resolution visible (HRV) channel were used to monitor daily variations of turbidity. The HRV reflectance was normalized throughout the day to correct for the changing geometry and then calibrated against available MODIS surface reflectance. Finally, hourly averaged reflectance maps are presented for summer and winter. The results show that turbidity is concentrated along the silty shores of the lake and the southern embayments, with a gradual decrease of turbidity values from the shoreline toward the center of the lake. This pattern is most pronounced following the nighttime hours of intense winds. A few hours after winds calm the concentric turbidity pattern fades. In situ and remote sensing observations show a clear relation between wind intensity, wave amplitude and water turbidity. In summer and winter similar concentric turbidity patterns are observed but with a much narrower structure in winter. A simple Lagrangain trajectory model suggests that the combined effects of horizontal transport and vertical mixing of suspended particles leads to more effective mixing in winter. The dynamics of suspended matter contributions from winter desert floods are also presented in terms of hourly turbidity maps showing the spreading of the plumes and their decay.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The turbidity of the Dead Sea (Figure 1) recently became a major issue regarding the future of this unique hypersaline terminal lake. Two major projects that are now under consideration are being assessed in the context of their potential influence on the turbidity of this blue desert lake. The Red Sea Dead Sea Canal project aims to introduce seawater in order to reduce the rate of drop of the lake level (currently more than 1 m/yr). A by-product of the addition of seawater to the Dead Sea brine is the precipitation of gypsum, which may add significant amounts of suspended particulate matter (SPM) to the Dead Sea [Gavrieli et al., 2011]. Another large-scale project is the dumping of salt from the industrial ponds into the Dead Sea, which will add a significant amount of halite grains to the Dead Sea (16 × 106 m3/yr) [Lensky et al., 2010]. Characterizing the SPM distribution and its dynamics within the Dead Sea is required to robustly evaluate the potential impact of additional SPM as an outcome of these projects.
 Turbidity dynamics in water bodies depend on the supply of SPM and on transport related processes. Mapping turbidity in coastal waters is critical to many scientific and environmental studies [Siegel et al., 2009; GutiÉrrez-Mas et al., 2006] as well as for decision makers [Miller and McKee, 2004]. Turbid water affects the physical, chemical, and biological condition of lakes through their effect on radiative transfer and hence on temperature and stratification of the lake. The major sources recognized of turbidity are runoff, shoreline erosion, and resuspension of bottom sediments. Space-borne sensors are useful tools to map SPM [Bignami et al., 2007; Chen et al., 2007; Feng et al., 2012; Hu et al., 2010], providing synoptic mapping capabilities [Doxaran et al., 2002; Chen et al., 2004; Miller and McKee, 2004; Wang et al., 2004, 2007, 2009; Dogliotti et al., 2011]. To monitor the dynamics of turbidity within the daily cycle, high-temporal resolution data are needed, which can be achieved by using data from geostationary satellites [Neukermans et al., 2009, 2012; Choi et al., 2012].
 The suspension of SPM in lakes and bays is highly correlated with the wind intensity and duration [Howick and Wilhm, 1985; Chen et al., 2007] as well as with circular water motion under waves and large-scale currents [Sheng and Lick, 1979; Petrusevics, 2005] and is a function of water depth [Howick and Whilm, 1985]. Cho  investigated the effects of diurnal and seasonal prevailing winds on the turbidity and concluded that winds increased the turbidity of littoral areas and that prolonged unidirectional winds can affect shores by increasing turbidity and wave energy.
 Turbidity of the Dead Sea was observed previously by in situ measurements, where higher SPM load was measured in the upper layer relative to the lower water body [Neev and Emery, 1967]. Following flood events, a lateral SPM gradient was recorded in the surface water by means of in situ measurements, with high nearshore concentrations (20 mg/L) decreasing toward the central part of the lake (4 mg/L) [Levy, 1981]. Herut et al.  combined an airborne multispectral sensor together with in situ SPM filtering along the western part of the lake and found local maxima of SPM concentrations at the Jordan River estuary and the Qidron River mouth (>100 mg/L) and a sharp decrease of SPM seaward (SPM concentration dropped to the background values ∼4 mg/L about 1 km offshore). Yet, the temporal and spatial variations of turbidity in the Dead Sea still need to be explored while they remain loosely constrained. Here we address this shortcoming by means of combining remote sensing, in situ continuous observations, and simplified conceptual/mathematical models. We exploit moderate spatial resolution together with high-temporal resolution satellite data to monitor daily variations of surface turbidity in the summer and winter, and the effects of winter floods.
