UNIDATA is an NSF funded program in the Atmospheric Sciences. Although the specific goals of UNIDATA are different from IRIS´s, one of their main missions is to provide data to the atmospheric community, similar to the IRIS DMS mission of providing data to the seismological community. UNIDATA has developed the Internet Data Distribution (IDD) and Local Data Management (LDM) systems to meet this need. Developed for atmospheric sciences, the technology has been easily adapted to seismological data.
The general concept is similar to a pyramid scheme. Any participating node within the IDD structure can inject products into the system. These products are forwarded to a small number of other locations running the IDD/LDM system and those in turn forward it to other nodes, and so on and so on. It should be clear that just having four layers of participating centers each forwarding data to only 5 other nodes, products can be forwarded to more than 150 locations. Figure 1 shows the basic scheme. Nodes that receive and forward products to lower levels in the hierarchy are called RELAY nodes and end members of a chain are called LEAFs. UNIDATA has developed a socket-based system that is far more robust than standard UNIX utilities and a monitoring capability is under development to identify problem areas and eventually to dynamically route products around problem RELAYs. A huge amount of data can be forwarded in this manner to a large number of institutions without saturating a single node of the Internet. A potential advantage for IRIS would be the use of this method for sending data to additional distribution sites in Europe, Asia, South America and Africa, eliminating the need to send multiple copies of identical data over the oceans.
Figure 1. The UNIDATA system allows any node within the system to inject data products. Each relay can treat the product locally using LDM utilities but will also send it to downstream RELAYs or LEAFs.
Once data products are received at a RELAY or LEAF, the LDM determines if a product is of local interest and what to do with it. In this manner the only local computing resources that are required are Internet capacity, temporary buffers, and storage for data extracted for local use. The remainder of this article will provide details of the initial test of the IRIS implementation of UNIDATA´s system.
To identify data packages with seismological content, the UNIDATA system will be expanded to include a feedtype of SEISMIC. Initially IRIS has only defined one product, SPYDER® , to be transferred with this feedtype. SPYDER® data are those data that the IRIS DMS recovers from broadband stations of the IRIS GSN, cooperating FDSN stations, and other stations in response to alerts sent by the NEIC. SPYDER®data are now available in full SEED format and it is these SEED volumes that are transmitted through the IDD/LDM system. When a SPYDER® alert message is received at one of three international data centers (ORFEUS, GEOFON, and IRIS DMC), up to 14 cooperating data centers running SPYDER® nodes are contacted automatically and requested to recover specific time windows from stations for which they have agreed to recover data. The connections to these stations are normally done using high speed modems, or in some cases, Internet connections to the stations. These data are then forwarded to IRIS, ORFEUS or GEOFON by the Internet and then these data centers exchange data between themselves so that in general all three data centers have all of the data recovered for each SPYDER® event. This process normally takes from a few minutes to a few hours.
Periodically, SEED volumes are built that contain all of the data that have been recovered up to that time. These SPYDER® SEED volumes are injected into the IDD/LDM system. Each SPYDER® product is tagged (in LDM protocol) with specific information about the time, location and magnitude of the event. When the SPYDER® data product is received at an IDD/LDM location, the LDM software scans the product information and if it is determined that the product is of local interest, a user definable process is invoked. In the IRIS implementation this process is called SLUICE, an appropriate name since it is the technique to recover that portion of data of highest value to the local scientists. For instance the SLUICE process can be used to indicate that one is only interested in events from the Tonga trench, deeper than some depth and for data from only specific stations and/or channels. Ultimately SLUICE invokes RDSEED, and data can be translated into whatever analysis format an individual scientist wishes to use. Only data of specific interest are left on a user´s computer system and those data have already been converted into the format desired at that institution.
Figure 2. The initial test of the IRIS implementation of the IDD/LDM system was conducted between the IRIS DMC, the UNIDATA Program Center, the University of Arizona, Harvard, and the University of Michigan.
The initial test of the IDD/LDM system took place between IRIS, UNIDATA, University of Arizona, University of Michigan and Harvard University as shown in Figure 2. The system is stable and we believe it is an appropriate time to begin propagating the IDD/LDM approach to seismic data transfer to as many IRIS universities as possible. By installing the IDD/LDM system you will receive all SPYDER® data automatically and you will be able to customize locally such things as stations extracted and analysis format. Installation of the IDD/LDM system and the related SLUICE system will take a small amount of a knowledgeable person´s time, normally that of a systems administrator. If your institution would like to participate in this new method of timely information transfer, please fill out the form found on the WWW at
URL = http://www.iris.washington.edu/FORMS/IDD.form.html
You will be contacted by Rob Casey when it is time to install the software at your location. UNIDATA has already installed IDD/LDM software at more than 100 different universities in the United States and you might find it useful to contact your local atmospheric science department since the software might already be installed at your university and you may already have local expertise that will be of great benefit. 