Volume 54, Issue 1 p. 150-160
Research Article
Free Access

Diurnal Course of Evaporation From the Dead Sea in Summer: A Distinct Double Peak Induced by Solar Radiation and Night Sea Breeze

N. G. Lensky

Corresponding Author

N. G. Lensky

Geological Survey of Israel, Jerusalem, Israel

Correspondence to: N. G. Lensky, [email protected]Search for more papers by this author
I. M. Lensky

I. M. Lensky

Department of Geography and Environment, Bar-Ilan University, Ramat Gan, Israel

Search for more papers by this author
A. Peretz

A. Peretz

Geological Survey of Israel, Jerusalem, Israel

Department of Geography and Environment, Bar-Ilan University, Ramat Gan, Israel

Search for more papers by this author
I. Gertman

I. Gertman

Israel Oceanographic and Limnological Research, Haifa, Israel

Search for more papers by this author
J. Tanny

J. Tanny

Institute of Soil, Water and Environmental Sciences, Agricultural Research Organization, Volcani Center, Rishon LeZion, Israel

HIT - Holon Institute of Technology, Holon, Israel

Search for more papers by this author
S. Assouline

S. Assouline

Institute of Soil, Water and Environmental Sciences, Agricultural Research Organization, Volcani Center, Rishon LeZion, Israel

Search for more papers by this author
First published: 27 December 2017
Citations: 23

Abstract

Partitioning between the relative effects of the radiative and aerodynamic components of the atmospheric forcing on evaporation is challenging since diurnal distributions of wind speed and solar radiation typically overlap. The Dead Sea is located about a 100 km off the Eastern Mediterranean coast, where and the Mediterranean Sea breeze front reaches it after sunset. Therefore, in the Dead Sea the peaks of solar radiation and wind speed diurnal cycles in the Dead Sea are distinctly separated in time, offering a unique opportunity to distinguish between their relative impacts on evaporation. We present mid-summer eddy covariance and meteorological measurements of evaporation rate and surface energy fluxes over the Dead Sea. The evaporation rate is characterized by a clear diurnal cycle with a daytime peak, few hours after solar radiation peak, and a nighttime peak coincident with wind speed peak. Evaporation rate is minimum during sunrise and sunset. Measurements of evaporation rate from two other water bodies that are closer to the Mediterranean coast, Eshkol Reservoir, and Lake Kinneret, present a single afternoon peak, synchronous with the sea breeze. The inland diurnal evaporation rate cycle varies with the distance from the Mediterranean coast, following the propagation of sea breeze front: near the coast, wind speed, and radiation peaks are close and consequently a single daily evaporation peak appears in the afternoon; at the Dead Sea, about a 100 km inland, the sea breeze front arrives at sunset, resulting in a diurnal evaporation cycle characterized by a distinct double peak.

Key Points

  • We present eddy-covariance evaporation measurements at the Dead Sea and two waterbodies along the path of the Mediterranean Sea breeze
  • Diurnal evaporation from the Dead Sea presents a distinct double peak induced by daytime radiation and nighttime Mediterranean Sea breeze
  • The inland diurnal evaporation cycle varies with the distance from the Mediterranean coast, following the propagation of sea breeze front

1 Introduction

Evaporation is a major component of the water, heat, and salt budgets of lakes (Assouline, 1993; Lensky et al., 2005). In many cases, there are other unknown components affecting such budgets, like inflow from ungauged basins or springs or water losses due to percolation from the bottom, and accurate estimate of evaporation can improve the evaluation of these unknowns (Assouline, 1993; Tanny et al., 2008). Therefore, efforts have been invested over decades in developing tools allowing an estimate of this variable (Henderson-Sellers, 1986; Lenters et al., 2005; Linacre, 1993; Rosenberry et al., 2007) but accurate estimate of evaporation rate is still challenging.