2. Approach and Methods
 A remote sensing methodology was developed to characterize the seasonal variability of turbidity using moderate resolution imaging spectroradiometer (MODIS) data (section 2.1.1) and daily variability (section 2.1.2) using geostationary satellite data. “Ground truth” in situ data for turbidity (section 2.2.1), surface reflectance (section 2.2.2), wind (section 2.2.3), and wave intensity (section 2.2.4) provide the physical basis for the remote-sensing proxies developed herein for SPM.
2.1. Remote Sensing
2.1.1. Seasonal Cycle
 MODIS on NASA Terra and Aqua sun-synchronous satellites (Terra-10:30; Aqua-1:30 AM/PM LT) data were obtained from NASA's Earth observing system data and information system (Available at http://reverb.echo.nasa.gov). MODIS 250 m/pixel top of atmosphere (TOA) radiance data from band 1 (620–670 nm) and band 2 (841–876 nm) were previously used to monitor coastal and estuarine turbid waters [Li et al., 2003; Hu et al., 2004]. The processing of such data requires atmospheric correction. In this context, the dark pixel method is often used to remove aerosol contamination from TOA radiances [Hu et al., 2000; Doxaran et al., 2002; Miller and McKee, 2004; Chen et al., 2007; Doxaran et al., 2009; Rodríguez-Guzmán et al., 2009; Feng et al., 2012]. The 250 m/pixel Terra/Aqua daily product (MOD/MYD09GQ) estimates the surface reflectance by correcting absorption and scattering of gases and aerosols [Vermote et al., 1997; Ahmad et al., 2010; Moreno-Madrinan et al., 2010]. Surface reflectance can be also derived using the SeaWiFS data analysis system (SeaDAS)—a comprehensive image analysis package for processing ocean color data [Baith et al., 2001] (http://seadas.gsfc.nasa.gov/). In the Dead Sea, both the MOD/MYD09GQ product and the SeaDAS corrected surface reflectance are affected by over corrections, i.e., negative reflectance [see., Feng et al., 2012], and from aliasing (distortion) due to coarse resolution of ancillary data used in the processing of these products.
 In this paper we corrected MODIS bands 1 and 7 for Rayleigh scattering and gaseous absorption (water vapor and ozone) to obtain the corrected water reflectance using the corrected reflectance science processing algorithm (CREFL_SPA), version 1.7.1, developed by the direct readout laboratory at NASA/GSFC. The data were geolocated using MODIS reprojection tool swath software. The CREFL_SPA processes MODIS Aqua and Terra Level 1B data to create the MODIS Level 2 Corrected Reflectance product. The algorithm performs a simple atmospheric correction with MODIS visible, near infrared, and shortwave infrared bands (bands 1 through 16). The corrected reflectance products created by CREFL_SPA are very similar to the MOD/MYD09GQ in clear atmospheric conditions, since the algorithms used to derive both are based on the 6S radiative transfer model [Kotchenova et al., 2006]. In contrast to MOD/MYD09GQ, the CREFL_SPA algorithm does not perform aerosol correction. Following Wang and Lu , Wang et al. , Gordon , and Vermote et al.  we corrected the effect of aerosols by subtracting band 7 (2105–2155 nm) from band 1. Sulfur particles have the largest contribution to aerosols concentration over the Dead Sea [Levin et al., 2005]. The spectral signature of sulfur (<1 µm) between 600 and 2500 nm is spectrally flat (http://speclib.jpl.nasa.gov/). Therefore, the aerosol effect at 645 nm (band 1) and 2130 nm (band 7) are practically the same. This method was found to be suitable for relatively clear water and is based on the assumption that water is usually considered as dark in MODIS band 7 due to the very high absorption by the water itself in this band. Under this assumption, the reflectance that remains in band 7 after Rayleigh correction can be attributed entirely to aerosols. To avoid pixels with high turbidity, we picked the lowest band 7 reflectance for the aerosol correction (band 1–band 7) and applied this correction to all the band 1 Dead Sea pixels.
 For the seasonal cycle we averaged 20 winter and 20 summer MODIS images (MOD/MYD02QKM) from 2009 to 2010 (Table 1). These images were selected after verifying (visual inspection) that all pixels are free of sun glint and clouds. Land pixels were masked using a 500 m buffer offshore.
