You are on page 1of 8

DETECTING AIRFIELD VERTICAL OBSTRUCTIONS USING DIGITAL PHOTOGRAMMETRY AND GIS

Ray L. Harris Jr., GIS Engineering Specialist Andrea R. Johnson, Systems Engineering Photogrammetrist
BAE SYSTEMS Mission Solutions 16250 Technology Drive Mail Zone 6102-U San Diego, CA 92127 ray.harris@baesystems.com andrea.johnson@baesystems.com

ABSTRACT
Historically, airfield vertical obstruction identification was done with conventional land surveying methods. Yet even with technological advances such as GPS, these surveys are expensive and time-consuming. Land based survey methods also have difficulty collecting obstruction data over the 7 nautical mile radius beyond the airfield. A recent review of such surveys has shown that important information regarding vertical obstructions and obstructing terrain around airfields is often missed or out of date. Consequently, many U.S. and international airfield surveys are backlogged or will not be performed. A new approach to aeronautical surveys using digital photogrammetric techniques combined with conventional survey data and geographic information systems improves obstruction identification for both terrain and manmade features. Computer visualization of the obstruction identification surface during stereo feature extraction enables the immediate detection of obstructions and reduces production time over traditional methods. A description of the techniques and tools developed for aeronautical surveys is presented with results from initial trials. In all trials to date, the use of digital photogrammetry has identified more terrain and object obstructions than conventional surveys. Maintaining the data in a digital format also allows both multiple product generation from the data and simpler update of airfield surveys over time.

INTRODUCTION
On April 3, 1996, Secretary of Commerce Ron Brown, business executives, and US military personnel met an untimely death while their aircraft was on approach to Dubrovnik, Croatia. Twenty-seven passengers and six crew members were on board the T-43 plane, a military Boeing 737 commercial airliner, when it went down in extremely poor weather conditions. The military aircraft, dependent upon instrumentation, was flying in bad weather without the aid of a Global Positioning System (GPS) receiver. The plane was attempting to make an instrument approach using a radio beacon at the airport when tower contact was lost. During their approach to the runway, which was bordered by mountains, the aircraft was off course and crashed into a mountainside. According to the official Air Force report, three major causes were identified. First, Air Force personnel were not aware that the Dubrovnik airport and its approaches had never been certified as safe by the Air Force. The airport approach was badly designed because the minimum descent altitude was set too low at 2,150 feet. By setting it at a higher level, off-course aircraft would not be able to stray into the nearly 2,400 ft. mountains. Second, the approach to runway 12 had not been designed properly by the Croatians partly because it used 1930s old-fashioned radio beacon landing systems. Third, according to the Air Force, gross pilot error contributed to the crash. The plane's pilots flew on a heading 9 degrees to the left of their proper course, and ultimately into the side of a mountain, St. John's Hill. The Air Force states the pilots failed to switch frequencies to the second beacon after passing the first radio beacon and were coordinating their course based on the wrong ground navigation beacon 12 miles out from the runway. The Air Force blamed its own commanders, pilot error and outdated navigation equipment for the plane crash. The Air Force punished 16 officers, including two generals and six colonels for their role in the crash. Two

were cited for "dereliction of duty" for failing to make sure that the correct instrument approaches were used for landings at Dubrovnik airport. Critics of the investigation have suggested that the inquiry was compromised because investigators began with the assumption the crash was simply an accident. Both the pilot and co-pilot had thousands of hours flying this type of aircraft. But the extensive Air Force Report, 7,000 pages and 22 volumes, detailed the sequence of errors that resulted in the crash. While multiple factors contributed to the cause of the crash, a change in circumstance to any factor may have avoided the accident. The report clearly indicated that improvements in Air Force command, and in the navigational data about the terrain and proper approach might have averted the tragedy. The US Congress, acting in honor of the victims and on behalf of aviation safety, funded the Ron Brown Airfield Initiative to improve navigational data around airfields. Congress tasked the National Imagery and Mapping Agency (NIMA) with collecting the improved data worldwide. Of the approximately 40,000 airfields worldwide, NIMA maintains navigational data on approximately 10,000 airfields. NIMA knew the task of improving the data would not be easy or simple.