Evaporation from a water surface is driven by wind, water-air vapor pressure gradients, and the internal heat stored in the water, which is supplied from external sources such as solar radiation and/or downward sensible heat flux (Brutsaert, 2005). Consequently, the main approaches generally applied to estimate evaporation rates are: the energy budget method that remains by far the most common for estimating surface fluxes, and has been widely applied to lakes; the flux-gradient approach for heat and mass transfer across the boundary-layer over the water surface; and a combination between the two (Brutsaert, 1982, 2005; Monteith & Unsworth, 1990). All these methods rely on the measurement of standard meteorological variables (solar radiation, air temperature and humidity and wind speed) and require the evaluation of the heat storage within the water body. Therefore, they provide indirect estimates of the evaporation rates. The development of fast-response sensors such as sonic anemometers and infrared and ultraviolet absorption hygrometers allowed the application of the eddy-correlation method (Swinbank, 1951) for direct measurement of latent and sensible heat fluxes from the water surface. During the past decades the Eddy-Covariance System (ECS) became the most reliable and accurate technique for direct evaporation rate (E) measurement, provided that it is applied under certain limitations (Itier & Brunet, 1996). ECS has been used for direct E measurements from lakes, reservoirs and seas (Allen & Tasumi, 2005; Assouline & Mahrer, 1993; Assouline et al., 2008, 2016; Liu et al., 2009, 2012; McGowan & Sturman, 2010; Sene et al., 1991; Stannard & Rosenberry, 1991; Tanny et al., 2008, 2011).

The ECS allows determining the diurnal behavior of E at a high temporal resolution. Since evaporation is responding to both solar radiation (via the heating of the water body) and wind velocity, the ECS enables the partitioning between these two factors if they are distinct in time. While radiation has a characteristic diurnal cycle, typically peaking at noon, this is not the case for wind speed that has a more erratic pattern, which may or may not overlap the radiation diurnal pattern. Therefore, distinction between the relative contributions of the aerodynamic and radiative effects on evaporation is challenging. For example, ECS measurements of E over the Eshkol Reservoir and Lake Kinneret in northern Israel (Figure 1), have shown that while net radiation peaked at noon, as expected, E peaked in the afternoon with the arrival of the Mediterranean Sea breeze front (Assouline & Mahrer, 1993, 1996; Tanny et al., 2008). The Dead Sea (Figure 1) offers a unique opportunity to resolve that, since in the summer the Mediterranean Sea breeze reaches the lake at sunset (Alpert et al., 1982; Hecht & Gertman, 2003; Lensky & Dayan, 2012). During daytime, when radiation is high, the wind velocity is low, whereas during nighttime when solar radiation is null, the Mediterranean Sea breeze front arrives and wind velocity peaks. Most previous studies on evaporation from water bodies reported an evaporation diurnal course with a single peak, associated with the diurnal wind speed and radiation peaks (McGloin et al., 2014; McJannet et al., 2011, 2013). To the best of our knowledge, only Sene et al. (1991) reported several peaks in the diurnal evaporation, which were associated with wind speed peaks.

Details are in the caption following the image

Maps of the study area including measuring stations: ER, Eshkol (Netofa) Reservoir; LK, Lake Kinneret; DS, Dead Sea. (a) Elevation model (blue are water surfaces) and (b) the fronts of the Mediterranean Sea breeze, displayed with labels of the local time (data from 13 July 2015).

The Dead Sea is a hypersaline terminal lake, at the lowest point of the globe (430 m below sea level). Evaporation rate estimates from such a location are interesting because of the extreme conditions: hypersalinity of the water and hyperaridity of the climate (∼60 mm annual rainfall). The high salinity of the Dead Sea brine reduces water activity, from 1.0 (for freshwater), to ∼0.65. This reduces the water vapor pressure by 0.65 compared to freshwater at the same temperature, and thus reduces evaporation by a similar factor (Salhotra et al., 1985). Thus despite the hyperarid conditions, evaporation rates over the Dead Sea, estimated between ∼1.1 m·yr−1 (Lensky et al., 2005) and ∼0.94 m·yr−1 (Kottmeier et al., 2016), are reduced compared to the neighboring fresh water Lake Kinneret (∼1.5 m·yr−1, respectively). Accurate E estimates from the Dead Sea are required because of its impact on the water budget of the lake and the resulting dynamics of its water level (Lensky & Dente, 2015). The latter was shown to have a great environmental impact as it is related to the occurrence of sinkholes on the shore (Abelson et al., 2003; Yechieli et al., 2002) and stream incision (Ben Moshe et al., 2008). E estimates are also required to evaluate the efficiency of the planned “Red Sea Dead Sea Canal” which aims in regulating the lake water level (Gavrieli et al., 2011). Previous estimates of E from the Dead Sea involved indirect methods of energy and mass budget and mass transfer method (Lensky et al., 2005; Neumann, 1958; Stanhill, 1985, 1990, 1994, Steinhorn, 1991, 1997) and measurement of E from pans that were located in the southern basin of the Dead Sea (Alpert et al., 1997; Salhotra et al., 1985). Yet none of these papers reported the diurnal course of E, which is critical to explore the modulation due to radiation and sea breeze.