Table 1. Dates of MODIS Images (LT) Used for Seasonally Turbidity Maps (See Figure 8)
2 Aug 2009, 11:25
1 Feb 2010, 12:10
4 Aug 2009, 11:15
7 Feb 2010, 13:10
5 Aug 2009, 13:35
23 Feb 2010, 10:00
6 Aug 2009, 14:20
10 Feb 2010, 10:25
22 Jun 2010, 14:15
11 Feb 2010, 12:45
24 Jun 2010, 14:05
12 Feb 2010, 10:15
28 Jun 2010, 13:20
16 Feb 2010, 13:05
1 Jul 2010, 14:10
2 Mar 2010, 10:00
3 Jul 2010, 14:00
2 Mar 2010, 13:15
4 Jul 2010, 11:25
1 Dec 2010, 13:05
7 Jul 2010, 13:35
2 Dec 2010, 10:30
15 Aug 2010, 13:40
3 Dec 2010, 12:50
16 Aug 2010, 11:05
4 Dec 2010, 10:20
17 Aug 2010, 11:50
5 Dec 2010, 12:40
17 Aug 2010, 13:25
8 Dec 2010, 13:10
18 Aug 2010, 14:10
15 Dec 2010, 13:15
19 Aug 2010, 11:40
16 Dec 2010, 10:45
20 Aug 2010, 13:55
17 Dec 2010, 13:05
23 Aug 2010, 11:15
18 Dec 2010, 10:30
24 Aug 2010, 11:55
22 Jan 2011, 12:40
2.1.2. Daily Variability
 The European geostationary meteorological satellite meteosat second generation (MSG) operated by European organisation for the exploitation of meteorological satellites (EUMETSAT) provides 15 min interval data from 11 spectral channels and one high-resolution visible (HRV) channel [Schmetz et al., 2002]. For the Dead Sea the spatial resolution is approximately 4 and 1.6 km for the spectral channels and for the HRV channel, respectively. The high-sampling frequency provided by MSG data provides a unique opportunity to resolve the daily variability of surface turbidity. The spectrally wide (0.37–1.25 µm) HRV band is usually used qualitatively. As the maximal width and length of the Dead Sea are ∼18 by 50 km, respectively, the higher spatial resolution of the HRV channel appears more suitable for monitoring turbidity. Therefore, we use HRV time series together with 250 m/pixel MODIS data, when available, to enable quantitative use of the HRV data. This requires treatment of the bidirectional reflectance (BDRF) of pure water, and would be valid as long as the absorption is constant or covaries with SPM. This is correct for MODIS (see horizontal arrows in Figure 4b) but less so for spinning enhanced visible and infrared imager (SEVIRI) where dissolved and particulate substances could modulate the signal in the blue. Nevertheless, the absorption in the blue is small and seems homogeneous throughout the lake [Boss et al., 2013].
 Fifteen minutes interval data from SEVIRI on MSG were downloaded from the EUMETSAT data centre (http://www.eumetsat.int/). The raw data were converted to TOA reflectance using EUMETSAT calibration coefficients. Images with clouds over the Dead Sea were manually removed. The HRV data are used here to examine the spatial-temporal dynamics of the daily variability of turbidity in summer and winter. One-hour interval HRV reflectance maps were averaged over 10 days in winter and summer (for detailed dates see Table 2). There is no standard procedure to apply atmospheric corrections to HRV due to the width of this channel. We took the following steps to calibrate all available HRV data between 8:00 and 16:00 local time (LT) in the winter and between 8:00 and 17:00 LT in the summer against MODIS surface reflectance:
 1. Land pixels were masked using a 500 m buffer offshore.
 2. The darkest pixel in the Dead Sea is assumed to represent clear water with minimal atmospheric effects and minimal contribution of turbidity to the HRV reflectance. As the lake is small, we assume that the sun elevation is the most dominant factor contributing to changes in the BDRF. Therefore we average the HRV darkest pixel in each 15 min time step (not necessarily the same pixel in each time step) over 10 days in each season (see Figure 2 for the summer season), assuming that the sun elevation does not change dramatically within ±5 days (in summer), with respect to the 15 min time step, or in winter when the rate of maximum sun elevation change is the smallest.
 3. The resulting curve serves as a “normalization factor” to correct the geometric effects of clear water BDRF. This is done by dividing the TOA reflectance at all the time steps from stage (2) by the maximal value, resulting with a curve with maximum of unity (see Figure 2)
 4. TOA reflectance of all the Dead Sea pixels in each time step were then divided by the corresponding normalization factor of the relevant time step from stage (3).
 5. HRV reflectance values were calibrated against spatially averaged simultaneous corrected MODIS reflectance. We averaged reflectance from all MODIS pixels to match the HRV resolution by using Thiessen polygons [Thiessen, 1911].
 6. This calibration was applied to the normalized TOA reflectance of all the time steps.
Table 2. Dates of Hourly Averaged MSG Images Used for Hourly Turbidity Maps (See Figure 5)
16 Aug 10
1 Dec 10
17 Aug 10
4 Dec 10
18 Aug 10
14 Dec 10
19 Aug 10
21 Dec 10
20 Aug 10
22 Dec 10
21 Aug 10
24 Dec 10
22 Aug 10
26 Dec 10
23 Aug 10
27 Dec 10
24 Aug 10
28 Dec 10
25 Aug 10
29 Dec 10
2.2. In Situ Measurements
 Figure 3 presents the setup of the observing stations in the Ein Gedi region of the Dead Sea.
