Chart of the number of landings and interdictions over time. Intervals represent each month of data. Areas are color-coded and stacked
Edward Swing, Vision Systems & Technology, Inc.
Student Team: NO
The Prajna Project is a Java toolkit designed to provide various capabilities for visualization, knowledge representation, geographic displays, semantic reasoning, and data fusion. Rather than attempt to recreate the significant capabilities provided in other tools, Prajna instead provides software bridges to incorporate other toolkits where appropriate. Prajna will be released to the Open Source community in the near future.
For this challenge, I developed a custom application using the Prajna Project. The Prajna Project provided the utilities for reading the XML data files, segmenting the data into feature groups, and parsing the various field values.
For this challenge, the software bridge to JFreeChart provided the charting capability for statistical displays. JFreeChart is an open source project that provides a Java toolkit for displaying graphical charts. Prajna includes a GUI component that allows dynamic selection of different chart displays, based on the fields contained within the data.
Prajna also includes the capability to export to KML data files. Rather than develop an entirely new geographic display, I used Google Earth to display the KML.
The Prajna Project is a toolkit developed by Edward Swing. The custom application was built at VSTI. Other VSTI programs have since incorporated some of the new components that were developed for this contest.
To discover the patterns of landing sites, We divided the geographic area of interest into several regions. These regions were the Florida Keys, the Miami area, the Atlantic Coast, the South Gulf Coast, the Tampa area, the Cancún area in Mexico, the area near Cuba, and the region near Isla del Sueño. The data records were parsed by the Prajna Project XML parser, and various specific field manipulations were used to extract particular features which could be charted in a statistical display. Field extractors also identified individual passengers, and spotted those who tried to leave the island.
The records were also converted to a KML file for display in Google Earth. We used the KML exporter within the Prajna Project to export geographic overlays, separating the layers by vessel type and record type. The capabilities of Google Earth also include a temporal component which allowed the display of geographic elements which have a time associated with them. We included the DateTime field in the KML to take advantage of this temporal display feature. Between the geographic layers and the temporal component, we can easily filter and view only certain aspects of the data as needed.
Chart of the number of landings and interdictions over time. Intervals represent each month of data. Areas are color-coded and stacked
During 2005, we note that all of the vessels either were interdicted near Isla del Sueño, or attempted to land either in the Florida Keys or near Miami. This is likely due to the proximity of the island to these areas of Florida. During the entire three year period, the Keys and Miami regions remained a common landing destination.
Landings and Interdictions in 2005. Plain paddles indicate landing, the square-icon paddles indicate interdiction.
Data is color-coded by vessel type: Red=Go Fast, Blue=Raft, Green=Rustic.
As we were scanning through the records using the Google Earth time slider, we noticed the spread of migration areas as time progressed, and began to examine likely causes. On Jan 1, 2006, a raft landed near Cape Coral, FL, with a single survivor. The identity of that passenger is unknown. Prior to this, there were neither any landings or inderdictions outside of the Miami or Keys regions. Following this landing, more landings occurred in the South Gulf Coast area starting in Spring of 2006. No interdictions occurred in this area during Spring 2006. Over the Summer of 2006, perhaps because the landings were all successful, more landings occurred on Florida's gulf coast, in a significantly wider area. Landings occurred as far north as the Tampa Bay area.
Another significant change also occurs in Spring 2006. Starting in late April, 2006, a number of Rustic and Go Fast vessels landed near Cancún, Mexico. The landings were much tighter than along the coast of Florida, suggesting some coordination among the migrants. During 2007, more vessels arrived in the Cancún area than any of the other regions.
Landing attempts of Eduardo Catalano and Jesus Vidro.
One factor of interest were those passengers who attempted to migrate more than once. These passengers were identified by name automatically by the record loader. When we reviewed the list of repeat migrants, two names stood out - Eduardo Catalano and Jesus Vidro. These men were together on three separate attempts to flee Isla del Sueño, ultimately succeeding in a landing near Cancún on Feb 23rd, 2007, after two unsuccessful attempts to land in Florida.
Landings and Interdictions in 2007, showing broader area of migration.
In 2007, the Florida landing areas also broadened. In early 2007, vessels started landing further north, on both the Atlantic and Gulf Coasts. A significant number of vessels began landing successfully north of West Palm Beach, reaching as far north as Daytona on the Atlantic Coast. Similarly, a number of vessels successfully landed north of the Tampa Bay area.
Of significant note is the cycle of migration has a significant seasonal variation. Despite the lack of temperature variation in southern Florida and the Carribean, the attempted migration rate is highest in the summer months, and significantly lower during the winter months.
By charting the geographic areas over time, and restricting the data to interdictions, we can identify the general trends in the patterns of interdictions. These trends are corroborated by displaying the KML data in Google Earth, and restricting the data layers to the various interdiction records.
Chart of Interdictions over Time. Areas are color-coded and stacked. The selector at the upper right filters the records used to generate the chart.
In 2005, all interdictions occurred near Isla del Sueño, Miami and the Keys. Even during the later migrations, these areas made up the vast majority of interdictions. In 2006, the Coast Guard began to patrol the sought Gulf Coast, in response to the increased landing attempts in the region. Similarly, in 2007, the Coast Guard began interdiction efforts along the Gulf coast area near Tampa. However, the interdictions in those areas were few.
Geographic display of interdictions for Jan-Jun 2006.
No interdictions occurred north of the Miami area along the Atlantic Coast during the three year period. No known interdictions occurred near Mexico.
Again, by charting the data in statistical charts, we see that the interdiction rate exceeded the successful landing rate in 2005. In 2006, the Coast Guard was able to maintain a high interdiction rate in the Keys and Miami, and intercepted over 60% of the vessels. However, they failed to extend their coverage quickly enough to prevent migrants from landing along the Gulf Coast. Similarly, in 2007, the Coast Guard was unable to keep up with the migrants spreading to other regions. The landing rates in the areas outside of Miami and the Keys remained over 60%
Chart of Record Type over Time, displayed in Bar Chart, for the Florida Keys. Interdictions are
indicated in Red, Landings in Blue. Popup over bar chart displays numeric data.
Overall, the landing rate rose from 30% in 2005 to 57% in 2007. However, when we remove the Cancún area from the data set, the overall landing rate remained relatively constant, only increasing from 30% to 37%. This suggests that the Cancún region may be skewing the statistical results.