Demonstrating UAV-Acquired Real-Time Thermal Data over Fires Vincent G. Ambrosia, Steven S. Wegener, Donald V. Sullivan, Sally W. Buechel, Stephen E. Dunagan, James A. Brass, Jay Stoneburner, and Susan M. Schoenung Abstract winter in the southeastern U.S. (initially in Florida) and Project FiRE (First Response Experiment), a disaster transitions both northward into the Appalachian states and management technology demonstration, was performed in westward. The severe fire season transitions into the south- 2001. The experiment demonstrated the use of a thermal west, north through the Rockies, west to the Pacific North- multispectral scanning imager, integrated on an unmanned west, then south through California, culminating in late aerial vehicle (UAV), a satellite uplink/downlink image fall/early winter fire events in southern California. This data telemetry system, and near-real-time geo-rectification extended fire season taxes the resources of local, state, and of the resultant imagery for data distribution via the Inter- federal agencies mandated to monitor and mitigate these net to disaster managers. The FiRE demonstration pro- events. During the extensive burning seasons throughout vided geo-corrected image data over a controlled burn to the United States, piloted aircraft employing thermal imag- a fire management community in near-real-time by means ing systems are deployed by some state agencies (California of the melding of new technologies. The use of the UAV Dept. of Forestry; CDF) and by the National Interagency Fire demonstrated remotely piloted flight (thereby reducing the Center (NIFC; Boise, Idaho) to collect large volumes of data potential for loss of human life during hazardous missions), over fires spread from Canada to Mexico, and from the and the ability to “linger and stare” over the fire for ex- Rocky Mountain Front to the Pacific Ocean. Large fires, tended periods of time (beyond the capabilities of human- such as the Yellowstone conflagration in 1988, the Cerro pilot endurance). Improvements in a high-temperature Grande, New Mexico fire in 2000, the western Montana calibrated thermal imaging scanner allowed “remote” oper- fires of 2000, and the Colorado and Arizona fires of 2002, ations from a UAV and provided real-time accurate fire burdened these remote sensing data-gathering crews, and information collection over a controlled burn. Improved taxed both the resources and the stamina of these personnel. bit-rate capacity telemetry capabilities increased the Through cooperative endeavors between agencies, a amount, structure, and information content of the image team of investigators have embarked on a research and data relayed to the ground. The integration of precision technology demonstration utilizing UAVs and advanced navigation instrumentation allowed improved accuracies sensor and data distribution technologies focused on in geo-rectification of the resultant imagery, easing data improving the information gathering over wildfire disaster ingestion and overlay in a GIS framework. We present a events. The rationale for exploring the development of the discussion of the feasibility of utilizing new platforms, platform, payload, data telemetry and geo-rectification pro- improved sensor configurations, improved telemetry, and cedures was to greatly increase the timeliness of the data new geo-correction software to facilitate wildfire manage- stream for utility in fire mapping, to reduce potential risks ment and mitigation strategies. to pilot and system engineers, and to provide improved and more accurate information on fire conditions than is Introduction currently realized. The culmination of those efforts resulted in the UAV-FiRE demonstration. Large-scale wildfires occur frequently throughout the The Uninhabited Aerial Vehicle (UAV) First Response United States every year. The “fire” season begins in late Experiment (FiRE) was created in 2001 to demonstrate the utility of integrating remotely piloted aerial platforms, advanced thermal imaging, cost-effective satellite data V.G. Ambrosia is with California State University– telemetry systems, and image geo-rectification for rapid Monterey Bay, Earth System Science and Policy Institute, data dissemination of a disaster event in near real-time M.S. 242-4, NASA-Ames Research Center, Moffett Field, (Figure 1). The primary focus of the FiRE demonstration CA 94035-1000 ([email protected]) was