2.2.1. Turbidity and Beam Transmission
 Time series of turbidity were measured near the EG coast station, 50 m offshore where the seafloor is 9 m deep, using a Campbell OBS3 sensor placed at a depth of 3 m. The turbidity data were recorded every 5 min in nephelometric turbidity units (NTU) (light scattered by particles by source light wavelength of 850 ± 5 nm). The accuracy of turbidity measurements is ±0.5 NTU or ±2% (whichever is larger). Unlike the remote sensing turbidity observations, which are based on reflectance of sunlight, in situ measurements are based on a light source in the sensor and thus are continuous. The relationship between OBS signal and sediment concentration is almost linear for many suspended sediments [Downing, 2006]. Beam transmission was measured in situ using a wetlabs C-star a 25 cm beam transmissometer (650 nm).
2.2.2. Surface Reflectance
 Highly turbid waters contain a large number of scattering particulates, and are therefore expected to reflect more light. The amount of solar radiation reflected from the water can be a proxy of water turbidity as long as the absorption is constant or covaries with SPM. Surface reflectance measurements were conducted from the “Taglit” research vessel along a 5 km transect from the Ein-Gedi coast toward EG100 station, using an “ASD Examiner” Full Range field spectrometer (spectral range of 350–2500 nm and wavelength resolution of 10 nm).
2.2.3. Wind Speed and Direction
 Wind speed and direction were measured at the two stations, EG coast (on a mast 10 m above sea surface) and EG100 buoy (on a mast at 3.7 m above the sea surface), every 20 min (for location see Figures 1 and 3; for more details see Hecht and Gertman ).
2.2.4. Wave Intensity
 Wave amplitude, the difference between maximum and minimum sea level recorded in 120 records of 5 s intervals, was measured using Campbell CS455 level meter (resolution: 0.175mm; accuracy: 2 mm) attached to the sea floor (∼3 m deep, see Figure 3).
2.3. Laboratory Measurements of Turbidity Versus Concentration of Suspended Particles
 Turbidity values of silty clay from the Dead Sea shores, sorted halite and Attapulgite clay (international lab standard) were measured in the lab. Turbidity was measured using a portable Hach 2100Q turbidity meter. The samples of turbid water were prepared by weighing the particular matter (accuracy of 0.001 mg), and then adding it to Dead Sea brine samples (50 ml with Resolution of 0.5 ml). After mixing, the solution was transferred to a ∼15 cc glass tube for turbidity measurement (NTU). Turbidity measurement accuracy is ±2% and resolution is 0.01 NTU.
3. Results and Discussion
3.1. Relation Between Surface Reflectance and Turbidity in the Dead Sea
 In situ turbidity measurements (section 2.2.1) were compared with spectral (0.4–2.5 µm) reflectance measurements (section 2.2.2) obtained simultaneously. Figure 4a demonstrates the correlation between measured in situ turbidity and surface reflectance at 645 nm. Figure 4b presents in situ surface reflectance 350–1250 nm along an E-W transect from EG buoy to EG coast (see Figure 1) reflecting the decrease of turbidity from the coast toward the central parts of the lake between four stations located 0.2–5 km offshore. The correlation between turbidity and surface reflectance indicates that with appropriate atmospheric corrections TOA reflectance data can serve as feasible proxies for water turbidity in the Dead Sea waters.
3.2. Role of Shore Abrasion on the Dynamics of Dead Sea Surface Turbidity
3.2.1. Daily Variability of Surface Turbidity—Role of Wind and Waves
 The dynamics of the daily surface turbidity distribution in the Dead Sea is presented in Figure 5 in a series of hourly maps of the Dead Sea surface reflectance. The maps are based on HRV surface reflectance averaged over 10 days during (a) winter and (b) summer. Patterns of high concentrations along the western shores and southeastern and southwestern embayment are typical (red color in the maps). Wind charts are also presented in Figure 5 based on hourly averaged wind data from EG100 (Figures 1 and 3).
 In the summer, a wide band along the western shore appears in the morning (8:00–12:00 LT in Figure 5b), whereas in the afternoon, with the decrease in wind speed (∼2 m/s), the turbid band narrows and is less pronounced. In the winter, these radial patterns are less pronounced (Figure 5a), as will be discussed in the next section.
