NASA & the Philosophy of the Tec Student: Failure is Not an Option

Cyber-Physical Hyper-Spectral Systems: Present and Future

Dust Plumes at Chihuahuan Desert and East-Lubbock/East-Midland Nov 6, 2011

Dust Plumes at Chihuahuan Desert and East-Lubbock/East-Midland Nov 6, 2011

As predicted by some folks, dust arrived to El Paso, TX. This was mainly over the Chihuahuan desert with winds and dust moving from south-west to north-east.
Also, some dust plumes could be observed on the east region of Lubbock, TX, and east Midland, TX, with winds and dust blowing from south to north.

Here is the true color NASA MODIS - Terra image:

Here is the output of my algorithm for dust aerosols detection:

If we use the GOES information, we can barely see any activity on the infra-red data. This is GOES East:

This is GOES West:

Have any comments or other pictures, please let me know.

White dust in El Paso, Texas, coming from White Sands, New Mexico - Nov/2/2011

Dear reader, I ran my super (maybe not so) smart algorithm to detect dust aerosols, I got some information, but I got other stuff that is a little confusing, so I need some of your input. Here is the story as reported by the US Weather service...

"It is a dusty day across El Paso -- especially east of the Franklin Mountains. Gusty north winds have lofted gypsum from White Sands National Monument, and the plume has spread south down the Tularosa Basin. The plume can be seen in Visible Satellite..."

This is the information I got from the web...
1) Aline Jaime's facebook photo (a friend of a friend)

2) Then, this from the US National Wheather Service for El Paso
https://www.facebook.com/US.NationalWeatherService.ElPaso.gov

3) The "true color" image from NASA - Terra - MODIS data at 1740 UTC

4) This the output of my algorithm to detect dust aerosols (soon to be fully published if my advisor is happy with the text):

5) Other view of the event using Hao's aerosol index:

QUESTIONS for the courageous reader:
a) Assume that the "white" area circled on the center is the reported event in White Sands, NM and El Paso, TX. What is that thing on the right? It looks like a wave of dust, but I didn't hear any reports. That area corresponds to Midland, TX, and Odesa, TX.
b) What is that on the left, right in south California? Did anyone reported any dust in California?
-----------Update Nov/3/2011----------
Recent reports show that there was "dusty" activity on the areas of Midland Texas as confirmed by the following NASA GOES videos. It seems like some sort of wave from north to south. Click play to see the Infra-Red GOES West and East data animation.

-------Update Nov/4/2011 ------------
The following is additional information that can be found on the internet.
All these information confirms a Haboob in the Midland - Odesa, TX, area.

"If you go to the RAMSDIS LW difference page found on:
http://rammb.cira.colostate.edu/ramsdis/online/sounder.asp
and choose the "archive" from both the East and West, you can see the
event in West Texas quite clearly.  Because of the poor resolution of the
sounder, the much smaller White Sands event is not so clear.  That "event" in the top of the Sea of Cortez is a feature that quite often shows up; but, because I almost never see any "plume" development, I think that it must be an artifact of the underlying surface or just a lot of "haze" as that is just below the Imperial Valley, Mexicali, and Yuma agricultural region.  On that day, we had dust in the Mesilla Valley but it was not too thick and does not show up well in the LW difference imagery -- LW difference from either MODIS or AVHRR might show it better (as well as these other events)." - M. P. Bleiweiss

 This was taken from a plane, unknown author, yet.

A YouTube video of a concerned citizen on the Odesa - Midland, TX, area.

This video confirms the observation on the ground.

What is a Haboob anyway, and what's the difference between it and a dust storm?
Well, in case you were wondering, here is what an expert on the area answered when asked this question.