to utilize the unique features of an unmanned aircraft as a data-gathering platform to test the feasibility of long S.S. Wegener, D.V. Sullivan, S.E. Dunagan, and James A. duration missions to support wildfire management activi- Brass are with the NASA-Ames Research Center, Moffett ties. Although UAVs are currently expensive to operate, the Field, CA 94035-1000 ([email protected]; purpose of the demonstration was to showcase their feasi- [email protected]; [email protected]; bility for future operational data gathering. Within the next [email protected]). S.W. Buechel is with Terra-Mar Resource Information Services, Hood River, OR 97031 ([email protected]). Photogrammetric Engineering & Remote Sensing J. Stoneburner is with General Atomics-Aeronautical Vol. 69, No. 4, April 2003, pp. 391–402. Systems, Inc., San Diego, CA 92127 (jay.stoneburner@ gat.com). 0099-1112/03/6904–391$3.00/0 S.M. Schoenung is with Longitude 122 West, Inc., Menlo © 2003 American Society for Photogrammetry Park, CA 94025 ([email protected]). and Remote Sensing PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING April 2003 391 Figure 1. FiRE Demonstration Concept Plan. Thermal infrared image data collected from a payload operating on a UAV is telemetered over-the-horizon through an orbiting communications satellite to a data command center. At the data archive center, the imagery is processed in near real time, archived, and also distributed to the World Wide Web for use in disaster visualization and mitigation response. few years, as operational costs decrease, UAVs will prove to sets. The integration of the various technologies and the be an effective alternative platform for long-duration data ultimate success of the project were due to this unique gathering missions. UAVs and the improved data telemetry partnership among all the participants (Brass et al., 2001). technologies described in this paper offer vast improvements in platform range capabilities (ability to cover continental- Background scale areas), timely delivery of data, and improved quality of information to the disaster manager. The employment of Fire imaging from remote platforms can be divided into a multichannel thermal imaging payload also offers im- two distinct categories: strategic and tactical. Strategic provement over the current use of forward looking infrared observations are those that provide a regional view of fire (FLIR) systems and single-channel thermal systems with occurrences. Tactical observations are those that provide poor (or non-existent) high temperature calibration, narrow detailed, frequent, and repetitive views of specific fire FOVs, and low spatial and radiometric resolution. Improve- events. Satellite systems such as AVHRR have been used to ments in on-board and ground-based image geo-correction provide strategic views (Matson and Dozier (1981), Matson technologies also allow vastly improved data integration et al. (1984), and others) but are ineffective for repetitive, capabilities and speed. timely observations of fires due to orbit cycles, design char- The success of the FiRE project was the result of a col- acteristics, and data relay (days, weeks to months after the laborative effort between government agencies and private fire). While the Forest Service and other agencies utilize industry. NASA-Ames Research Center organized the project satellite data to provide an overview of fire distribution at and the enhancement and development of the AIRDAS scan- regional scales, they rely on the use of airborne platforms ner as well as the integration of all the components needed as tactical systems to provide continuous coverage and for the demonstration. Ames cooperated closely with Gen- rapid data accessibility over individual fire events. The eral Atomics Aeronautical Systems, Inc., (GA-ASI) who de- ability to change aircraft position, altitude, and data acqui- veloped, built, and flew the ALTUS® II UAV airborne plat- sition time makes them essential for real-time fire fighting form. System integration was accomplished at the GA-ASI and management decisions. This paper will focus on the Flight Operations Facility in El Mirage, California. The use of improved tactical airborne remote sensing systems. telemetry system, provided by Remote Satellite Systems, Airborne fire imaging has changed dramatically in the Inc. of Santa Rosa, California, was modified by NASA-Ames United States since