The main objective of this study is to measure directly the diurnal course of E in the Dead Sea, by means of ECS, and to present new observations and insights on the dynamics of E from the Dead Sea in response to the heating by solar radiation and wind speed. The specific goals are: (i) characterize the diurnal course of evaporation rate from the Dead Sea water surface; (ii) relate the observed evaporation rate diurnal dynamics to the atmospheric diurnal forcing, i.e., net radiation and wind speed; and (iii) compare the diurnal course of E from the Dead Sea to that from two freshwater bodies, Eshkol (Netofa) Reservoir (ER) and Lake Kinneret (LK), along the propagation path of the sea breeze front (Figure 1).

2 Methodology

2.1 The Study Sites

Figure 1a presents the study area, including the location of three water bodies that were equipped with ECS: Eshkol Reservoir, Lake Kinneret, and the Dead Sea. These stations are situated at an increasing distance from the Mediterranean coast, and thus the sea breeze front typically reaches each station at a different hour during the day.

The Dead Sea is located in the southern part of the Jordan Rift Valley, between the Judean Mountains in the West and the Moab Mountains in the East. It is located ∼100 km east of the Mediterranean coast. It has a surface area of ∼625 km2, a maximal depth of 290 m and an actual water surface level at −430 m . The Dead Sea water surface is steadily declining at a rate of ∼1 m·yr−1, due to water losses by evaporation and diversion of water inflows from the surrounding watersheds. The Dead Sea is a hypersaline lake with brine density ∼1.24·103 kg m−3, saturated to halite (Arnon et al., 2014, 2016; Sirota et al., 2016, 2017).

Lake Kinneret (also known as the Sea of Galilee; 32°50'N 35°50'E; −212 a.m.s.l.) is situated in the upper Galilee (northeastern Israel), in the central part of the Jordan Rift Valley. It has an area of 166 km2, a maximum depth of 40 m, and a semiarid climate with ∼400 mm annual rainfall. The lake is located ∼60 km east of the Mediterranean coast. The microclimatic conditions at the site, the components of the lake water and energy budgets, and the characteristics of the evaporation process from the water surface were reported in previous studies (Assouline, 1993; Assouline & Mahrer, 1993, 1996; Rimmer et al., 2009; Shilo et al., 2015).

The Eshkol (Netofa) reservoir is located in the Bet-Netofa valley in northern Israel (32°46'N; 35°15'E, 145 a.m.s.l), a site characterized by a Mediterranean climate (mean annual rainfall of ∼480 mm). It is a square reservoir with a 600 m side and 3.5 m depth, located ∼30 km east of the Mediterranean coast, acting as a settling reservoir of the National Water Carrier. Evaporation from the reservoir was investigated by Tanny et al. (2008) and Assouline et al. (2008, 2016).

2.2 The Mediterranean Sea Breeze Front Propagation

The Mediterranean Sea Breeze is a mesoscale thermally induced phenomenon, most pronounced in the summer season when the land-sea temperature difference is the largest and the large-scale winds (synoptic conditions) are weaker (Lensky & Dayan, 2012).

Figure 1b presents the propagation of the Mediterranean Sea Breeze front in the study area, based on two indicators. The first is retrieved using time series of wind speed data from meteorological stations. The second depicts the timing of the peak thermal anomaly, i.e., the cooling of surface temperature associated with the sea breeze front as detected by a geostationary satellite, as described in Lensky and Dayan (2012). Animation of the Sea Breeze propagation is presented in the supporting information SI-1.

