Can I get the time-mean mass fields that are removed from the GRACE Tellus fields?
We get this question a lot. Our answer is always similar: the time-mean fields that are removed in the prosessing represent essentially the Earth's mean gravity field. As such, there isn't really a 'time-mean mass field', in particular not in terms of average water storage height. Keep in mind that GRACE is a system that measures the time-variable gravity component, which we then take and interpret to represent the near-surface land water storage or ocean bottom pressure variations. Other effects (e.g., earthquakes) might also be present in the data and need to be corrected for.
GRACE cannot directly provide the time-invariant water storage components (e.g., groundwater and aquifer absolute storage).
Will GRACE data be useful to study the aquifers in an area of 83,600 sq km? (a real user question)
GRACE resolves scales of approximately 300x300km, which is about the size of this aquifer. At this small scale, one has to carefully evaluate possible ‘signal-leakage’ from surrounding areas, as well as carefully assess signal damping. All that can depend quite heavily on the particular case at hand (e.g., local geography, surrounding areas etc). So while GRACE is certainly ‘useful’ here, errors and uncertainties are larger at these small spatial scales (compared to larger-scale spatial averages), and it’s thus prudent to carefully assess the signal characteristics.
Why 'GRACE Tellus'?
Tellus refers to the Roman goddess of the planet Earth. In English it offers a wordplay, so we can ask "What can GRACE, and time changes in gravitational acceleration TELL US about our changing planet?" Incidentally, Tellus was also "a citizen of ancient Athens who was thought to be the happiest of men." (Source: Wikipedia)
Mass anomalies? Mass changes? Monthly mass change rates? Negative mass values? I'm confused ... Help!!
First-time users often wonder what exactly GRACE measures, and what the gridded values of positive and negative mass (in units of equivalent-water-height) mean.
In a nutshell:
- GRACE measures Earth's gravity field every month;
- to convert this to surface mass changes, we need to subtract a 'Mean Field' (a.k.a. 'Static Field') - this mean can be thought of as a spatially-varying-but-constant-in-time reference field;
- a mean field is essentially a convention: since the gravity field is always slightly changing due to many processes, a mean field is typically referenced to an epoch (i.e., a time-period); All GRACE Level-3 data sets that we provide have information in the header as to which time-mean or reference field has been removed (look for 'time_mean_removed')
- after the mean field is subtracted, the remaining signals vary around zero: there are mass gains (positive change), and mass losses (negative change);
- Units: surface mass change is mass/area, and since most of the temporal changes are related to water, we divide mass/area with water density (1000 kg/m^3), which yields the commonly used unit of 'equivalent-water-height' (meters or centimeters);
- Thus, GRACE does not measure total/absolute water storage; rather, think of it as an accurate scale that can detect and quantify when mass changes, but we typically don’t know where ‘the zero’ is.
- there are some instances where we can infer absolute values; for example, GRACE has been used to measure total snow-water-equivalent (SWE) in high northern latitudes. This works because at the end of boreal summer, there is no snow on the ground, which effectivbely represents the 'zero' of the scale (of course, this approach implies there are no other mass changes than snow accumulation during the winter).
And lastly: 'mass anomalies' vs 'mass rates' vs 'monthly mass changes' ....
- 'mass anomlies': deviation from a time-mean or static reference field (i.e., the mass in the bucket);
- 'mass rates': the temporal derivative of mass anomalies (i.e., the flow rate into or out of the bucket);
- 'monthly mass change': somewhat ambiguous, but mostly used with the meaning of 'mass rate';
Is it necessary to apply scaling factors to derive groundwater storage anomaly from GRACE?
The scaling factors are necessary to compare (or combine) GRACE TWS with model data, such as GLDAS. This procedure aims at making the data combination ‘apples-to-apples’, in the sense that it restores some signal amplitudes lost in GRACE due to smoothing and post-processing. Groundwater (GW) storage can then be derived by GW = scaled_GRACE - GLDAS. However, there are some important caveats: foremost, the scaling factors are optimized for seasonal signals, and in that mostly for soil_mositure and snow. If groundwater changes have a significantly different spatial pattern than the near-surface water changes (which the scaling factors are based on), the GW signal estimated as above is potenitally biased. By how much exactly depends on the specific case at hand and likely requires some sensitivity kernel analysis. Please see this paper for details and an actual case study: http://onlinelibrary.wiley.com/doi/10.1029/2011WR011453/full.
Does the baseline time (e.g., 2004-2009) include all months, or is it different for each month?
The monthly anomalies are typically computed relative to a time-mean baseline that includes all months. That way, the anomalies show the full seasonal cycle and any other longer-than-monthly variations. A monthly varying baseline (the average for each calendar month over several years) is called a ‘climatology’ (a.k.a. average annual cycle). We do not remove a climatology as this signal is of interest to many users and can easily be removed by end-users if they require it for their application.
In other words: the baseline we remove consists of a constant value for each grid point. If your application requires a different baseline (e.g., instead of 2004-2009 you need 2005-2010), you can simply compute that by averaging each grid point over the 2005-2010 baseline, and subtract that value from all time steps.