 The solar reflectance maps, limited to daytime, provide spatial coverage of the entire lake surface, with good temporal resolution. Continuous in situ turbidity observations from the EG coast station (for location see Figure 3) provide the full daily cycle. Figure 6a presents a time series of wind speed, wave amplitude and in situ measured turbidity. A clear correlation is seen between wind intensity and wave amplitude with minimal delay; waves develop as the wind speed exceeds ∼4 m/s, and waves diminish as the wind calms below this threshold value. Turbidity shows a similar diurnal pattern, with a time lag of a few hours, in which waves of >20 cm amplitude are related to the formation of turbid water. Figure 6b shows similar time series for the solar reflectance from satellite images, in a few pixels along transect from EG coast to the 5 km offshore EG100 station as a proxy for turbidity. The spatial and temporal pattern is clear––turbidity decreases with increasing distance from shore, and with reduced wind intensity throughout the day. Figure 6c presents similar time series, turbidity was measured in the coast station and simultaneously at the offshore EG100 station. The turbidity values at the coast station are much higher than the offshore station (1–7 NTU and 0.5–0.7 NTU, respectively). This is in agreement with the remote sensing data shown in Figures 5 and 6b. We suggest that the offshore turbidity changes are the result of halite precipitation, driven by nighttime cooling [Steinhorn, 1983; Stiller et al., 1997; Gavrieli, 1997]. The halite precipitation signal is much weaker than the shore abrasion signal recorded in the coast station, since the halite crystals are typically large (150–400 µm according to Levy ) and have thus higher settling velocity, than the clay-rich sediments in the shores [Neev and Emery, 1967; Garber, 1980]. Thus, the effect of halite precipitation on water turbidity is hard to observe nearshore, since the sediment suspension there creates a much higher turbidity signal.
 To translate turbidity of the brine, as an optical property, to the mass concentration of SPM in the brine, we conducted laboratory measurements of turbidity of Dead Sea brine containing various fractions of silty clay samples from the Dead Sea shores (see Figure 7 and section 2.3 for the procedure). Two samples were taken as representative of the fine grained silty shore material (from EG coast). A linear relationship exists between the turbidity (NTU) and the concentration of SPM (mg/L), although the slopes of the lines can differ significantly. As an example, minimum turbidity measured in the Dead Sea is ∼1.3 NTU (Figure 6a), which according to Figure 7 corresponds to suspended matter concentrations that vary between 6.4 mg/L to 39.4 mg/L (green and blue curves, respectively). Similarly, maximum typical turbidity measured in situ is ∼8 NTU, which corresponds to solution concentrations that vary between 39.2 and 242.2 mg/L (green and blue curves, respectively). In flood event, where turbidity exceeds 100 NTU (see below), concentrations of 490–3030 mg/L are expected, depending on the type of SPM. For reference, also presented in Figure 7 is a turbidity-concentration relation for Attapulgite clay (international lab standard). The high variation of the measured relation between the optical property and mass concentration of SPM in natural mud mixed with the Dead Sea brine emphasizes the difficulty to obtain SPM concentrations based on turbidity measurements (NTU or reflectance). Halite turbidity concentration relation is also presented for sorted size grains of three intervals. The larger grain size interval (0.074–0.125 and 0.125–0.25 mm) creates a significantly less turbid effect than the smaller size (0.062–0.074 mm). The low turbidity of the larger grain size can explain the relatively low effect of halite precipitation on the water turbidity.
 The observed turbidity dynamics in the Dead Sea suggests that the SPM source is primarily from shore abrasion by wave action, and that the turbidity daily variation is governed by the diurnal wind cycle. The spatial turbidity patterns during the day are thus highly linked to the hours of suspended matter flux from shores. In the Dead Sea, the availability of unconsolidated sediments is continuously “supplied” by the continuous drop of the Dead Sea lake level (>1 m/yr), which exposes silty shores to wave action. On land, scars of wave action are clearly visible as coastal stairs (“cliffs”) that follow topographic contours, and represent erosion along previous coastlines by wave action.
 Dust deposition over the Dead Sea surface occurs mainly in discrete events of dust storms. Deposition rate ranges between ∼10 gr m−2 yr−1 in summer and winter, to an order of magnitude larger rate in spring and autumn, when desert storms occurs [Singer et al., 2003]. The effect of dust on Dead Sea turbidity will be uniform along the Dead Sea surface, and thus cannot explain the concentric patterns observed in remote sensing and in situ measurements, and will not explain the correlation of turbidity time series nearshore to wind and wave intensity over the diurnal cycle.
3.2.2. Seasonal Variations of the Dead Sea Turbidity—Role of Water Column Stability
 Figure 8 shows the seasonal surface reflectance maps, based on averaged MODIS band 1 surface reflectance for summer and winter (20 maps each), with the corresponding wind charts. In the summer a relatively clear concentric pattern appears, as presented above, with high-reflectance values along the shore, and high reflectance in the major embayment (the northern and two southern embayments), with a gradual decrease of reflectance values offshore. In the winter, a much narrower turbid concentric band appears, and the average surface reflectance of the entire lake is also lower (2.4% compared to 2.2%, respectively). In the winter the wind intensity is similar to the summer, but the direction in the winter are dispersed with respect to the summer, when northern winds, attributed to the Mediterranean Sea breeze, dominate from late afternoon to late morning [Hecht and Gertman, 2003].