Figure 2 presents the diurnal course of solar radiation and wind speed measured at the three sites. The incoming solar radiation is uniform throughout the study area while the wind speed distribution varies significantly with the distance from the Mediterranean coast. Eshkol Reservoir and Lake Kinneret are located 30–60 km east of the coastline; hence, the peak of the sea breeze is reaching them at about 13:00–15:00 (LT, i.e., - GMT + 2). The Dead Sea, however, is located ∼100 km off the Mediterranean coast, and the Sea breeze front arrives ∼7 h after leaving the Mediterranean coast, at about 19:00 (LT).

Details are in the caption following the image

Solar radiation and wind speed at the three waterbodies, average diurnal courses (10–16 July 2015, standard deviation is indicated). Local time (GMT + 2) is indicated in the x axis. Note that the peak velocity of the sea breeze reaches the Dead Sea around midnight, whereas it arrives at the Eshkol Reservoir and Lake Kinneret early in the afternoon. The solar radiation in the three stations is similar; data presented here are from the Dead Sea (red curve).

2.3 Eddy Covariance and Meteorological Stations

We present here data from three sites. Measurement systems at the Dead Sea are presented below. The detailed description of measurements at Eshkol Reservoir is presented in Tanny et al. (2008) and that of Lake Kinneret in Assouline and Mahrer (1993). At the Eshkol Reservoir, the EC station was located in the center of the reservoir (Figure 3c). At Lake Kinneret, the station was located in the lake, 200 m offshore.

Details are in the caption following the image

Measurements setup in the Dead Sea. The Eddy covariance station at the waterline at (a) Mishmar and (d) the offshore meteorological station based on the EG100 buoy. (b) Location map of the Dead Sea with elevation model (elevation contours 50 m interval, elevation is indicated in m above sea level); current coastline, bold contour. (c) The ECS station at Eshkol Reservoir is also presented.

2.3.1 Eddy Covariance Coastal Station: Dead Sea

The ECS station is located at the tip of a narrow peninsula, termed here “Mishmar Station” (MS), in the southern part of the Dead Sea (Figure 3b), exposed to the lake water from azimuth 310° to 210°, clockwise. The dominant winds during summer are northerly, thus the station is measuring wind from the open water surface. The station's footprint (see Appendix Appendix A) is out of the local influence of freshwater discharges, which are concentrated along the northern shore (mainly Jordan River and the Ein Feshkha springs), at least 30 km to the north. Such discharge locally reduces the surface water salinity and temperature near the shore. The station is located ∼10 m off the shoreline, on a very smooth and flat area, covered by halite salt crust, indicating that the brine is halite saturated and no water emerges in the area. The ECS was installed at the top of a 4.5 m height tripod, ∼5 m above the water surface. The system is equipped with a 3-D Sonic Anemometer (WindMasterPro, Gill, UK) and a Li-Cor Open Path CO2/H2O Gas Analyzer (LI-7500A, Li-Cor, USA). The high frequency data (20 Hz) are stored in a Li-Cor data logger and half-hourly mean fluxes are calculated on site using the EddyPro software by a SMARTFLUX module (Li-Cor, USA). The footprint of most of the data is from the open lake direction (excluding 3% of the measurements that reached from the land, see Appendix Appendix A).

Meteorological sensors (response time of seconds) were located on the same tripod, and their output was stored on a data logger (CR1000, Campbell Scientific, USA). The sensors used are a 2-D Sonic Anemometer (WindSonic, Gill, UK) located at 4.5 m above ground; and an air temperature and relative humidity sensor (HC2S3, Campbell Scientific, USA) located 2 m above ground in a radiation shield.

2.3.2 Dead Sea Open Lake Hydrometeorological Buoy (EG100)

The second station at the Dead-Sea is a hydrometeorological buoy (Figures 3b and 3d) located ∼5.0 km offshore, which therefore, represents well the conditions on the open lake. The offshore data are used here for two purposes: (i) to evaluate the differences in the meteorological conditions of the open lake station and the onshore ECS station, and (ii) to provide measurements of net radiation and of the rate of heat storage in the water, typical of the Dead Sea water body. The station is equipped with a data logger (CR1000, Campbell Scientific, USA) and meteorological sensors similar to those on the shore station: A 2-D Sonic Anemometer, air temperature and relative humidity sensor, both located at 4.5 m above the water surface. Net radiometer (Kipp & Zonen CNR4), installed at ∼1.5 m above the lake water surface at the edge of a ∼2.5 m long horizontal arm (see also Nehorai et al., 2009, 2013a, 2013b). We added an automatic daily washing system to wash sea spray and salt precipitants off the 4 domes on the net radiometer. A thermistor chain of 19 sensors (12 Aanderaa, and 7 SBE39) covering the top 40 m of the water column is measuring the water temperature profile every 20 min.

