Investigations on Aral Sea Regressions from Mirabilite Deposits and Remote Sensing

Abstract

Remote sensing techniques including radar (Topex/Poseidon, Jason-1 and Envisat) and laser altimetry (Icesat), and moderate resolution spectro-radiometer (MODIS) images, are used to estimated current level and surface extent time variations of the Aral Sea. During the Holocene several phases of regression occurred, leading to desiccation of the Aral Sea. During the last 50 years, Aral Sea has drastically shrunk due to intense use of river’s water for irrigation purposes. It is currently separated into four distinct water bodies, namely, the Small Aral in the North, the Tchebas Bay in the North West, and the South West and the South East basins. The Kulandy strait connected the SW and SE basins until very recent times. These basins are now almost separated and salinity becomes very high (140–180 g/l) in the Eastern part. Rubanov discovered past deposits of mirabilite in the years 1970–1980. We investigate the significance of these deposits in the light of current evolution of the four water bodies that constitute the heritage of Aral Sea contemporary desiccation. Using remote sensing techniques, we have attempted to calculate the water balance of south Aral Sea during the last 3 years. We conclude in strong probability that the Kulandy strait carries water most of the time from the Eastern Basin to the Western Basin. We have demonstrated that it should have been the same process in the past to explain the Mirabilite deposit, but unfortunately, due to recent artificial water monitoring of the Aral Sea (dam in the Berg’s strait, new reservoirs in the Amu Darya’s delta), it is impossible to make definitive conclusion from actual Aral Sea water balance.

Appendix

1.1 Remote Sensing Monitoring of Aral Sea

The satellite altimetry technique has been developed in the early 1970s with the launch of Seasat (1978). Initially designed to study the oceans, it has also been a successful technique for monitoring continental surface water, such as inland seas, lakes (Crétaux and Birkett 2006), and rivers, (Calmant and Seyler 2006). It primarily measures the surface water level of water bodies in a terrestrial reference frame with a return time varying from 10 to 35 days depending on the orbit cycle of the satellite, with fairly good accuracy (a few centimetres over large bodies’ such as Lakes to tens of centimetres over river stems). The concept of the satellite altimetry measurement is rather straightforward. The onboard radar altimeter transmits a short pulse of microwave radiation towards the nadir. Part of the incident radiation is bounced back to the altimeter, providing distance between water surface and the satellite position, which is then transformed to the instantaneous water height above a reference fixed surface, a geoid model for instance.

We have used the radar altimetry data of different satellites that all passed over the Aral Sea: Envisat, Jason and GFO. We also used laser altimetry data gained by the ICESat satellite that NASA launched on 12 January 2003. It operates the first Lidar onboard a satellite, namely a dual beam laser (GLAS), which provides water surface with precision never reached before (Zwally et al. 2003) but which unfortunately operates over continental lakes only 2–3 times per year, hence limiting the potential uses of this technique for lake level monitoring on high time resolution. It however provides height of the Aral Sea, independent from radar altimetry and in situ measurements. Altimetry data are distributed relatively to the GRS80 ellipsoid of the WGS84 global system. To make a levelling of the radar and laser altimetry height over the Aral Sea in a reference frame consistent with the bathymetry of the lake’s basin that is available with respect to the Russian geoid and that used to estimated water volume variation used in water balance calculation, we also used geodetic levelling measurements made by P. Zavialov, from 2002 to 2007 almost twice a year (Zavialov, personal communication) that are referred to the same geoidal reference as the bathymetry. These 3 sources of information were used to calculate the Big Aral sea level variations from 2002 to 2007 on a monthly basis, for WB and EB, respectively, in order to detect possible differential level variations between both basins (Fig. 7), and to calculate the water balance of WB. More detailed applications of radar altimetry for Aral Sea can be found in Crétaux et al. (2005). From Fig. 7, a lot of information could be derived. All sources of data are very coherent, providing the same rate of water level decreasing over the period of study. This information is crucial for the further water balance calculation, which will rely on these data. We can also see that current level of big Aral Sea is now below 29 m, which is the elevation of the Kulandy Straits. This indicates that probably EB and WB are close to being physically separated, at least during the dry season in Summer–Autumn. Finally, due to the fact that WB is almost not fed anymore by Amu Darya river (except from some very small river input from the Sudoche reservoirs, P. Micklin and N. Aladin, personal communication), and that EB still receives some water from Eastern branch of Amu Darya Delta, some underground seepage, and also from release of water in the Berg’s strait when dam’s gates are opened (generally in Spring) it is probable that EB and WB will evolve in a different way in the near future.

