What is the difference between smoothed and unsmoothed data in the AE_L2A data set, and how should I use them together?
Each frequency is looking at a different size footprint. Each frequency has its own feedhorn, and is thus susceptible to independent pointing errors. The most important benefit of resampling (or smoothing) is to create a suite of frequencies that are all looking at the same scene. So, when the higher-resolution channels are resampled to match the footprints of the lower-resolution channels, the data are smoothed.
In the not-resampled (unsmoothed) channels (denoted by 'o' in Table 5):
- The unsmoothed 23.8 GHz channel does not spatially correlate with any other channels, and should generally not be used in conjunction with any other frequencies
- The 18.7, 36.5, and 89.0 GHz channels are not resampled to their own footprints, because it is intended that they be used in their native (not-resampled) form
- The 89A and 89B GHz channels are not aligned with the lower-resolution channels, by design
In the resampled channels (denoted by ‘•’ in Table 5):
-Pointing errors in the 6.9 and 10.7 GHz channels led to resampling of these channels to their own footprints in order to line their boresights up (or spatially correlate) with the higher-resolution channels
- The 23.8 GHz channel is resampled to spatially correlate with the 7.9, 10.7, and 18.7 GHz channels (it was determined that the footprint size of the 23.8 GHz channel was close enough to that of 18.7 that it did not warrant a suite of channels spatially correlated to the native 23.8 GHz footprint)
- For 18.7 and 36.5, each higher frequency is resampled to correlate with those native footprints
The unsmoothed and smoothed channels (see Table 5) are useful for constructing sets of “spatially correlated” channels, such as:
Resolution 1: 6.9, 10.7, 18.7, 23.8, 36.5, 89.0
Resolution 2: 10.7, 18.7, 23.8, 36.5, 89.0
Resolution 3: (18.7), 23.8, 36.5, 89.0
Resolution 4: (36.5), 89.0
Resolution 5: (89.0)
where, () = unsmoothed
Over homogeneous areas of ocean, the different size footprint observed by each frequency is not a big problem. But if there is heterogeneous meteorology, each channel may be looking at different amounts of cloud, rain, land, etc.
Start with the lowest frequency (largest footprint) that is important to any particular application.
For example, 10.7 GHz wind speed retrieval. Then ask, what would 18.7 GHz say if it were looking at the exact same patch of earth as this 10.7 GHz observation? And how bright would this exact same patch of earth (“scene”) be at 36.5 GHz?
The unsmoothed 18.7 and 36.5 GHz channels do not answer these questions directly, but there are more than enough observations to construct “virtual” observations which do “measure” the same scene. These are the smoothed channels.