2.3.3 Similarity Between the Coastal and Open Lake Dead Sea Measurements

Figure 4 presents the time series of measured meteorological data in the open lake (EG100) and coastal (Mishmar) stations. The two stations (Figure 4a) measured almost identical wind speed and direction, relative humidity, and air temperature as shown in Figures 4c–4f, respectively. The dominant wind is blowing from EG100 to the coastal station (azimuth 10°, Figure 4b), thus the majority of the ECS data recorded at the MS coastal station arrives from the open sea, without land “contamination.” This data set suggests that the ECS of the coastal station well represents evaporation from the open lake.

Details are in the caption following the image

Meteorological conditions of the open lake (EG100, green) and coastal (Mishmar [MS], purple) stations. (a) Map of the stations, (b) wind rose diagram, the dominant northerly winds blow from the open lake station into the coastal Mishmar station (measurements from Mishmar), (c, d) wind direction and speed, (e, f) air relative humidity and temperature. Note the high similarity between the two stations (the coefficients of determination between the stations data are: urn:x-wiley:00431397:media:wrcr23045:wrcr23045-math-0001 = 0.95, urn:x-wiley:00431397:media:wrcr23045:wrcr23045-math-0002 = 0.87, urn:x-wiley:00431397:media:wrcr23045:wrcr23045-math-0003 = 0.85).

3 Results and Discussion

3.1 Heat Fluxes Over the Dead Sea Surface

3.1.1 Diurnal Evaporation Rate From the Dead Sea: A Distinct Double Peak Induced by Solar Radiation and Night Sea Breeze

Evaporation rate and corresponding meteorological data from the Dead Sea are presented here for a period of 7 days (10–16 July 2015, DOY 191–197) which represent the summer period in this region. Figure 5 presents time series of measured atmospheric forcing components and the resulted evaporation rate. The net radiation presents diurnal cycles peaking at noontime at ∼900 W·m−2 and a minimum nighttime value of ∼ −100 W·m−2. Surface water temperature (measured at 1 m water depth) peaks about 3 h later. The wind speed is relatively low during the day, intensifying from ∼18:00 local time (LT, GMT + 2) as the sea breeze reaches the Dead Sea and peaking (8–10 m·s−1) at night, 8–12 h after the net radiation maxima. As a result, the measured E shows a unique diurnal cycle characterized by two peaks. The first peak appears when surface water temperature peaks, ∼3 h after the net radiation maximum, while the second peak appears at night, when the wind is at its maximum velocity. The maximum E during the observation period reached ∼0.37 mm·h−1. The two minima of E appear at sunrise and sunset, when both wind speed and net radiation are at minimum. The mean daily evaporation rate for the 7 days presented is 0.16 mm h−1 (i.e., 3.84 mm·d−1). In Figure 5, evaporation rates that are higher than the hourly average value (0.16 mm·h−1) are colored; blue stands for evaporation rate at high wind speed and low radiation, and red stands for evaporation rates during daytime when radiation and surface temperature are high and wind speed is relatively low.

Details are in the caption following the image

Evaporation rate (lower panel) and atmospheric forcing (net radiation and wind speed; upper panel) measured at the Dead Sea during typical summer conditions. Surface water temperature (Tw 1 m, dashed) peaks 2–4 h after radiation peak. E values higher than the average (0.16 mm·h−1) are colored: blue fill relates to the E peak at high wind and low radiation, and red fill relates to the E peak when radiation (and Tw) is high and wind speed is low.