Conventional Aeronautical Navigational Data Collection Process


Originally, aeronautical navigation data was intended to get the pilot into the general area of the airfield, and to provide obstruction and facility information. Once in the vicinity of the runway, the pilot was advised to use navigational aids (NAVAIDS) associated with the airfield to attempt an approach. Pilots were (and still are) expected to plan for their flight and use their vision or airfield operated landing aids. In the US, airfield NAVAIDS are usually well maintained but abroad these systems may not be as modern or reliable. For each runway, a set of Terminal Procedures is created to establish a safe and orderly approach to the airport. Terminal procedures are published by NIMA for each of the 10,000 airfields for which NIMA has data. To construct a terminal procedure, detailed information is needed on the runways, the airport facilities, and the obstructions around the airfield. Civilian or Airforce Terminal Procedure experts, nicknamed terpsters, use the information to establish runway coordinates, approach headings, and minimum safe altitudes. When procedures are established, they are passed back to NIMA for publication in the form of Terminal Procedure booklets know as FLIPs. Since airfield data can change quickly, the booklets are updated annually. For changes occurring before an update, information is passed to pilots using the NOTICE to AIRMEN system known as NOTAMS. NOTAMS are updated every 28 days. The Terminal Procedures process begins with an aeronautical survey. An aeronautical survey is a conventional land survey to determine important ground point coordinates to a high degree of accuracy. An aeronautical survey is a unique type of survey because it includes obstruction detection. The FAA no. 405 Standards for Aeronautical Surveys and Related Products sets rules for conducting the surveys and producing the chart products. Conventional aeronautical surveys start with a review of the airfield. For access permitted airfields, aerial photography is collected over a 7 nautical mile area around the airfield. A survey team establishes monuments and control, then locates and surveys all runways. NAVAIDS on and in the vicinity of the airfield are surveyed. Ground control from the survey is used to triangulate and control the aerial photography. Once the ground survey is complete, the photography is used to identify vertical objects around the airfield. Traditionally, the photogrammetrist, using either an analytical stereoplotter or a DPW, can only extract vertical objects that are candidate obstructions. Conventional devices did not have the ability to perform basic analysis of features. To determine which objects are obstructions requires the construction of and comparison against the imaginary obstruction identification surface (OIS). The OIS is composed of surfaces or facets that create an imaginary stadium above the airfield (Figure 1). Each runway has an approach ramp that extends at a set slope up from the runway ends out through the imaginary stadium. The whole OIS is composed of a primary surface, an inner and an outer horizontal surface, a conical surface, approach surfaces, and transition surfaces. The dimensions that define these surfaces vary depending upon the use (military, civilian, or NIMA) and the type of landing approach (precision instrument, visual, etc.). For military airfields, the primary surface is a rectangle the length of the runway and extending 1,000 feet on either side of the centerline. The inner horizontal surface is a plane at a height of 150 feet above the established airfield elevation. The outer horizontal surface is a plane, located 500 feet above the established airfield elevation, extending outward from the outer edge of the conical surface for a horizontal distance of 30,000 feet. The conical surface connects the inner horizontal surface outward and upward at a slope of 20 to 1 to the outer horizontal surface. The approach clearance surface is an inclined plane, symmetrical about the runway centerline extended, beginning 200 feet beyond each end of the primary surface at the centerline elevation of the runway end and extending for 50,000 feet. The slope of the approach clearance surface is 50 to 1 along the runway centerline