 The turbidity gradients from the shores to the central parts of the Dead Sea are much higher in the summer compared to the winter. The seasonal variation can be explained by two processes. First, the prolonged unidirectional winds in the summer is more likely to affect the shores by increasing turbidity and wave energy [Cho, 2007], compared to the more variable directions of winter winds. In addition, the summer / winter limnological state is significantly different: thermal layering exists in the summer months, whereas vertical mixing occurs in the winter [Hecht and Gertman, 2003]. The lateral gradients in a stratified layer are expected to be higher than in a vertically mixed state. Summer stable layering reduces the effective depth of mixed upper layer from 300 to 25 m [Hecht and Gertman, 2003]. This implies that the fine turbid particles that are added in the shorelines are mixed in the summer within a much smaller effective water body and may thus be more concentrated near surface. A somewhat similar seasonal change was observed in the thermal properties of the Dead Sea surface from satellite images, in which higher surface temperature variability was observed in the summer, compared to winter [Nehorai et al., 2009].
 To further understand the physical processes responsible for the seasonal differences in the turbidity gradient we consider a simple, Lagrangian trajectory model in which passive, neutrally buoyant particles can be transported by three distinct processes: (1) turbulent mixing (horizontal and vertical), (2) a near surface direct wind drift current, and (3) the subsurface thermohaline currents. This is described by:
where X is the three dimensional displacement of the position of the particle, with (X,Y,Z) representing the cross-shore, along shore, and vertical directions, respectively; UW is the near surface, horizontal wind drift current; US is the subsurface, horizontal current; dt is the time step; and dXT is the turbulent displacement simulated by a random walk model with components
where α, β, and γ are random numbers in the range (−1,1) and KH, KV are the horizontal and vertical eddy diffusivities, respectively. Values of the eddy diffusivities were estimated from the modeling results of Gavrieli et al. . KH is assigned a constant value of 15 m2/s, which is toward the lower end of the range of values estimated by Carlson et al. , for the northern tip of the Gulf of Eilat which has physical dimensions, and climatic conditions similar to the Dead Sea. Stratification is indirectly introduced through KV which is assigned values representative of the surface mixed layer in the respective seasons. In winter it is given a value of 0.05 m2/s from the surface down to a depth of 100 m, and 1 × 10−5 m2/s below 100 m. In summer it decays exponentially from 0.05 m2/s at the surface down to a depth of 20 m, with a scale height of 5 m. Below 20 m it is assigned the same deep water winter value. UW is computed from the wind vector using common drift factors of 0.02 for the speed and rotation of 10° to the right. The speed decays exponentially with depth with a scale height of 5 m. The wind is assumed to be northerly (shore parallel) with a speed of 8 m/s. Based on a general assessment of the current measurements and observed drifter tracks, US is assumed to be mainly southward (shore parallel) with a rotation of 20° offshore and a speed of 15 cm/s.
 In Figure 9 we show the results from four 24 h long simulations run in summer and winter with various combinations of the three transport mechanisms turned on and off. The simulations were run for the western shore which is a primary source of mud and sand. In each simulation 2000 particles were released from points randomly distributed along a cross section extending 50 m eastward from the shore and in the upper 0.5 m of the water column. Particles that drifted onto land were randomly returned to the water in a 25 m wide strip adjacent to the shore. The results are expressed in terms of the cross shore distribution of the number of trajectory points falling within the upper 5 m of the water as a percent of the total number of points in this layer.
 Figure 9a shows the simulation with the turbulent mixing only. Three distinct zones can be seen here. In general, the diffusive mixing creates a 300–400 m wide strip adjacent to the coast with relatively high and uniform particle concentrations. This zone is separated from the low-concentration offshore zone by a very sharp boundary. The values in summer are somewhat higher than those in winter due to the deeper winter mixing and the removal of particles from the near surface layer implied by the vertical profile of KV. The third zone is a very narrow strip of higher concentration at the shore. The concentration of particles in this zone is very high compared to the diffusive mixing zone and the open water zone due to the particles that drift onto land and are returned to the sea.
 Figure 9b shows the results from the simulation with the wind drift turned off but with the subsurface turned on. The net effect of the subsurface current is to enhance the transport of particles from the shore toward the open water. In this case the three zones are still visible with the main effect being felt in the diffusive mixing zone. This zone widens to 500–600 m and consequently the relative particle concentrations drop to less than one half of the values in the previous simulation. The boundary between this zone and the low-concentration open water zone is also not as sharply defined, but the very high concentration nearshore zone is preserved. The seasonal effect is still apparent with higher values in summer than in winter. Due to the generally lower values in winter, the boundary between the mixing zone and the open water zone is weakened.
