djhoese
commented
11 years ago After looking into this it's hard to say if there is anything different or if it's just the difference between how the tools show the data. I can't remember atm, but I'm pretty sure it also had something to do with WMS not getting the negative radiances. By that I mean, the polar2grid version generating the binaries to be put through McIDAS ignores negative radiances which causes it to look worse.
There has always been "grainy" DNB images during this time of the year (I think?). The new DNB scaling (local equalization) should help to reduce this graininess once it is perfected, if not already.
I'm creating this because bugs need to be documented, not magically fixed.
Russ Dengel pointed out that images he is putting into his WMS server look very grainy. He is using DNB binary files provided by Kathy on the production system. These file are the binary files produced by fornav, but before they go to rescaling. Russ is doing his own manipulations using McIDAS to go from binary to geotiff that he is providing to the WMS server as the SSEC.
@evas-ssec and I have looked at a few test cases. One to note is the 211e grid case on 2013/01/08 07:14. @wstraka3 believes the angle and luminosity/phase of the moon is causing the graininess. We believe this is causing the majority of the graininess, but comparing our results with Russ', we find his to be much grainier; to the point that some features are lost in the image quality.
@evas-ssec can correct me and @wstraka3 has stated this before, the graininess is not new. It has always been a problem. The key point with Russ' issue is that it is worse and as far as we can tell it only shows up in WMS. I have created 100 dpi and 500 dpi versions of images created from the binary files Russ should be getting and the AWIPS NetCDF files. The 100 dpi images look grainy, but this is mostly due to the quality of the image, the 500 dpi images look fine/normal. I used the most recent development version of polar2grid and made images using negative radiance values and ignoring them. The version that ignores negative radiances looks much grainier.
We should verify that the binary images Russ is getting are from an older (<1.0.0) version of polar2grid. I believe what Russ is seeing is not necessarily a bug in the sense of seeing something that wasn't there before, it still isn't perfect, but this is more of a DNB scaling problem that we are aware of. I believe if Russ uses a newer binary file it will look better. I will try to get him one later today.