Figure 6 presents the average diurnal course (for the 7 days presented in Figure 5) of the atmospheric forcing (net radiation and wind speed) and the corresponding evaporation response characterized by two diurnal peaks (“double peak”). The maximal mean Rn value is 850 W·m−2 and the maximal mean wind speed is ∼7.5 m·s−1. Interestingly, for these respective values a similar mean maximal E rate of ∼0.23 mm·h−1 is measured, suggesting that net radiation and wind speed equally contribute to evaporation under these conditions.

Details are in the caption following the image

The 7 day averaged diurnal course of (a) net radiation and surface water temperature (Rn and Tw 1 m), (b) wind speed, and (c) evaporation rate, for the period 10–16 July during summer 2015 (Figure 5).

3.1.2 The Water Heat Storage

The Dead Sea is stratified during summer, with a warmer epilimnion (Arnon et al., 2016; Gertman & Hecht, 2002). Figure 7a presents the temperature profile time series, measured along the thermistor chain, showing the clear distinction between the warm epilimnion (∼33°C, upper ∼25 m) and the cooler hypolimnion (∼24°C, lower ∼25 m). In Figure 7b, the epilimnion's upper 5 m are showing daily warming, following the solar radiation input whereas the lower layer (30–40 m) is hardly affected by such input. The water heat storage, HG (Figure 7c), is calculated as the integral of temperature with depth from the lake's surface (srf) to the thermocline depth (thm):
urn:x-wiley:00431397:media:wrcr23045:wrcr23045-math-0004(1)
Details are in the caption following the image

The heat storage in the water column. Time series of (a) depth profiles of temperature, (b) temperature at different depths (depths in meters are indicated in the right); note the distinction between epilimnetic warm water and hypolimnetic cooler water, (c) water heat storage (HG), and (d) the rate of change of water heat storage ( urn:x-wiley:00431397:media:wrcr23045:wrcr23045-math-0005).

Since the thermocline depth fluctuates within a few hours period due to internal waves (Arnon et al., 2014; Sirkes et al., 1997), we integrate the temperature profile from the surface down to 25 m deep, where the depth interval of 15–25 m was vertically uniform, best presented by the 20 m thermistor which was not influenced by the upper edge of the thermocline fluctuations.

The rate of change of heat water, heat storage per horizontal unit area, G, is:
urn:x-wiley:00431397:media:wrcr23045:wrcr23045-math-0006(2)

Figure 7d shows a time series of G, showing daily cycle with daytime warming (positive derivative) and nighttime cooling (negative derivative values during late night). The average G of the entire 7 day period is positive, 105 W/m2, as can be identified by the increasing water temperature in the epilimnion (Figures 7a and 7b) and the positive slope of HG (Figure 7c). The uncertainty in determining the average depth of thermocline is about 2 m (around the 25 m), resulting in G values of 113 and 98 W·m−2 for the deeper and shallower thermocline depths, respectively. The temperature below the thermocline was nearly constant during the observation period, meaning that there is no measurable heat flux along the temperature gradient (see more on diapycnal fluxes in the Dead Sea in Arnon et al. (2016)).

3.1.3 The Surface Energy Budget of the Dead Sea

The measured heat fluxes are presented in Figure 8. Figures 8a and 8b show the 7 day time series whereas Figures 8c and 8d show the corresponding average diurnal courses of the same 7 days. The net radiation (Rn) during daytime is mostly converted to the rate of water heat storage (G) with additional dissipation to latent and sensible heat fluxes (LE and H, respectively), LE being larger than H. During nighttime G is negative; heat is being dissipated mostly by LE corresponding to the second evaporation peak induced by the incoming sea breeze (see Figure 6), and by negative net radiation (long wave radiation emitted from the lake surface is larger than the atmospheric input). The rate of water heat storage diurnal cycle has a time lag of a couple of hours after net radiation. The sensible heat flux is always positive. On average, LE is more than 7 times higher than H (average Bowen ratio is 0.14), but daily fluctuations are significant, with LE minima that sometimes could be slightly lower than H.

Details are in the caption following the image

Measured components of the energy budget of the Dead Sea: (a) net radiation and the rate of water heat storage (Rn and G, respectively), (b) sensible and latent heat fluxes (H and LE, respectively). (c, d) Daily averages of Figures 8a and 8b, respectively. The error bars in Figures 8c and 8d represent the standard deviation; note the relatively large error bars of G.