"All haboobs are dust storms, but not all dust storms are haboobs. Sort of like the saying that the Eskimos have thirteen different words for snow. And some old-timers I met up in the Panhandle differentiate between a "twister" and a "tornado."
A haboob is a stark, well-defined, cloud of dust, advancing as a wall- like an avalanche coming down a snowy slope or an oncoming flash flood (or like when you spill your coffee on a table top and a wave of liquid flows out), a massive wave of oncoming dust, where the dust is confined to a dense layer of onrushing air near the ground. Contrast to the "garden variety" dust storm where the dust rises up from the ground over a large area and it just gets hazier and hazier and the dust sort of thins out steadily as you go upwards in the air, without the discrete, stark "wall" of dust.
Nowadays, the haboob-type dust storms are usually caused by turbulent downdrafts spreading away from thunderstorms (those ones in Phoenix this past summer were that type). They are dramatic, but relatively small in size. Haboob-type "walls" of dust associated with the typical cold-season dry cold fronts spreading down the Plains had been almost unheard of for at least 20-25 years, until the last few weeks... BUT this type of "non-thunderstorm haboob" is EXACTLY the class of dust storm that was so prevalent during the Dust Bowl days of the 1930s. I'm not sure exactly what that means... but it can't be good. (Perhaps not surprising though, given that much of west Texas is in a drought WORSE than that of the 1930s)." - Dr. Thomas Gill

Algorithms for training large-scale linear programming support vector regression and classification

Abstract

The main contribution of this dissertation is the development of a method to train a Support Vector Regression (SVR) model for the large-scale case where the number of training samples supersedes the computational resources. The proposed scheme consists of posing the SVR problem entirely as a Linear Programming (LP) problem and on the development of a sequential optimization method based on variables decomposition, constraints decomposition, and the use of primal-dual interior point methods. Experimental results demonstrate that the proposed approach has comparable performance with other SV-based classifiers. Particularly, experiments demonstrate that as the problem size increases, the sparser the solution becomes, and more computational efficiency can be gained in comparison with other methods. To reduce the LP-SVR training time, a method is developed that takes advantage of the fact that the support vectors (SVs) are likely to lie on the convex hull of each class. The algorithm uses the Mahalanobis distance from the class sample mean in order to rank each sample in the training set; then the samples with the largest distances are used as part of the initial working set. Experimental results show a reduction in the total training time as well as a significant decrease in the total iterations percentage. Results also suggest that using the speedup strategy, the SVs are found earlier in the learning process. Also, this research introduces a method to find the set of LP-SVR hyper-parameters; experimental results show that the algorithm provides hyper-parameters that minimize an estimate of the true test generalization error. Finally, the SVR scheme shows state-of-the-art performance in various applications such as power load prediction forecasting, texture-based image segmentation, and classification of remotely sensed imagery. This demonstrates that the proposed learning scheme and the LP-SVR model are robust and efficient when compared with other methodologies for large-scale problems.

Buy my dissertation here

But if it seems too expensive for you ($37 the PDF and $59 the soft-cover) you may contact me for a free copy, i.e., if you want to torture yourself in 335 pages, just let me know.

"A Signal Processing Love Letter

I want you to love me like the impulse response of an unstable recursive filter.
I want a sampling interval that will be enough just for you.
I want a new sampling theorem that will sample only you without losses and aliases all the others.
I want a quantizer that will quantize only you without any losses.
I want you to be a single tone signal so that I can filter out all the others and keep you.
I want a non-linear filter that will select you and reject the others.
I want you to be the pitch of my voice so that each time I speak people will recognize you.
I want you to be pixels in my eyes so that each time someone looks me in the eye will see you.
I want to discover a new FFT that will transform any image into yours.
I want a new JPEG technique that will compress only you without losses.
I want a special decoder that will produce you out of any coded signal.
I want you to be my cepstrum.
I want a fast processor that can decode you in zero time.
I want you to keep me in your ROM memory.
I want a data bus that will only transfer your bits."

by Hussian Al-Ahmad
-----
This quote was taken from an IEEE magazine check the last page of the following article:

Farrell, D.; Oakley, A.; Lyons, R.; , "Discrete-time quadrature FM detection," Signal Processing Magazine, IEEE, vol.22, no.5, pp. 145- 149, Sept. 2005
doi: 10.1109/MSP.2005.1511836
URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=1511836&isnumber=32367