the early assessments performed with a and integrated by GA-ASI into the ALTUS® II UAV fuselage. Polaroid Land Camera in the late 1940s and early 1950s All geo-rectification procedures and data handling were co- (Arnold, 1951; Johnson and Thomas, 1951). Throughout the ordinated and performed with Terra-Mar Resource Informa- 1960s the U.S. Forest Service–Northern Forest Fire Labora- tion Services of Mountain Ranch, California. Disaster man- tory (Missoula, Montana) pushed the forefront of fire imag- agers from the U.S. Forest Service (USFS) and the State of ing technology with various programs and research efforts California Resources Agency participated as technology re- (Hirsch, 1962; Hirsch, 1963; Hirsch, 1964; Hirsch, 1968; viewers and provided feedback on the resultant fire data Hirsch et al., 1971; Wilson, 1968). During the 1970s, exper- 392 Month 2003 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING imentation with newly developed digital imaging systems the development of faster geo-rectification algorithms for led to an expansion of the aerial fire assessment programs correcting the telemetered thermal data when it arrived at within the U.S. Forest Service and also within NASA (Brass the data server location. The data were required to be geo- et al., 1987; Ambrosia, 1990). The Forest Service maintained corrected and in a universal GIS-compatible format. The a program of fire research at the Missoula Lab and at other success of that development effort was evident in the WILD- facilities including the Forestry Applications Program, then FIRE project (Ambrosia et al., 1998). AIRDAS data, captured at Johnson Space Center in Houston, Texas (now the Forest from a NASA Lear jet, over a controlled burn, were relayed Service Remote Sensing Applications Program (RSAC) in to a field-portable computer at an Incident Fire Command Salt Lake City, Utah). NASA’s activities were largely struc- Center via the aforementioned Flitefone 800® system. The tured around development of technological tools and the data were geo-corrected using a single control point and an employment of high altitude aerial platforms (such as the approximated two-dimensional model, and were available U2/ER2) to assist in Earth resource assessments (Brass et al., within 30 minutes of collection. 1987; Ambrosia, 1990). These programs gravitated naturally In the late 1990s, cooperative efforts between the USFS into cooperative efforts focused on development of plat- and NASA-Ames established new requirements for improv- forms, payloads, and software for improvement of fire imag- ing wildfire monitoring from airborne platforms. These ing technologies. requirements included over-the-horizon (OTH) telemetry During the 1980s the National Interagency Fire Center capability, greater bit-rate throughput telemetry, multiple (NIFC) initiated research and development activities in spectral band data collection and transmission, and im- infrared fire imaging. These activities included the use of proved geo-rectification procedures. The final requirement forward looking infrared (FLIR) systems (George et al., was to provide these elements within one hour or less of 1989), and development of the Flying Infrared Enhanced data collection. With the advent of remotely piloted aerial Maneuverable Operational User Simple Electronic Tactical vehicles and their attendant large payload capacity, the part- Reconnaissance And Patrol (FIRE MOUSE TRAP (FMT)) imag- ners began efforts to develop new technologies to exploit the ing system (Dipert and Warren, 1988; Warren, 1990; Warren UAV as an aerial remote sensing platform. That effort led to and Celarier, 1991; Scott, 1991). In the late 1980s and early the UAV FiRE project in 2001. The demonstrations of these, 1990s the Forest Service and NASA-JPL (Jet Propulsion Lab- and emerging technologies, are the focus of this paper. oratory) jointly developed the FIREFLY (FF) thermal fire imag- ing system (Warren, 1992; Warren, 1994). That system was Goals of FiRE Demonstration utilized for a number of years and was reconfigured in the mid/late 1990s into the PHOENIX system. Project FiRE focused on demonstrating engineering concepts Following a parallel development track, NASA-Ames for future use of UAV platforms, integrating appropriate dis- Research Center began research on improved thermal aster imaging payloads, integration of improved data teleme- infrared scanning systems