Satellite altimetry is an efficient tool to monitor the level of enclosed seas, lakes and even large rivers as long as the satellite flies over them.
The Issyk-kul Lake is a great salted mountain lake in Central Asia, in Kyrgyzstan. Its name means “warm lake” since it never freezes. It is located around 77.40°E and 42.20°N.
Contrary to the “Aral Sea” Data Use Case, we will not use the altimetric height in order to plot its sea level variations along the years. For this example on the Issyk-kul Lake, we directly use the altimetry waveforms which are the records of the reflected echoes over time. From this waveform, several parameters are computed using several algorithms (retrackers) which describe its shape (leading edge width, trailing edge, epoch at mid-height, skewness, etc.) and provide the satellite-ground range.
We use here the OSTM/Jason-2 SIGDR which can be obtained from Aviso and contain the 20 Hz along-track waveform information. OSTM/Jason-2 offers the opportunity to use an algorithm enabling a better tracking over lake or over non-ocean surfaces. Its altimeter is equipped with an open-loop tracking technique for which a Digital Elevation Model (DEM) has been developed. The altimeter’s onboard memory contains the elevation values of areas overlayed by the ground tracks. These data, combined with DORIS, are used to position the receiving window in advance, in order to anticipate the contrasts of the topography and to give priority to measurements over water. This one corresponds to the DIODE/DEM tracker. During the OSTM/Jason-2 validation phase, two others trackers were used: the Split Gate Tracker (SGT) and the Median Tracker. The SGT tracker is equivalent to the Poseidon-2 tracker on Jason-1. The Median tracker works on the barycenter of the waveform shape and improves data availability on coastal zones (water/land transitions) and on continental water areas (lakes).
We only use the three first cycles of OSTM/Jason-2 (during its validation phase):
- from 2008/07/12 to 2008/07/21 for OSTM/Jason-2 cycle 001 and corresponding to the SGT tracker,
- from 2008/07/21 to 2008/07/31 for OSTM/Jason-2 cycle 002 and corresponding to the Median tracker
- and from 2008/07/31 to 2008/08/10 for OSTM/Jason-2 cycle 003 and corresponding to the DIODE/DEM tracker.
To select the right track, you can have a look at the pass locator using Google Earth available through Aviso. Thus, we can directly download the pass #131 we wish to work on (and we do not need the whole cycles of OSTM/Jason-2 data) on the “Datasets” tab. After that, in the “Operations” tab we limit the geographical boundaries around the lake from the latitudes: 41°N – 45°N, so a 4°-segment.
In the “Datasets” tab, we have selected the three first OSTM/Jason-2 cycles, each in a different dataset: Datasets_wvf_c001, Datasets_wvf_c002 and Datasets_wvf_c003.
First, in the “Operations” tab, create three operations. From the native OSTM/Jason-2 SIGDR files, create one operation per cycle as follows: select the latitude at 20 Hz that is to say the field “lat_20hz” for “X”, and select in the “Data”, the field “waveforms_20hz_ku” which contains the 104 Ku-band waveform samples for each waveform.
In the “Selection criteria” expression, the geographical boundaries are limited as follows: is_bounded(41, lat_20hz, 45). Execute this first operation for a given cycle from the native OSTM/Jason-2 SIGDR files.
In the “View” tabs, you can choose between two different kind of graphs:
- Y=F(X) which plots the waveforms individually and represents the return echo power in function of time.
- Z=F(X,Y) which plots a set of cumulated waveforms in function of latitude, as if seen from above.
For the Y=F(X) graph, you have just to select the operation and execute it.
In order to properly plot the Z=F(X,Y) graph, a second step is necessary. The Broadview Radar Altimetry Toolbox fits the pixels of this kind of figures to span the space between two values. However, when there are gaps in the values, this leads to extending artificially some data over those gaps. So, to have a figure with gaps correctly plotted, we first have to compute the date, then instead of plotting the Data expression directly from the Operation result, we use the NetCDF file thus created by the Broadview Radar Altimetry Toolbox in a new dataset.
You can select in the “Dataset” tab this NetCDF output, from the location where you have saved your workspace on your computer, in the “Operations” folder (named CreateOperations_wvf_c001.nc). This file contains three different fields that we integrate again into a new operation. Be careful to give a new, meaningful, operation name mentioning the NetCDF format. Select the field “lat_20hz” for “X”, the “wvf_ind” for “Y” and the “waveforms_20hz_ku_c001” for the “Data”. No expression is described here, but a new resolution is given according to the total number of waveforms for one-degree of latitude. Since one-degree of latitude is about 111 km, and since the ground distance covered by OSTM/Jason-2 during 1/20 s, is about 0.350 km, the step for the X resolution is 0.350/111 = 0.00315. This last value must be filled in the “Operations” tab, by clicking on the “Set Resolution filter” button. The number of intervals is computed and gives the total number of waveforms for your pass segment.
In the “Views” tab, select this last operation made from the BRAT NetCDF output file and execute it. The same steps are performed for others cycles by duplicating the operations.
Results and comments
On the first kind of graph, Y=F(X), each waveform can be seen either individually, by browsing one by one, or through an animation, displaying all the selected waveforms. This animation cannot be saved, but each frame/waveform can be. We have selected and saved 5 waveforms per cycle, which are the closest ones to a given location from one cycle to another (altimetry measurements are not taken exactly at the same point from one cycle to another). We observe many waveform shapes along the lake: Brown echoes, specular, multipeaked, etc (see From radar pulse to altimetry measurements). But the waveform shapes are also quite different for the same location, due to the kind of tracker used or due to a change of surface conditions between the cycles.
On the second kind of graph, Z=F(X,Y), we observe a set of juxtaposed waveforms as a function of latitude, as if seen from above. The segment pass is covering approximatively 520 km, from 41°N to 45°N, so a long segment before and after the lake.
The altimetric signal is not present all over the segment, there are a lot of missing values (white areas on the graph) corresponding to the mountainous land. The signal loss is different in function of the tracker used : the Median tracker (middle) seems to be locked and gives a signal more often.
The Southern boundary of the Lake Issyk-kul is detected for the three trackers at the same time, around 42.15°N. But the Median tracker (middle) seems to give a longer tracking over (or beyond) the Northern boundary of the lake.
Over the lake, the thermal noise of the waveforms corresponds to black areas. The leading edge position corresponds to the black/dark blue transition and follows the thermal noise. The leading edge position varies a lot over the lake for the trackers SGT and Median: a difference of 20 gates can be seen for the Median tracker, corresponding to a range difference of almost 10 m. Such discrepancies are corrected using the adapted retracking algorithm.
Another feature can be underlined with the red area measured over the lake by all the three trackers and at the same time, around 42.7°N. This feature might correpond to a highly reflecting surface nearly the Northern coast.
fig 2.Set of juxtaposed waveformsas a function of latitude (41°N – 45°N, pass #131) and as seen from above for the first three cycles of OSTM/Jason-2. Each cycle corresponds to a different tracker: SGT for cycle 001, Median tracker for cycle 002 and Diode+MNT for cycle 003.