Integrating multiple GIS predictions and optimizations at scale

Mosaic Software provides world-class GIS consulting for all businesses.

This is blog post #1 of a 3-part series focused on how technology can support GIS decision making using analytics and innovative design-thinking software development methodologies.

Read post #2 | The Strategic Use of Weather Data

Geographic Information Systems (GIS) benefit organizations of all sizes and in almost every industry. There is a growing need for and awareness of the revenue and strategic value of GIS-based software. In business and government, when operations require situational awareness of quickly changing conditions, today’s smart maps empower quick and authoritative communications.

Increasing Efficiency and Sharing Information with Users

Maps provide the means to capture knowledge and share data in a simple and comprehensible way. Maps capture reality using visual information that communicates understanding, shares perspectives and encourages coordination. Maps also form the basic unit of output from a GIS.

Technical advancements and new capabilities of GIS provide technology professionals with innovative new ways to present and communicate data. Business decision-makers expect unparalleled support in the form of data when making decisions, and GIS provides a great opportunity for technology groups to make an impact in their day-to-day business lives.

Designing and deploying a robust GIS system allows organizations to:

• Realize significant cost savings from greater efficiency,
• Significantly improve decision making,
• Improve enterprise communications,
• Promote better geographic information recordkeeping, and
• Manage business geographically.

GIS Technology Barriers

With all the benefits provided by a functional enterprise GIS program, it seems like a no-brainer for businesses to implement, yet so many fail to manage a successful GIS program. This is due to several barriers, including the difficulty of data integration and the multiple non-traditional software engineering skillsets required to deploy a fully functional GIS system.

For example, GIS uses particular types of data input files, and extensive data engineering is often required to make disparate data sources compatible. This can be even more challenging given the scale of relevant information in today’s business environment.

Even after data sources have been integrated into a single repository, the information may not be easily interpretable. In other words, robust GIS applications not only display data exactly in the form in which it is generated, but also map relationships between important data sources directly onto geographic locations. The process of discovering the right ways to combine various datasets to provide maximum insight typically requires the expertise of data scientists. Data science isn’t done in a vacuum, so software developers must be involved throughout the model development process to understand how to best capture and visualize the modeling results.

Finally, since data scientists’ time is usually at a premium, software professionals are expected to take these analytical insights and deploy them into business users’ hands. This requires another set of skills required to take advantage of GIS-based analytics. Not only are software engineers expected to visualize this data, but they need to understand aspects of human-factors management to best aid users in decision making.

Given the difficulties involved in making use of location-based data and the breadth of skills required to build GIS applications, it is no wonder that few software teams get it right.

GIS Analytics at Scale: Aircraft Routing Optimization

Mosaic Software, an innovative GIS consulting company, has been working with GIS systems in many industries for over a decade. Our parent company, Mosaic ATM has been providing air traffic management (ATM) analytics in the form of custom decision-support systems to NASA and the FAA for many years. In that time, we have learned quite a bit about how to tune GIS to get the most out of the data and technology.

With such a deep background in GIS consulting, Mosaic Software would like to share some insights through a tangible example focused on helping users decide the optimal trajectory for an airplane to follow during departure and takeoff.

Airports and airlines need to be able to predict how long a flight will take to fly its trajectory. Quite often, it has been adequate and possible to use the outputs of a one-off predictive analysis tool for this purpose. Mosaic has built custom tools to predict both the arrival time (ETA) and multiple intermediate time checkpoints along common flight paths.

But sometimes, predictive analytics based on existing routes aren’t good enough – such as when new flight routes are being developed, or when adapting existing routes for new aircraft. Mosaic Software has run into these challenges in the past; two of our GIS consulting experts engineered a modeling solution to predict flight durations for flights on new trajectories in order to select the most efficient route.

Data Engineering: Matching Weather & Routing

To estimate a flight’s duration, one needs to estimate two parameters: the total distance traveled, and the speed at which the plane can travel. The distance traveled includes the distance over the earth’s surface plus the vertical distance covered as the plane ascends and descends from the air.  The plane’s speed is a bit more complicated: air speed is a function of engine power and air density. As the plane climbs to higher altitudes, the air gets less dense, and the same amount of engine thrust has a smaller effect on the plane’s movement. Therefore, air speed must be estimated separately for every altitude at which a plane might fly. Furthermore, wind can passively affect a plane’s speed, either increasing or decreasing speed depending on the wind direction relative to the plane’s trajectory. As a result, to model plane performance, the GIS consulting team needed to aggregate data from multiple sources. Data was gathered from sensors monitoring fleet performance from similar aircraft over a pre-identified time horizon and atmospheric ‘nowcast’ data with air temperature, pressure, and wind speed from that same horizon.

The team then estimated the speed parameters of multiple types of aircraft by aligning historical data on flight times and routes with the atmospheric measurements over those same time points. Once the team had robust estimates of each plane type’s possible air speed at different altitudes, they could select possible flight trajectories between new origin and destination points and calculate the projected flight duration for those routes.

Deploying a custom Predicting & Prescriptive Enterprise GIS Tool

The GIS consulting team named the custom tool the Route Network Optimizer. This application helps users find optimized trajectories between an origin point (which may be the aircraft’s current position en route) and a destination. The software delivers a true 4D path drawing recommendations synthesized from aircraft performance (sensors), aircraft category, engine type, weather (atmospheric), and climbs/descents/cruise segments.

Due to the complexities of the national airspace, to provide decision makers with the optimal flight path, the custom software also pulls algorithmic recommendations from 3 different predictive/prescriptive models. These include the, aircraft performance model (described in the section above) a network-based search algorithm, and a tool known as the Clearable Route Network (CRN) Generator.  

The CRN Generator is a program that acquires and processes large amounts of flight data, then outputs a network with link-specific properties. The input flight data consists of 1) flight meta-data: ID, aircraft type, departure time, origin, destination, etc.; 2) one or more flight/route strings consisting of at least one full route and possibly multiple partial route updates and; 3) the flight’s trajectory points (4D positions). This flight processing includes stages for route aggregation, checking and link-building. These different stages take the model outputs and compare different routes in real-time. Mosaic Software provided valuable GIS consulting during these stages and built the infrastructure to be able to handle multiple algorithms running in the background.

The model retrieves aircraft performance information from data lookup tables through which a large amount of pre-computed aircraft trajectory information can be accessed by aircraft type.

Each table was engineered to aggregate and clean several data types, including: aircraft type, current state on a flight path (climb, cruise, descent), altitude, true-air-speed, climb/descent rates, and weather.

The custom software next uses the A-Star variant of the Dykstra search algorithm to search the network for an optimal path. For those unfamiliar with Dykstra’s algorithm, you can read more about it here.

The Route Optimizer designed by the GIS consulting team at Mosaic Software, fuses all of these algorithms together by finding the ‘optimal’ flight path between an origin and destination node in the network. Our custom software is designed to run a constrained optimization, accounting for aircraft performance calculations and costs along the flight path.

Our software provides users with real-time insights and allows decision makers to make snap decisions driven by real-time data. Mosaic Software is comfortable providing GIS consulting to any company looking to take advantage of real-time routing opportunities. We can help integrate new data sources, design software to interface with multiple algorithms, and deploy into an enterprise environment used by decision makers. Figure 9: Lateral Projection of Route Optimizer Search