for fire imaging. NASA-Ames try systems, and data rectification and distribution. The researchers focused on developing complete end-to-end goals of the project were to systems that would deliver accurate fire temperature data ● Integrate an imaging system and telemetry equipment on a and deliver the information remotely in near real time. The high performance UAV, Airborne Infrared Disaster Assessment System (AIRDAS) was ● Operate the payload remotely from a ground station via the result of those activities. AIRDAS was designed to fill a telemetry protocols, critical gap in remote sensing fire research and to support ● Telemeter payload data from the UAVto a communications fire applications related activities. Those critical gaps relate satellite and OTHto a ground receiving station, to the need for design of appropriate thermal channel and ● Provide automated geo-rectification and correction of the data, and band-pass regions, improving high temperature calibrations, ● Globally distribute that data to the Internet and disaster improvements to electronics systems to facilitate high data managers. quantization, and development of high temperature resolu- tion systems. The AIRDAS was conceived to overcome short- The first goal of the FiRE demonstration was to success- comings of previous instruments not designed for accurate fully integrate and fly the AIRDASscanner on a UAVplatform. discrimination of high temperatures found in a wildfire The ALTUS® II UAV payload integration of the AIRDAS was condition. Those shortcomings led to the development of performed at the GA-ASI Flight Operations Facility in El narrower thermal bandpass filters, improved detector devel- Mirage, California during late summer of 2001. opment, and innovative design changes to signal pre- The second goal, to operate the AIRDASpayload remotely amplifiers to facilitate high temperature discrimination with from a ground station via command and control telemetry, appropriate detector designs (Ambrosia and Brass, 1988). required modification of the instrument control software and Perceiving the need for rapid payload data delivery, all functionality performed by a systems engineer. NASA-Ames shifted development to aircraft telemetry sys- The third goal, to provide satellite data communica- tems (Ambrosia, 1990). Low-bit data rates (4.8 Kbs) were tions telemetry, required modification and integration of a achieved by enhancing a Flitefone 800® air-to-ground digi- satellite phone antenna system onboard the ALTUS®II UAV. tal cellular phone system for data relay of “frame grabbed” The telemetry system provided a communication and data scenes from the AIRDAS controller’s station on-board the link to the INMARSAT (International Mobile Satellite Orga- acquiring aircraft. This procedure necessitated dialing into nization) satellite system. a computer server at Ames Research Center and transmit- The fourth goal, to provide improved geo-rectification ting a single spectral band image file from the aircraft. The of fire image data, required customized image correction file was sent unrectified and was available for analysis software provided by Terra-Mar. The software required within approximately 15 minutes. This procedure was inputs from the platform, and payload navigation informa- employed successfully for a number of years in the 1990’s tion collected in conjunction with the AIRDAS image data. and was used over numerous fires both in the western U.S. The fifth goal, to globally distribute the real-time cor- and in Brazil. rected thermal imagery via the Internet, allowed the use of The requirement for geo-rectified, GIS-compatible the resultant data both at an Incident Command Center thermal-infrared image data drove the development and (ICC) and at other facilities, including remote fire camps. further refinement of the AIRDAS and telemetry system. An uncorrected image file and geo-rectified file were sent Cooperative efforts with private industry (Terra-Mar) led to to a server at NASA-Ames and accessed via a web address. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING Month 2003 393 The following sections describe the FiRE experiment proximately 35 meters. After the propane was ignited, ver- components, the demonstration, results, and future direc- tical flame lengths reached approximately 3 to 5 meters. Pits tions for our research and development program. containing fuel were disbursed around the propane feeder pipes (Figure 3). These fuel pits were ignited and provided FiRE Demonstration added flame and thick black smoke, obscuring the flames, as would occur in a natural fire. The controlled burn was Overview ignited immediately prior to the ALTUS® II rollout and The FiRE demonstration mission occurred on 06 September takeoff and continued for the duration of the mission (over 2001 at the GA-ASI Flight Operations Facility in El Mirage, one hour). California. Disaster managers and fire management person- A GA-ASI PREDATOR B® UAV was launched prior to nel viewed the FIRE demonstration and participated in an the ALTUS® II to provide overhead video coverage of the evaluation of the procedures and products. Integration of mission. The video data were relayed to the Command and the payload and telemetry suite occurred during the week Control station and distributed via closed-circuit monitors preceding the mission, and a full, end-to-end UAV/payload/ to the attending disaster managers and audience. A piloted telemetry test flight was also performed the previous day. GA-ASI Cessna 182 aircraft was used for in-flight photo- The FiRE Demonstration controlled burn site was located graphy while the aircraft flew within the test range. at the GA-ASI Flight Operations Facility at El Mirage, Cali- The controlled burn was ignited adjacent to the aircraft fornia. The first phase of the FiRE project involved local runway and the UAV was launched at 0800 (PDT). After at- flights over controlled burns in the vicinity of El Mirage. taining the planned data collection altitude of about 945 me- The fire location was moved to the GA-ASI facility to im- ters AGL (about 3100 ft), the ALTUS® II flew “racetrack” pat- prove control of the burn (during high fire season on sur- terns in an east/west orientation over the controlled burn. rounding wildlands), and to better coordinate the project This altitude provided a pixel spatial resolution of about observers and fire managers. The ALTUS® II UAV was there- 2.5 meters. The first data collection pass was made at fore limited to flights within the GA-ASI Flight Operations 0822 (PDT). Range. The Flight Operations Facility at El Mirage is situ- ated at 872 meters (2861 ft) elevation at 34°37(cid:10)30(cid:11)north lat- ALTUS®II UAVDescription itude and 17°36(cid:10)21(cid:11)west longitude and has a 1525-meter The ALTUS® II aircraft used for this demonstration is a (5000-ft) runway situated on a dry lakebed. The field is derivative of the GA-ASI PREDATOR® UAV. The ALTUS® II 30 kilometers (19 miles) northwest of Victorville, California builds on the proven technology of the PREDATOR® sys- and 47 kilometers (29 miles) east of Palmdale, California, tem with an altitude optimized propulsion and control sys- in the Mojave Desert of San Bernardino County (Figure 2). tem to create a UAV that is capable of high altitude, long The demonstration burn was located approximately endurance flight (Table 1). The ALTUS® II was developed 50 meters (50 yards) north of the runway. The burn was under the NASA AeroSpace Enterprise, Environmental created by a controlled ignition of a large supply of propane Research Aircraft and Sensor Technology (ERAST) program fuel fed through a series of vented feeder pipes. The feeder and is designed as a technology demonstrator for scientific pipes were laid out in a linear pattern and extended ap- research and commercial applications (Figure 4). Figure 2. Location of General Atomics Aeronautical Systems, Inc. Flight Operations Facility and FiRE test burn site. The demonstration site is in the Mojave Desert near El Mirage, San Bernardino County, California, northeast of Los Angeles. 394 Month 2003 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING Figure 3. Controlled burn site at GA-ASI Flight Opera- tions Facility, adjacent to runway. The controlled burn was created with a propane pipe feeder system and fuel Figure 4. The ALTUS® II UAV in flight during the FiRE pits ignited immediately prior to ALTUS® II take-off. demonstration mission, September 2001. The telemetry antennas are integrated into the fuselage and located on both sides of the aircraft at the FiRE logo, aft of the pay- load section. The payload (AIRDAS scanner) is located in The ALTUS® II UAVis controlled via a command and the nose. control (C2) link that is separate from the data telemetry link. The C2link is used strictly for flight and instrument opera- tions. The aircraft can be easily disassembled, crated, moved, and reassembled rapidly (assembly in under five hours). This facilitates operations in remote areas away from base opera- UAV and passing real-time payload data at ranges up to tions and provides rapid response for emergency situations. 