 Figure 9c shows the results when the wind drift transport is turned on but the subsurface current is turned off. The net effect of the wind drift is to enhance the onshore transport of particles in the upper layer. Consequently the mixing zone narrows to less than 300 m. Compared to the diffusive mixing only case, here the concentrations in the mixing zone are somewhat higher while in the open water zone they are noticeably lower and thus the boundary between these two zones is much sharper. The enhanced shoreward transport of particles is strong enough to hide the seasonal differences.
 Figure 9d shows the results when all three transport processes are turned on. The combination of all three effects emphasizes the seasonal differences in the characteristics of the mixing zone. In winter the role of the subsurface transport is evident in spreading the particles to the open water zone, thereby blurring the differences between the two zones and resulting is a nearly flat, relatively low-concentration profile. In summer the mixing zone is maintained as a distinct ∼500 m wide region separating the very concentration nearshore zone and the low-concentration open water zone. Furthermore a clear boundary is maintained between the mixing and open water zones as observed in the composite seasonal images in Figure 8.
 Finally, to test the sensitivity of the results to the intensity of the horizontal turbulent mixing, additional simulation were run with all three processes turned on with values of KH ranging from 0.1 to 100 m2/s. For very low values there is a peak concentration located ∼150 m from the shore which decays smoothly to the open sea zone. The broad mixing zone seen in Figure 9d is absent. For very large values the entire profile is smoothed to low, nearly uniform values. The ∼500 m wide mixing zone with a sharp transition to the clear open sea develops only for values on the order of 10 m2/s, thus confirming the initial value chosen. The model results suggest that the difference between the ∼10 cm/s typical current intensity [Gavrieli et al., 2011] and the much faster propagation of the coastal turbid water could be explained by turbulent mixing.
3.3. Surface Turbidity Dynamics Associated with Flood Events
 Winter floods are an additional major source for SPM affecting the Dead Sea turbidity. Such floods can be regarded as rare wintertime point-source events in which fluvial waters loaded with sediments reach the Dead Sea and spread from the shoreline at the lake's surface. Unlike shore abrasion discussed above, lateral spreading of turbid flood plumes is driven by the momentum of the river discharge and the density differences between the plume and the underlying dense brine (the Dead Sea density is ∼1.24 gr/cc). Typically, plumes loaded with sediments entering into lakes tend to sink and form dense hyperpycnal plumes due to the sediment load. However the Dead Sea is an exceptional case because of its very high density and therefore the turbid water forms a hypopycnal plume that remains near the surface. Our methodology allowed us to follow the spatial and temporal spread and lifetime of winter desert flood plumes. Such plumes can be recognized by their high-turbidity values and their lower temperature (Figure 10), since the floodwaters are more turbid and are cooler (typically 15–20°C according to the Hydrological Service) compared to the winter Dead Sea temperature of 24–25°C [Hecht and Gertman, 2003]. Continuous in situ turbidity measurements captured a full record of a flood plume, measured at two stations (EG coast 50 m offshore and EG100 5 km offshore, both 3 m deep) during the event of 29 February–3 March 2012 (Figure 11). The coast station, located 50 m offshore near the Nahal Arugot outlet, recorded two major plumes entering the Dead Sea, with turbidity values of 40–130 NTU lasting for approximately 1 day. This flood event added a volume ∼106 m3, with a flux of up to 30 m3/s (Hydrological Service of Israel). A much weaker signal of the plume (4–10 NTU) reached the EG100 buoy, 5 km offshore, a day later and lasted for a much shorter time.
 A rough estimation of the spreading and dimensions of such a plume can be obtained from the steady state equations describing the two dimensional advection-diffusion of a hypopycnal, turbid plume emanating from a river mouth [Syvitski et al., 1998]. The equation for the sediment inventory, I, is
where (x,y) are the horizontal coordinates with respective velocity components (u,v). The first term on the right hand side represents a first order removal (settling) process with rate λ, while the second and third terms are the turbulent diffusion with diffusivity K. The suspended sediment in the river discharge is composed of a mixture of clay and silt with removal rates of 2 and 4.8 d−1, respectively. Assuming a 100 m wide, 1 m deep river mouth, a flux of 30 m2/s would produce a jet of turbid water with a velocity of 0.3 m/s. At a distance of 5 km from the discharge point, the sediment concentration is a narrow jet is expected to decay to 7% of its input value at the river mouth. This is in close agreement with the measurements at station EG100 where the plume signal is 7–10% of its value at the nearshore EG coast station as shown in Figure 11.