extended until it reaches an elevation of 500 feet above the established airport elevation. It then continues horizontally at this elevation to a point 50,000 feet from the point of beginning. The width of this surface at the runway end is the same as the primary surface, it flares uniformly, and the width at 50,000 from the runway end is 16,000 feet. Transitional surfaces are sloped planes that connect disjointed surfaces. The slope of the transitional surface is 7 to 1 outward and upward at right angles to the runway centerline. Airfield vertical obstruction feature extraction requires techniques that compare the elevations of objects against the OIS elevation. An airfield obstruction is any object that blocks the approach of aircraft on normal landing or takeoff. If an object is above the OIS, then it is an obstruction. Examples of obstructions are antennas, buildings, power lines, roads, rails, vegetation, construction equipment, etc. Vertical objects around the airfield Figure 1. OIS Visualization (source: NIMA). are not necessarily obstructions unless they penetrate the OIS. Since each runway can be a different length, the OIS must be constructed for each individual runway, regardless of the techniques used to extract height and elevation information on objects. Airfields that have more than one runway or different approaches with thresholds must have a separate OIS for each. Specific runway end coordinates are needed to build the surfaces in order to adapt the particular approach surface. The first order of business is to obtain the required runway data needed to construct the OIS. If reliable ground survey data is available, it can be used to establish runway ends and build an OIS. If reliable ground survey data is unavailable, the runway end coordinates are photogrammetrically extracted from stereo imagery. A product of the FAA aeronautical survey is the airport obstruction chart, a 1:12,000 scale drawing depicting the OIS, objects that penetrate it, runways, taxiways, and apron areas, NAVAIDS, prominent airport buildings, and a selection of roads and other planimetric details (Figure 2). Each chart has an accompanying obstruction data sheet that provides the coordinates of the features found on the chart. The NIMA airfield initiative intends to produce digital Geographic Information System (GIS) data layers that contain similar information.

Production Problems
Several obstacles (pardon the pun) limit the speed and turnaround for aeronautical survey projects. Ground surveys require access to the airfield and runways, which means that a busy airfield may need to suspend operations for the survey. International airfields may deny access. For national security reasons, some countries may decline to publish the precise locations of their airfields. Access permission can be a major problem that delays data acquisition and updates. Aeronautical surveys are also time-consuming and expensive, in part because surveying a 7 nautical mile radius and separating the obstructions and non-obstructions in a timely manner is a difficult but critical task. Thus, even though the FAA establishes that a periodic survey be conducted every 4 years, there are too many airfield and too little resources to meet the requirement. Figure 2. Aeronautical Obstruction Chart Example (source: NOAA NGS).

A NEW APPROACH TO AERONAUTICAL SURVEYS


The NIMA, recognizing the impracticality of improving obstruction data using conventional aeronautical survey techniques of ground access and aerial photography, proposed to Congress the Ron Brown Airfield Initiative (RBAI). The initiative incorporates the use of satellite imagery and digital photogrammetric workstations (DPWs) with stereo feature and terrain extraction to find obstructions. At the beginning of the initiative, NIMA intended to collect all vertical objects around an airfield and populate a vertical obstruction database (VOD) and GIS data layers with all vertical objects. This idea would bypass the access problem and allow NIMA to collect vertical objects even if they could not get on the airfield. The NIMA Airfield Initiative Document (http://www.nima.mil safety of navigation link) establishes the specifications for the production and output of the obstruction data. The specifications have recently been completed, but are still subject to revision in part due to the new techniques in development. A mission of the RBAI was to quickly and to efficiently collect, by survey and photogrammetric means, comprehensive and accurate airfield vertical obstruction data and to provide that data in a uniform and standardized format. As NIMA prepared to collect vertical object data, it became apparent, especially in densely populated areas, that extracting all vertical objects was potentially as impractical as ground surveying the obstructions. Accurate estimates of the time required to extract all features above a threshold elevation were not available. However, experienced stereo analysts expected at least one full labor week would be needed for complete feature extraction within each runways area of interest. Assuming an average of 1.5 runways per airfield, and 1,000 airfields, the estimated time needed to extract all features was roughly 30 person-years. The drawback to this brute force extraction approach was that it did not differentiate between an obstruction and a non-obstruction. Feature extraction of every object above a threshold elevation within the 7 nautical miles would still need to be filtered to remove the objects that do not penetrate the surface. This could be done with a subsequent process using a geospatial analysis package. Extraction of all objects within the OIS was inherently wasteful and time-consuming because the operator had to over-collect the data. What was needed was a new, faster approach to obstruction identification. Exactly how to represent the OIS during stereo feature extraction on the DPW to ensure proper selection of obstructions was a major problem. The work was performed on NIMAs Integrated Exploitation Capability (IEC) workstations, DPWs with GIS software. At the beginning of the initiative, no commercial software in the IEC suite had a tool for OIS generation. The only way to build the OIS using the IEC workstation was to use the SOCET SET sketch tools to manually draft the three dimensional object. To complicate matters, the OIS dimensions were dependent upon the runway length that meant simple scale and rotation of a template OIS were not possible. No complete estimates of time required to manually draw an OIS were available. However, prototyping research determined at least one full labor day would be needed for each runway. Assuming an average of 1.5 runways per airfield, and 1,000 airfields, the estimated time needed to build all the OIS was 6 person-years. Without solving this problem, stereo operators had no choice but to over-collect features.