We also used another type of remote sensing data to confirm this scenario. The Moderate Resolution Imaging Spectro-radiometer (MODIS) was launched in December 1999 on the polar orbiting Terra spacecraft and since February 2000 has been acquiring daily global data in 36 spectral bands with spatial resolution of 250 and 500 m. The surface reflectance product that we have used is defined as the reflectance that would be measured at the land surface if there were no atmosphere. It provides information on the type of surface, which reflects the incident solar energy. Classification methods developed by Crétaux et al. (2008) enable monitoring the water extent of the Aral Sea. Here, we present 2 images that are used to point out and confirm results deduced from altimetry. First image (Fig. 5) displays the Aral Sea basin in May 2005. We have chosen this date because, at that time, we know that the Amu Darya River released more water discharge to Aral Sea than anytime during the last 10 years (P. Micklin, personal communication). Even though one can observe that the Eastern Branch of Amu Delta is the only water channel currently reaching Big Aral Basin (small feeding from Sudoche Reservoir in the western Amu delta is however still possible but certainly remains very small (a release of the Sudoche’s dam occurred in 2006 producing probable inflow to the WB). This conducts to the conclusion that WB basin is principally fed by precipitation (around 10 cm/yr) and eventual outpouring from the Kulandy strait (see the discussion on water balance computation, Sect. VII. 2). The second image (Fig. 6) was taken in October 2007, at the end of the dry season. It is clear from this image that the process of EB and WB separation has started, which is coherent with results obtained by altimetry data. Note that using a high-resolution Satellite Image, P. Zavialov observed that a very narrow channel remains open in the Kulandy Strait, not wider than 200 m, smaller than the resolution of the Modis images we used (P. Zavialov, personal communication).

Fig. 5

Modis image in May 2005: Black colour represents open water. The Amu discharge is marked on the Eastern branch of the delta, and inexistent on the Western branch. This image was taken at the maximum of the 2005 flooding on the Amu River. It proves that WB is quite isolated from Amu River at the present time

Fig. 6

Modis image in October 2007. Black colour represents open water. The Kulandy strait is now very narrow

1.2 Current Water Balance of Western Big Aral Basin

To assess the direction of the flow in the Kulandy Strait, we have calculated the water balance of the WB. Indeed, it is clear from remote sensing that the components of the water balance in the WB are limited to Evaporation and Precipitation rates, underground inflow, and to the water runoff in or out of the WB through the Kulandy Strait (the releases of Amu Darya to the Big Aral is now diverted to EB with high probability). Conversely, to calculate the water balance of EB we should take into account the different sources of water releases, which are not actually well known; exact water discharge from Amu Darya, outpouring from the Small Aral when the gates of the dam are opened, underground inflow and possible unknown discharge from Syr Darya Eastern branch.

The water balance of WB is simply given by the following equation:

$$ {{{\text{d}}V} \mathord{\left/ {\vphantom {{{\text{d}}V} {{\text{d}}t}}} \right. \kern-\nulldelimiterspace} {{\text{d}}t}} = \left( {P - E} \right){\text{S}}\left( t \right) + {\text{Rks}} + {\text{Gi}} $$

(1)

where Rks is the runoff in the Kulandy strait, P is precipitation, E is evaporation, S(t) the surface at the time t, Gi is possible underground inflow and dV/dt is the variation of volume in time of the WB. On this equation we neglect some possible water inflow from the dam of the Sudoche reservoir in the WB. From satellite Modis imagery, this did not occur in 2005, but probably in 2006 it has been operated in order to release this reservoir. The exact quantity of water that reached Aral sea WB is however not known (Miklin, personal communication).

We have estimated the variation of volume, by combining the altimetry level variation (Fig. 7) with a precise digitised bathymetry of the Aral Sea basin (see Crétaux et al. 2005 for more details). As far as evaporation is concerned, we used the model given in (Bendhun and Renard 2004), with absolute value of 1,160 mm/yr. For precipitation input, we used data from Bortnik 1999 also widely discussed in Crétaux et al. (2005), the variability of which is about 100 mm/yr depending on wet and dry year as measured over the 90s with some in situ precipitation data collected around the Aral Sea (from the Aral Sea web site: www.cawater-info.net/index_e.htm, the precipitation measured near Aral Sea until the end over the 90 s was between 40 and 170 mm/yr depending on the year). Possible inter-annual variability and trend of evaporation has been discussed in Small et al. 2001. They have estimated the change of lake surface temperature and E – P term of the water balance due to global warming and regional climate effect connected to the desiccation of the Aral Sea. Their main result is that between 1960 and the mid 1990s the net effect of the Aral Sea desiccation is to decrease (in average) the P – E by 40 mm/yr, while the net effect of global warming is to decrease it by 100 mm/yr. They also estimated SST variation due to global warming, and they conclude that it has increased by few degrees. From the Aral sea web site, where monthly air temperature and humidity are given over the 90 s, the general observation is that inter-annual variability which could provoke change in evaporation rate is small and the general trend confirms the conclusions of Small et al. (2001), that there is probably a continuous increase of Evaporation over the last 10–15 years. The value used in our calculation for Evaporation rate is hence probably underestimated. To take into account possible evaporation inter-annual changes, we considered that the total uncertainty on the value used for our calculation of P – E is about 200 mm/year. For the calculation of underground water, the issue is more complicated which led us to some simplifications. We have calculated the water balance over the Big Aral in the year 1996 and 1998, because during those 2 years, WB and EB formed a unique basin and the presence of a dam in the Berg’s strait at the same period allowed to annoy possible release from Syr Darya River. The equation used is:

$$ {{{\text{d}}V} \mathord{\left/ {\vphantom {{{\text{d}}V} {{\text{d}}t}}} \right. \kern-\nulldelimiterspace} {{\text{d}}t}} = \left( {P - E} \right){\text{S}}\left( t \right) + \left( {1 - {{\upalpha}}} \right){\text{Rsd}} + \left( {1 - {{\upbeta}}} \right){\text{Rad}} + {\text{Rks}} + {\text{Gi}} $$

(2)

Fig. 7

Level variations of EB and WB from radar and laser altimetry and from levelling data provided by P. Zavialov. Note that the last measurement of P. Zavialov (diamond in blue) in November 2007 was done during an intense storm, which certainly shifted this measurement and is hence not realistic to the mean lake level (P. Zavialov, personal communication). Coherence between laser and radar, collected at different dates over each month, are much more pronounced

where α and β are the percentage of water lost in the delta of Syr Darya and Amu Darya, respectively, Rad the river discharge from Amu Darya, and the other terms of the equation as defined in (1). From Aladin et al. 2005, α was estimated to 0.2, so we had to solve for the β and the Gi parameters with water balance of year 1996 and 1998 from (2). For β we obtained 0.1 and for Gi 2 km³. The calculation here is obviously limited to the small period used, and also to the fact that we assumed a constant value for each of them, which is not evident. From those results and the in situ observations listed above, we have then calculated annual values of the term Rks from (1). The results are listed in Table 2.

Table 2 Water balance of the WB of Big Aral for the years 2005 to 2007 (Rks is the volume of water passing through the Kulandy strait each year)

From runoff data on Amu Darya and Syr Darya one can also calculate the water balance of both Small Aral and South Eastern Aral to estimate the Runoff in Kulandy strait. It is given by Eq. 2:

$$ {{{\text{d}}V_{\text{t}} } \mathord{\left/ {\vphantom {{{\text{d}}V_{\text{t}} } {{\text{d}}t}}} \right. \kern-\nulldelimiterspace} {{\text{d}}t}} = \left( {P - E} \right){\text{S}}_{\text{t}} \left( t \right) - {\text{Rks}} + \left( {1 - {{\upbeta}}} \right){\text{Rad}} + \left( {1 - {{\upalpha}}} \right){\text{Rsd}} + {\text{Gi}} $$

(3)

Here S_t(t) is the total surface of Small and SE Aral, and dV _t/dt the total volume variations (the other terms in (3) are the same as that defined in (1) and (2)). As data on both Amu Darya and Syr Darya were available only for the year 2005 we have solved the water balance given by (3) to estimate Rks for 2005. The runoff of Syr Darya and Amu Darya were, respectively, 10.7 and 9.5 km³ in 2005, variation of volume of Small Aral and SE Aral was +2.7 km³, average surface 11,200 km³, we hence have obtained from Eq. 3, a value of 4.5 ± 2.5 km³ for the term Rks, which is in quite good agreement with the value obtained in 2005 from the water balance of WB (Table 2).

The results listed in Table 2 are unambiguous: when there is high inflow from the rivers, Rks is significantly positive, which indicates a Eastward flow in the Kulandy Strait in good agreement with altimetry results stating that nowadays, the WB starts to shrink a little bit quicker than the EB. Due to the differential level of both basins, in spring and early summer, there is probably a release of water from East to West, until level equilibrium between both basins is reached, after which (in autumn), the direction of the water flow mainly depends on wind stress. This conclusion is reinforced by the fact that we consider the total underground water in WB (Gi) as those calculated for the whole Big Aral by (2), so we even probably over-estimated this term in (1), with consequently under-estimation of the Rks term.

However, the uncertainty associated to the in situ data, and the fact that actually Small and Big Aral water balance are also influenced by artificial control through the dams, it is currently not possible to give definitive conclusion on the direction of the flow in the Kulandy strait at the time of the Mirabilite deposit found by Rubanov from contemporary water balance, but such a scenario is realistic.