 The high reflectance and lower temperature of the flood plumes are clearly identified in the MODIS sea surface temperature (SST) and surface reflectance maps following a flood event from 10 February 2010 (Figures 12a and 12b, respectively). Two cold and turbid plumes are recognized, one in the north, starting from the Jordan River, and the other in the south-western part of the lake starting from Nahal Ze'elim outlet and extending northward as elongated plume. Figure 13a provides hourly surface reflectance maps of the Dead Sea, following a flood event in 17–18 January 2010. Two major sources for turbidity are observed, a plume from the Jordan River in the north, and a plume in the south west, from Nahal Arava and Nahal Zeelim (See Figure 1 for locations). Figure 13b presents reflectance transect across the turbidity plume of the Nahal Arava at different times along the day. These transects capture the propagation of the plume in the morning (9:00–12:00 LT) and the decay of the plume during noon (12:00–15:00 LT). The plume faded after 1 day (remnants can be seen in the south-west embayment on the lower right map in Figure 13a).
 Remote sensing allows continuous monitoring of the extent and intensity of the turbidity of these plumes in cases where cloud-free skies immediately followed the flood event. The observations, backed up with in situ measurements, show that the plumes change their shape and properties within a few hours and their lifetime is limited to a few days. Mixing processes erode the original properties of the plume (temperature turbidity, and salinity), as the plume spreads and mixes, despite the initial stability of the plume (∼20% initial density difference).
 We developed a methodology to attain daily variability of turbidity and explored the dynamics of post flood plumes in the Dead Sea by means of remote sensing. The results show that turbidity is concentrated along the silty shores of the Dead Sea (north, west, and south) and the southern embayment, with a gradual decrease of turbidity values from the shore line toward the center parts of the lake. This pattern is most pronounced following the hours of intense winds (nighttime). A few hours after wind calms the turbidity concentric pattern fades. In situ observations show clear relation between wind intensity, wave amplitude and water turbidity near coast. The observed turbidity dynamics in the Dead Sea suggests that the source of suspended matter is from shore abrasion by wave action, and that the daily turbidity variations are governed by the diurnal cycle of the wind, which transfers unconsolidated sediments from the shore to the water body by wave action. In the winter and summer a similar turbidity pattern is observed, but with a much narrower turbid concentric structure in winter. This is explained by the combined effects of horizontal transport processes and vertical homogenization in the winter, and thus suspended matter from the shores is mixed much more effectively in the winter. The dynamics of suspended matter contributions from winter flash floods are also presented with hourly surface reflectance maps showing the spreading of the plumes. These floods introduce large amount of suspended matter. From the inlets, the flood water spread as surface plumes due to the lower density compared to the Dead Sea extremely high density.
 The turbidity of the Dead Sea is a major issue regarding the future of this unique lake. This paper aimed to characterize the SPM distribution and its dynamics within the Dead Sea and to robustly evaluate the potential impact of additional SPM which will be an outcome of several planned projects. The introduction of seawater into the Dead Sea through the Red Sea Dead Sea Canal project will result in gypsum precipitation, which may result in higher water turbidity [Gavrieli et al., 2011]. This is physically related to hypopycnal flood plumes, due to the lower density of the introduced seawater. The dumping of salt from the industrial ponds into the Dead Sea [Lensky et al., 2010] is physically related to the current state of particles added from shore abrasion. These particles tend to create turbidity patterns that are formed and decay within the daily cycle, since the particles are transported horizontally while settling. Water quality turbidity criteria for the Dead Sea should take into account aesthetic aspects (biota and water quality are not relevant in the hypersaline sterile brine). According to the Canadian Council of Ministers of the Environment, the turbidity should not exceed 50 NTU for recreation and aesthetic aspects. This turbidity threshold is higher than the turbidity measured in the Dead Sea (<10 NTU), except for flood events which transport large quantities of sediments from the rivers. The remote sensing methodology developed and applied here has led a preliminary understanding of the processes that determine the SPM patterns resulting from various sources. Further use of the remotely sensed data will provide a powerful tool for short and long term monitoring of the turbidity of the sea as well as forming a basis for developing models for predicting the fate of SPM in the lake.
 We thank Ittai Gavrieli for helpful discussions, Raanan Bodzin, Ali Arnon, Hallel Lutzki, and Tal Ozer for helping with the installation and calibration of the instruments. Onn Cruvi for discussions on dust contribution. Silvy Gonen and the “Taglit” team Meir Ifrach and Shachar Gan-El provided research vessel services. Udi Galili from the Hydrological Service of Israel supplied data on floods intensity and temperature. Yehudit Harlavan and Amir Sandler helped in the laboratory. The reviewers Emanuel Boss, Sergey Stanichny, and an anonymous reviewer are acknowledged for their significant comments and suggestions that significantly improved the manuscript. EUMETSAT provided the MSG SEVIRI data, NASA provided the MODIS data, and DRL provided the Corrected Reflectance algorithm (Version 1.7.1). The research was supported by the Earth Science Research Administration, the Ministry of National Infrastructures (Israel).