FAQ – Product Choice, Accuracy, and Datums

Which product should I select?

Products can have a 10-day or 35-day temporal resolution (i.e., revisit time). For the 10-day products end users have a choice of either the TPJO.1 or TPJO.2 products.

TPJO.1 products are based on a long-term datum – a datum that has been formed from utilizing 9years of smoothed height variations from the Topex/Posiedon mission (1993-2001). These products are utilized by USDA to form the Lake Status Map which shows the current status of the lake compared to the 1993-2001 mean. This helps to identify those lakes experiencing more longer-term hydrological drought conditions. Combined with Jason-1 (2002-2008) and Jason-2 (post 2008) measurements, these TPJO.1 products potentially span 25years and in essence are providing a new climatic index based on water level variations.

The TPJO.2 products are based on a more recent datum - one that has been formed from lake height measurements on a single Jason-2/OSTM overpass, i.e. on one day. This select date is variable between the lakes, i.e. each datum is unique to a particular lake. This product was introduced because not all lakes have a long-term record, there are many which have only valid Jason-2/OSTM data.

We suggest using TPJO.1 for longer-term climatic research, though if both products exist for a lake, TPJO.2 may be (slightly) more accurate with respect to the datum formation method.

How accurate are the time series of lake level variations?

Validation of the lake products is performed via comparison with ground-based gauge data and case studies are often performed on lakes and reservoirs in North America. Selecting a gauge site closest to the satellite overpass, mean daily stage variations are compared to the altimetric height variations. Because the measurement datums differ, the altimetric series is shifted on the y-axis (a constant offset) to achieve a best match. The standard deviation of the mean offset is then recorded and used as an estimate of the overall error on the altimetric time series.

This standard deviation value can vary depending on i) the size of the lake, ii) the complexity of the surrounding topography, iii) the roughness of the lake surface (altimetric range resolution) and iv) knowledge of the various corrections that are performed to obtain a single altimetric height.

Typical values can range from,

3-5cm for the largest “open” (wind roughened) lakes (such as the Great Lakes in the USA) or 15-20 cm for more smaller or more sheltered lakes or >20cm for reservoirs situated in steep valleys and/or where the instrument only observes a narrow expanse of water.

In addition to the fact that a single altimetric height value on the lake surface is actually an average value within the instrument footprint (which can be a few hundred meters to several kilometers in diameter, depending on surface roughness) the satellite technique relies on averaging height measurements across the lake surface to reduce noise. The validation exercises are thus comparing a spot height from a gauge at a set geographic location, with averaged satellite measurements, both within the footprint and across the lake surface.

Validation examples can be found in the published literature.

How can the lake products be converted to a different datum?

The individual satellite elevation values (geodetic heights) are given with respect to a defined reference ellipsoid datum with Equatorial radius = 6378136.3 meters and Flattening = 1/298.257. This datum is the standard for most altimetry missions and was introduced during the Topex/Poseidon mission. During the product creation process this datum is removed as we substitute a “reference profile” which is either i) a height profile across the lake on a given day (for the TPJO.2 products) or ii) a mean height profile across the lake (for the TPJO.1 “lake status” products) based on 9years of Topex/Poseidon measurements 1993-2002. Each of these reference profiles is unique to the lake, and acts as its new datum. Each satellite overpass height profile is then compared to the reference profile, a mean height difference is recorded, and a time series of RELATIVE height variations is constructed.

a) LOCAL DATUM? Translation to a local datum is only possible if the local datum has been historically tied into an established global datum at some point in time, or if there is a partial overlap in time between satellite and gauge data set and so an estimated mean difference in elevation can be applied.

b) WGS84 Reference Ellipsoid Datum ?WGS84 is essentially a reference ellipsoid datum using the same axes as the satellite reference ellipsoid, but radius and flattening coefficients differ. End users wishing to translate from the G-REALM product datum to the WGS84 ellipsoid datum (but still in a geodetic not orthometric frame) should,
  • From the product header, Subtract the “WGS84 mean shift” value from the “Mean geodetic lake height” and then ADD the resulting value to the G-REALM product heights
c) MEAN SEA LEVEL? (To convert the G-REALM product height variations into a mean sea level (or orthometric) frame, the geoid height along the satellite ground track has to be considered. EGM2008 and EIGEN6C4 are two well-known geoid models, but others exist in the research domain. However, all will have uncertainty at a given location, and here we must consider a profile across the lake, not a single geographical location.
  • To convert to an orthometric (mean sea level) datum, end users should apply one of the “Conversion Factors” given in the product header. Currently the choice is EIGEN6C4, EGM2008, or the experimental v052615A.

For each lake, the original reference profile, the shift to the WGS84 datum, and the geoid height profiles can be viewed by selecting the link in the left-hand margin.
The G-REALM team welcomes feedback on these conversions.


For lake 0012.Winnipeg,

  • A) To convert to the WGS84 geodetic frame apply (185.73-0.71) =185.02m to the G-REALM product height values.
  • B) To convert to the EIGEN6C4 orthometric (mean sea level) frame apply 218.01m to the G-REALM product height values.

Using the information below in the product header,


c       Mean lake height along the Reference Profile:

c       (i.e. geodetic height wrt Topex ellipsoid)

c       mean = 185.73 m std = 0.74 m



c      Shift from Topex ellipsoid to WGS84 ellipsoid:

c       mean = 0.71 m



c       Geoid height along the Reference Profile:

c       v052615A mean = -32.99 m std = 0.73 m

c       EGM2008 mean = -33.00 m std = 0.73 m

c       EIGEN6C4 mean = -32.99 m std = 0.72 m

c       EGM2008 mean = -33.00 m std = 0.73 m




c      (to shift from satellite product datum to an orthometric/mean sea level datum)

c       (ADD this value to each elevation in the lake product)

c       v052615A = 218.01 m std = 0.06 m

c       EGM2008 = 218.02 m std = 0.06 m

c       EIGEN6C4 = 218.01 m std = 0.06 m


Close Window