Average values of the fluxes presented in Figure 8 were (in W·m−2): Rn = 225, G = 105 ± 8 (depending on thermocline depth), LE = 108, H = 15. The measured energy balance closure, (LE + H) / (Rn – G), varies between 0.97 and 1.10 for the daily averages, which verifies the data quality. The momentary heat budget closure cannot be regarded as reliable measurement due to the large error of the momentary estimate of G (Figure 8c). As shown in Figure 7c, the daily average of G is much more robust than the momentary estimate of G (Figure 7d). These large momentary variations of G may result from local variations in the temperature profile due to several processes, such as turbulence in the waterbody and internal waves (e.g., Arnon et al., 2014). Previous studies concluded that determination of G from thermistor chains and determination of energy budget closure are reliable over a period longer than one day (e.g., Tanny et al., 2008).

3.2 Evaporation From Water Bodies Along the Mediterranean Sea Breeze Propagation

Figure 9 presents the summer diurnal cycle of measured E and atmospheric forcing from three water bodies along the propagation of the sea-breeze: the Eshkol Reservoir (ER), Lake Kinneret (LK), and the Dead Sea (DS). Evaporation rate over Eshkol Reservoir peaks during the afternoon (∼4 PM, ∼0.5 mm h−1), synchronous with the wind speed peak (∼8 m·s−1). Evaporation rate minimum values (∼0.15 mm h−1) persist during nighttime, from about an hour after sunset when wind speed decreases to its minimum value (∼2 m·s−1) until about 2 h after sunrise. Lake Kinneret, located eastward to Eshkol Reservoir, shows similar behavior, with a more abrupt change in wind speed, due to the rapid descent of the sea breeze from the upper Galilee mountains (400 a.m.s.l.) into the lake area (–200 a.m.s.l.) (Mahrer & Assouline, 1993). The evaporation responds quite closely to the temporal distribution of wind speed. At these two sites, the solar radiation partially overlaps with the wind speed distribution and E seems to rise due to the coinciding effects of wind speed radiation.

Details are in the caption following the image

Diurnal atmospheric forcing (net radiation in grey and wind speed in black with colored circles) and the corresponding evaporation rates (colored shaded area) in the (a) Eshkol Reservoir (Tanny et al., 2008) colored in blue, (b) Lake Kinneret (Mahrer & Assouline, 1993) colored in green, and (c) the Dead Sea (this study) colored in red. Observation period, ER, 2–10 and 13–17 September 2005; LK, 24 September 1990; DS, 10–16 July 2015. (d) Map of the Mediterranean Sea breeze front propagation, displayed with labels of the local time (see Figure 1b). Note the increasing time lag between radiation peak and wind speed peak down the MSB course.

In the Dead Sea, located further off the Mediterranean coast, the Sea breeze arrives after sunset, and high wind velocities (>5 m·s−1) persist during nighttime, decreasing only before sunrise. A peak in evaporation rate was measured around midnight that coincides with the peak of wind speed. However, another peak was measured in the afternoon, following the net radiation peak when wind speed is relatively low. This particular feature of the wind speed in the Dead Sea allows revealing the specific contribution of the radiative component of the climatic forcing on the evaporation process. In the Dead Sea, during the experimental period, the relative contribution of the radiative forcing appears to be quite similar to that of the aerodynamic one, but they peak at different hours during the day. Sene et al. (1991) observed three evaporation peaks in a tropical lake; however, in their study the peaks were associated with an unusual wind speed distribution, with three wind speed peaks, that occurred on a certain day. This is different from the present observation of double peak in the Dead Sea, which is a result of a consistent summer pattern of the Mediterranean Sea Breeze.

