Lead Investigators

Paul Davis, Deborah Estrin, Richard Guy, Allen Husker

Overview

The field deployment in Mexico of the CENS Seismic Application continued this year. The sites have had a number of problems that required a field engineer from the US to be stationed in Mexico for the duration of the project. The field management was done from UCLA through the success of the radio communications. That is we were able to determine most site problems from logging into the network and probing the sites. Despite being able to manage the network and fix sites within a week of becoming aware of a problem, we were unable to keep up with the problems. As a result some of the potential data has been lost. A paper is being written about the field installation and how to do a similar deployment for Seismologists.

Preliminary data analysis is quite promising. There are many proposed projects within the 3 universities collaborating in Mexico: UCLA, Caltech, and UNAM. The specific project being analyzed by UCLA researchers is to determine position of the subducting slab through velocity tomography of the slab. Current position analysis suggests that the slab has most likely broken as proposed by a previous geochemical study by Ferrari (2004). A paper is being written analyzing this data.

Approach

The proof of concept of the CENS Wirelessly Linked Seismic Network (WLSN) has worked quite well. The management of WLSN has been done almost entirely from the US while being deployed in Mexico. Unfortunately, many problems have lead to significant loss of data. These problems include: (1) Hardware failures; (2) Software bugs; (3) Weather related failures; and (4) Poor design of the CDCC for disk swapping. An example of data gaps are shown below for the months of July through September 2006 (Fig 1-4).

(1) The most significant hardware failure is the Stargate looses communication with the SMC radio and is no longer accessible. The station has to be reset in order to regain communication. A fix of automatic reboot daily has been applied to those stations that exhibit the problem. The problem only appears on certain stations. The problem appears to have no connection to temperature. The time between reboot and loss of communication occurs from a month to within a few hours and is CDCC dependent.

The flash disks are starting to become corrupted at an increasing rate. It is unclear why. Both WLSN and the Caltech sites are experiencing this problem. Many times they can be reformat remotely.

The CDCC often times need to be reflashed. The boot sequence becomes odd leading to failure or the Ethernet connection is damaged or other failures. We have had to reflash about 20% of the Stargates deployed.

(2) The most significant software bug was only in the Cuernavaca line as this is where the Disruption Tolerant Shell (DTS) has been deployed. When a file reached 100% transfer, if there was a delay on the last 1% the file would reset to 0% and start sending again. The purpose of DTS is for use in Challenged Networks. As Cuernavaca is such a network, there were constant delays and so nearly no data was transferred through the repeater at CUCE due to this being the most challenged location. Thus, all 8 sites behind it suffered huge losses of data. The bug was fixed a year after deployment of DTS. DTS’ current bug is that the final 2 stations quit transferring data and need to be reset. Thus, all data transfer is blocked unless DTS is reset by hand. Apart from this, DTS appears to work as it should.

The software watchdogs did not get deployed on all of the stand-alone sites immediately. Notably, the disk-space watchdog clears up un-used processes that bog-down the Stargate and eventually kill data collection was not installed on CIRE, MIMO, or ARBO for some time.

(3) The biggest failures due to weather was flooding and the subsequent failure of the solar panel charge controller. All parts of the system typically survive flooding as long as water can drain. However, the controller dies every time. We have had to replace about 20 charge controllers.

Wind moving antennas out of alignment has been rather minor. It has only effected 3 sites.

Lightning struck near one site, killing the CDCC.

(4) The disk swapping of the CDCC requires transfer of configuration files from and to the correction locations. In addition, with few 4GB cards, the data from the cards is downloaded directly to a laptop in the field many times. The complicated process of copying data to and from the correct folder leads to copy the wrong folder or deleting the incorrect data after data is copied. Often times the cards to not mount correctly as well leading to longer field time.

 

Figure 1: Cuernavaca Line data stored.

 

Figure 2: Pachuca Line data stored.

 

Figure 3: Huejutla Line data stored. Note PLSA was permanently uninstalled during this period.

 

Figure 4: Stanalone sites data stored.

 

The data collected thus far has shown that there is a fast structure at depth and that if it is connected to the surface subhorizontal slab it is dipping at an incredibly steep angle of 65° (Fig 5). It is unlikely that the slab is diving at such a steep angle given the buoyancy of such a young slab. Moreover, the poor depth resolution (Fig 6) reveals that if there is a break it is currently not seen in our images. There for it has most likely torn away from the upper slab as suggested by Ferrari (2004).

 

Figure 5: The y-axis is the depth in km of the study area. The x-axis is the distance along the strike of the line in km.  The color axis is the difference in slowness (s/km) with respect to a 1d background velocity model. The blue or fast part, is dipping at a 65° gradient.

 

Figure 6: The inversion done with a ‘checker board’ starting modeling for the purpose of resolution testing. The y-axis is the depth in km of the study area. The x-axis is the distance along the strike of the line in km.  The color axis is the difference in slowness (s/km) with respect to a 1d background velocity model. The left figure has a 100 km X 150 km starting model. The right figure has a 100 km X 100 km starting model. The resolution with depth is quite poor with 100km depth resolution.

Future Directions

The stations will be uninstalled in the months of March and May. Three papers will be submitted for publication as part of Allen Husker’s Ph.D.

People  

Faculty:

• Paul Davis, ESS, UCLA

Staff:

• Richard Guy, CENS, UCLA
• Igor Stubailo, CENS, UCLA
• Irving Flores, CENS, UCLA

Graduate Students:

• Allen Husker, UCLA
• Luis Antonio Domínguez, UNAM

External Research Partnerships

Caltech (current)
UNAM (current)