Combining Digital Photogrammetry and GIS


By re-examining the requirements and the goals of the mission, we created a new method for the feature extraction of airfield vertical obstructions using photogrammetry, GIS, and geospatial analysis techniques that focused on rapid obstruction identification. The new, more efficient approach was to perform feature extraction on only those objects that penetrate the OIS. This was done using a surface model to represent the OIS during photogrammetric data collection. The first part of the new approach was to create a reflective surface (RS) elevation model from stereo imagery, then create a surface model from the OIS, and intersect the two surfaces. The resulting objects above surface (OAS) contained the ground and objects whose elevation was greater than the OIS. Automatic terrain extraction methods generated the reflective surface model, and the OIS vector model was converted to a raster surface model with the same spatial resolution as the reflective surface model. The intersection was simple spatial algebra where OAS = RS OIS . In effect, the OAS is the set of vertical obstructions defining an obstructing area. The OAS model had limitations because automated terrain extraction produced some elevation errors that must be reviewed by an operator using quality statistics and terrain shaded relief as visual aids. Vertical obstructions that are smaller than the spatial resolution of the reflective surface model were not detected by this approach. However, the resolution of the imagery was much greater than the spatial resolution of the reflective surface model. Therefore, obstructions such as power lines, antennas, or flagpoles required direct stereo photogrammetric collection on the imagery. The safe approach was to use the OAS as an indicator of potential obstruction rather than relying on the RS and spatial algebra alone.

The combination of stereo imagery, a wireframe of the OIS, and the polygons of potential vertical obstructions on the DPW helped the operator identify and measure obstructions more quickly than without these aids. With the development of the new Commercial-Off-The-Shelf software called ClearFlite, an OIS that used to take a day to manually draw now takes less than 5 minutes. Obstruction detection that took a week now takes two days. The OAS polygons were displayed with the stereo imagery and the OIS wireframe boundary. These aids helped guide the operators to areas that contained obstructions. Without the addition of the wireframe OIS and the OAS potential obstructions, the operator was not able to visualize the surface to determine if an object met the obstruction criteria. The step-by-step method was as follows: 1. Extract or input runway points 2. Build obstruction identification surface model 3. Build high resolution reflective surface model 4. Intersect reflective surface and obstruction identification surface 5. Convert result to vector polygons and identify maximum elevation points within polygons 6. Display surfaces and objects in stereo a. obstruction identification surface model b. objects above surface polygons and maximum elevation points c. stereo imagery d. obstruction objects 7. Extract features and automatically populate attributes 8. Output feature coordinates and attributes in geospatial format

Figure 3. Example of airfield imagery, OIS blue lines, red circle cursor indicating an obstruction, and automatically extracted features as yellow polygons. Stereo analysts also needed visual indicators and information about the features above the OIS to help determine which features to extract. A list of features were developed and packaged into ClearFlite. Some of the key features are:

Lets the stereo operator select the runway endpoints and automatically creates a 3D vector and elevation model of the OIS over any runway Creates the NIMA and FAR 77 OIS Stereo view of terrain, features, and obstructions around an airfield Provides a cursor tracking window that continuously displays elevation and height above the OIS Has a visual indicator when objects are above or below the OIS Automatically calculates height and populates feature attributes Tracks inspected areas of the OIS Export to geospatial formats Export an FAA Universal Data Delivery Format data file Automatically delineates terrain obstructions

Initial Trials
The new method was applied on a test case of the San Diego International Airport, on seven airfields at NIMA, and on eight airfields under private contract. The San Diego International Airport was a good test because the airfield has terrain and building obstructions on three sides, as well as offset thresholds at each end of the runway. An FAA Airport Obstructions Chart and Universal Data Delivery Format file for the runway data already existed. All Figure 4. Perspective view of the 3D OIS and wireframe terrain geographic coordinates of the airport, for the San Diego International Airport. runway, NAVAIDS, and obstructions were published in the data file. The data file for the San Diego International Airport was used to verify the runway endpoints. The ClearFlite software generated an OIS 3D vector shown in Figure 4. The OIS was automatically scaled and oriented when the operator entered endpoint coordinates. Imagery draped over the terrain perspective view shows the aircraft approach on descent toward runway 27 at the bottom of the imaginary stadium (Figure 5). Elevation was exaggerated to show the slopes of the OIS Areas of terrain that exceed the elevation of the OIS model protrude into the center of the imaginary stadium. Applying the spatial algebra of subtracting the reflective surface from the OIS yields the objects above surface (OAS). Rendering the wireframe with an image of the airfield, the OAS tinted in red, and the 3D vector shown in blue gives a perspective view of obstruction areas around San Diego International Airport (Figure 6). The OAS identified by this process generally agrees with the obstructing building area polygons found on the Airport Obstruction Chart but over indicates obstructions. Thus, the automatic method is used only as an indicator. The raster OAS converted into a 3D vector is displayed as yellow polygons in Figure 3. During the stereo extraction process the operator used the OIS, and optionally the OAS, as aids for selecting obstructions. Once the tops of the obstructions were described and measured for elevation, the critical attributes of elevation, height above ground, and height above the OIS were automatically populated. The obstruction features were exported and easily displayed for quality control as represented in Figure 7. Similar work has occurred in NIMA and at private contractor locations. NIMA analysts have used Figure 5. Imagery draped over the terrain with the the ClearFlite tool to identify more obstructions than OIS. Terrain protrudes when the red OIS lines were currently documented in the digital vertical disappear. obstruction file (DVOF). The DVOF database ranks the

quality of information on a scale of 1 to 5, 5 being the worst and 1 being the best as ground survey. Since the IEC is a DPW with stereo analysis ability, all obstructions found by using the ClearFlite tool and SOCET SET are automatically a 2 on the quality scale. In one example for an airfield in Greenland, the DVOF listed only 5 obstructions, while the RBAI team identified 35 obstructions. The RBAI team also generated the Airfield Elevation Model (AEM) that clearly showed the rugged terrain bordering the airifield. Private contractor ADR, Inc. used the ClearFlite tool to reduce the amount of time needed to complete an aeronautical survey. ADR analysts state that the new approach cuts their production time in half. Figure 6. Terrain, tinted red, extending above the OIS.