For the periods presented in Figure 9, the average evaporation rate over the Dead Sea is about two thirds of that over each of the two other waterbodies. This is despite the fact that the Dead Sea area is much more arid than the two other locations, and characterized by a higher net radiation and a sea breeze of equivalent peak velocity (note that the measurements at the three sites shown in Figure 9 are not synchronous). Previous estimates of E from the Dead Sea (Kottmeier et al., 2016; Lensky et al., 2005) and Lake Kinneret (Assouline, 1993; Mekorot Watershed Unit, 2008), showed that Dead Sea E is ∼25% lower than in Lake Kinneret (0.95–1.1 and ∼1.5 m·yr−1, respectively). This is attributed to the much higher salinity of the Dead Sea brine than Lake Kinneret water, which reduces the water vapor pressure deficit through the reduced water activity (Lensky et al., 2005; Salhotra et al., 1985, 1987). This difference in evaporation rates is attributed to the extreme salinity of the Dead Sea brine (∼340 g·kg−1), compared to the low salinity of the two freshwater bodies (∼0.6 g·kg−1).

4 Summary

This study presented direct measurements of evaporation rates and climatic variables in summer over the Dead Sea at a high temporal resolution by means of ECS, enabling to characterize the dynamic variation of these variables during the day. The following insights were gained:
  1. A distinct evaporation double peak characterizes the diurnal course in summer.
  2. The early afternoon peak is related to the radiative heat supply and corresponds, with a short delay, to the peak in net radiation, whereas the night peak is concomitant with the peak in wind speed.
  3. During the observation period (10–16 July 2015), the amplitudes of the two peaks in a mean diurnal evaporation rate cycle were similar, indicating an equal contribution of the radiative and the aerodynamic components to evaporation during that period.
  4. The diurnal course of evaporation at the regional scale of Israel varies with the distance from the Mediterranean coast (Figure 9):

    1. Close to the coastal area, as seen for the Eshkol Reservoir and Lake Kinneret, the sea breeze front arrives early in the afternoon, and wind speed and net radiation temporal distributions partially overlap. Accordingly, only a single daily evaporation rate peak appears in the afternoon, reflecting the combined contributions of the aerodynamic and radiation forcing.

    2. At about one hundred kilometers inland, at the location of the Dead Sea, the sea breeze front arrives after sunset, and wind speed peaks at night, thus inducing a second peak of evaporation rate at night.

Acknowledgments

We thank the three anonymous reviewers and the associated editor for insightful comments that improved this manuscript. The measuring setup required a significant collaboration with the following great teams: The Geological Survey of Israel—Ali Arnon, Hallel Lutzky, Raanan Bodzin, Ido Sirota, Ziv Mor, Itzik Hamdani, Haggai Eyal, and Assaf Mor. IOLR—Tal Ozer and Boris Katsenelson. Taglit R/V—Silvy Gonen, Meir Yifrach, and Shachar Gan-El. Meteo-Tech—Denis Kuchuk and Igor. The research was funded by the Israeli Government under GSI DS project 40572. This study is a contribution to the PALEX project “Paleohydrology and Extreme Floods from the Dead Sea ICDP core”, funded by the DFG (grant no. BR2208/13-1/-2). Interested readers can access our data at: https://zenodo.org/record/1118297#.Wjj5PVWWY6Q.

    Appendix A: ECS Data Quality Checks

    The EddyPro (Li-Cor) software calculates the half-hourly fluxes using the 20 Hz raw data. The software performs a series of corrections on the raw data, as indicated by (Burba, 2013). Figure A1a presents the latent and sensible heat fluxes (LE and H, respectively) for the 7 days under study. The EddyPro reports a quality check flag for the calculated flux ( Figure A1b): 0, good data; 1, intermediate; 2, noisy data. Only 5% of the data was graded “noisy” (flag 2), regarded here as invalid data and was not included in the analysis.

    Another quality check is the “CO2 signal strength,” which is a measure for the cleanness of the IRGA optical path. During the observation period the CO2 signal strength was high (>92%, Figure A1c), meaning that the IRGA was clean and data are valid.

    The flux footprint of the ECS data, as calculated by EddyPro using the following models (Kljun et al., 2004; Kormann & Meixner, 2001), is presented in Figure A1d, where the distance of 90 and 10% of the flux contribution are presented. The model used is presented in Figure 11e with the initials of the authors KM and KJ. Figure 11e presents the corresponding 30 min average wind direction. As noted above, the prevailing wind direction is from the open lake; only 3% of the data were excluded since wind arrived from the land.

    Details are in the caption following the image

    Time series of (a) mesured EC fluxes, (b) QC, (c) CO2 signal strength, (d) footprint, (e) footprint model, and (f) wind direction.