From: Andrew Tupper

Hi all,

Fred Prata (CSIRO) has done some dust analyses on the MODIS images - with Fred's permission I've put the split-window analysis from the images that Jason refers to below on the NT internal page at:
http://www.nt.bom.gov.au/ntregion/volcano/training/MT231002-0045.gif  (Local copy here)

The split-window method is used by NTRO to analyse volcanic ash and (in clear skies) works well with AVHRR, GMS-5 and MODIS.  However it is also great for dust storms.  Spring dust storms are a *huge* topic in Asia because of the worsening situation in China.  There are many recent papers about the sources, properties, and detection of Asian dust - for example
there's a good one by Xuan and Sokolik in Atmospheric Environment 36 (2002) 4863-4876 .  There's lots of AVHRR & GMS split-window images of Asian Dust at:

http://www-sci.edu.kagoshima-u.ac.jp/sing/kosa/kosa-e.htm
('kosa' is the Japanese term for Asian dust)

Anyway, it may be worthwhile for whoever is writing up these events to look at some of the papers relating to Asian dust, and to do lots of split-window satellite analysis!  GMS should be able to track the development and dispersion of the dust clouds during the day and night very well.

cheers Andrew Tupper

Harald Richter

What is the essence of a "split-window analysis", i.e. how does it work?
>
>Most curiously, Harald

Andrew Tupper
Subject: [synoptic_discussion] Re: Split Window / MODIS plug

Dear Harald,

I apologise for using jargon in my first E-mail.  The split-window or 'reverse-absorption' algorithm uses the different absorption properties of volcanic ash (or mineral dust, in this case) and water cloud in the 11 & 12 micron infrared channels on NOAA/AVHRR or GMS-5/VISSR, to discriminate between dust and cloud during the day or night.

There's various bits of info about it on the web (eg http://www.geo.mtu.edu/volcanoes/vc_web/ ) and many examples (some on the Darwin Volcanic Ash Advisory Centre web site, www.bom.gov.au/info/vaac ) or on the internal site at:
http://www.nt.bom.gov.au/ntregion/volcano/training/gms.2002.09.1540.jpg
for a really good example from the recent Ruang eruption.  Or you can collar  Rod Potts or Beth Ebert or anybody from Satellite Section to ask them about it....

The main problems with the algorithm are overlying clouds, overshooting  CB  tops causing false alarms in the tropics, or water in the dust/ash cloud.  However for dust storms over a dry continent this is much less of a  problem than for volcanic ash detection in the tropics.

<start_MODIS_plug>
The exciting thing about MODIS is that it has 250 metre resolution visible  channels, and lots of extra 1 km resolution channels including one at 8.6 microns, which is really good for detecting ash/sulphates from eruptions,  and should also be excellent for mineral dust.  With two satellites now  carrying the MODIS sensor the data is relatively frequent (every six hours for most locations) and can be analysed using McIDAS....  so using this together with GMS and AVHRR should be a great way to get a handle on real-time dust monitoring.... when we get the hardware to receive it in  real time!  In the meantime, NASA have the MODIS data available for download in 300 MB (!!) chunks at their archive centre (http://daac.gsfc.nasa.gov), so it's at least good for dust case studies.
<end_MODIS_plug>

cheers Andrew