Accuracy
The Ron Brown Airfield Initiative specification contains accuracy requirements for planimetric and vertical components of each airfield feature. NIMA, the FAA, and the International Civil Aviation Organization (ICAO) all have different accuracy requirements. In nearly all cases, NIMA has the most stringent accuracy requirements for airfield points of interest. Because of these accuracies, and per requirements, some airfield points of interest must be captured using ground survey methods. These features include PACS (Primary Airport Control Stations) and SACS (Secondary Airport Control Stations), runway endpoints, threshold endpoints, overrun (or stopway) endpoints, etc. These points require measurements in the WGS 84 system of 0.42 meters in horizontal, and 0.42 meters in vertical. These accuracies are not typically achievable in photogrammetry using commercial satellite or frame photography. Dr. Ing Ricardo M. Passini of BAE Systems, ADR division, researched the accuracy and resolution of

Figure 7. Perspective view of San Diego International Airport showing terrain obstructions tinted red, yellow polygons representing obstructing areas, and symbol obstructions. digital mapping from aerial photography to take into consideration atmospheric and scanning degradation. He outlines a case typical of airfields to process the least number of models and still meet/exceed the accuracy requirements of airfield features and obstructions. Using a typical aerial camera with a 6 focal length, imagery with a scan resolution of 15m and an aerial photography scale of 1:24,000 and the assumption of a pointing accuracy of 15m (i.e.: not subpixel, although subpixel measurements are possible), the standard deviation in elevation (Z) would be 0.815m, and the standard deviation in planimetric direction (,) of 0.509m. This meets and exceeds all NIMA, FAA, and ICAO accuracy requirements.

This is an ideal scenario for airfields where ground access is permissible to perform a survey and fly aerial photography. However, NIMA estimates that as many as 700 of the 1,000 airfield on the initiative plan are cases where access to airfields is restricted or denied. In these cases, the only imagery sources are satellite imagery, and it is not possible to achieve the accuracy outlined in the current RBAI spec for NAVAIDS and important runway points. However, the accuracy requirements for vertical obstructions can be met with satellite imagery, which is the essence of the Airfield Initiative.

CONCLUSION
Safety of aeronautical navigation is dependent upon accurate and current airfield obstruction data. Manual methods, including land surveys and airfield product generation, are not sufficient to meet the time, cost and data management demands. Collecting that data rapidly and efficiently with aerial imagery and digital photogrammetry is an improvement over conventional manual techniques. The process is faster and more accurate with the marriage of digital photogrammetry and ClearFlite. There are several advantages to digital production. One is the improved obstruction collection. For the most accurate products, ground survey is still crucial. But the advantages that a SOCET SET and ClearFlite system provides for identification of obstructions is a significant advancement because more obstructions are being identified for less cost. The second advantage is timesaving. Features are easily visited and determined to be obstructions because the software continually indicates to the operator what to collect. Third is the ability to maintain and update the products. SOCET SETs ability to output digital features to formats compatible with many GIS packages makes it easier to produce updated products by simply overlaying existing data with new imagery. Lastly, the data collected is the first component for aeronautical navigation to accomplish the directives that spawned the Ron Brown Airfield Initiative and the future use of GPS guided landing of aircraft with graphical pilot heads-up display. In the future, pilots will no longer have to rely on outdated 1930s technology to guide them onto a runway in unsafe conditions but they will always need accurate and timely obstruction data. Initial trials have been performed by the NIMA and by experienced contractors aided in the development of the software and workflow. With these digital tools, it is more likely that the RBAI can be completed in a timely manner for the 1,000+ airfields. The NOAA National Geodetic Survey (NGS), who maintains airfield data on 3000 civilian airfields, is currently evaluating the tool for their use in aeronautical surveys.

REFERENCES
BAE Systems, (2000), ClearFlite Users Manual Version 1.1. Mission Solutions, Talisin Business Unit McIntyre, J. (1996). Air Force punishes 16 officers in Brown crash, CNN, http://www.cnn.com/US/9608/06/ron.brown.crash/ Passini, I. (2000). Resolution and Accuracy. BAE Systems, ADR, Inc. Ruddy, C. and H. Sprunt, (1997). Questions linger about Ron Brown plane crash, Tribune Review. http://tribunereview.com/ruddy/112497.html U.S. Department of Transportation, Federal Aviation Administration (1996). Standards for Aeronautical Surveys and Related Products, FAA No. 405, Fourth Edition National Imagery and Mapping Agency (2000). Airfield Initiative Document, http://www.nima.mil Navigation and Safety link

You might also like