280 km (150 NM). In addition to the LOS link, a high data rate Ku-Band satellite data link for routine over-the-horizon The remote operations for the UAVplatforms are provided by personnel in a Ground Station C2 trailer, whichis described operations is available and has been used by the military. GA-ASI has also developed a high mobility ground control in detail in the following section. station and a portable ground control station that allows GA-ASIALTUS Ground Station control of an unmanned aircraft to be passed to remote Remotely piloted flight operations for the ALTUS® II UAV pilots at an advanced location. The pilot and co-pilot rely on aircraft attitude and positional information provided by platform are performed on the ground at the GA-ASI Ground Station (Figure 5). The ALTUS® II and all GA-ASI aircraft the Litton LN-100G INS/GPS system as well as forward- are controlled by a portable common solid-state digital ground control station (GCS) through a C-Band line-of-sight (LOS) data link. The GCS is capable of direct control of the TABLE1. GENERALATOMICS-AERONAUTICALSYSTEMS, INC. (GA-ASI) ALTUS®II UAVSYSTEMSPECIFICATIONS Dimensions Wing Span 55.3 ft.; Wing Area 132 sq. ft.; Length 23.6 ft.; Height 9.8 ft. Weights Max Fuel Wt. 550 lb; Payload Wt. 330 lb.; Max GTOW 2,150 lbs. Propulsion/Fuel Rotax 914-2T Dual Turbo; liquid-cooled four cylinders. Rated 100 HP @ 52,000 ft. Performance Max Altitude 65,000 ft.; Endurance: 8 Hours (at 60K), 18 Hours (at 30K), 24 Hours (at 25K); Max Speed 100 KIAS, Cruise/Loiter Speed 65 KIAS Payload Specs Size: 58(cid:11)L (cid:8)26(cid:11)H (cid:8)27(cid:11)W, (Adaptable); Max Wt. 300 lbs., Payload Power Available 1.8 kW Figure 5. The GA-ASI Ground Control Station with the pilot (left) and co-pilot seats (right) located in the mobile Shipping Size 319(cid:11)L (cid:8)48(cid:11)H (cid:8)57(cid:11)W trailer. Communications link between the pilot and the Navigation Litton LN-100G INS/GPS(P-Code GPS) UAV are provided by a C-band command and control data Avionics GA-ASI PCM, C-Band Line-Of-Sight RF, adapt- link or Ku-band, OTH link. The flight crew utilizes a mov- able for Over-The-Horizon Operations. ing map display (upper left) and a UAV-mounted, forward- looking video camera display (lower left), in conjunction FTS GA-ASIRocket deployed parachute and NASA with attitude and navigation information (lower right) to Flight Termination System maintain the appropriate flight profile. Landing Gear Normal tricycle-type retractable landing gear. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING Month 2003 395 looking real-time video camera data collected on the UAV and streamed to the pilot console via the same C2 link. The video, in conjunction with UAV positional information displayed on a moving map base, allows the crew to monitor the aircraft condition and location at all times. AIRDASImaging Payload The AIRDASthermal scanner (Ambrosia et al., 1994) was flown on the ALTUS®II for the FiREdemonstration (Table 2; Figure 6). The AIRDASwas integrated on the ALTUS in late summer 2001, following system checkout flights on piloted aircraft. Integration was facilitated by the design elements of the UAVas a NASAscientific imaging platform. As such, only minor modifications were made to the platform to allow integration of the scanner. Significant software modifications were made to the AIRDASoperating system to allow remote control and operations by the payload engineer from the ground station. These modifications required the greatest time allocation for integration and mission performance. The AIRDAS has been laboratory-calibrated to resolve fire intensities up to 973 K (700°C). Accurate higher tem- perature discrimination of the thermal channels is possible Figure 6. The AIRDAS system integrated into the GA-ASI but is restricted by the peak temperature efficiency of the ALTUS® II UAV payload pod. The scan head is located in laboratory-calibrated thermal source in use (maximum at the forward portion of the payload bay above the open about 700°C). The NE(cid:5)T of bands 2 and 3 is less than 0.5°C port window, with the computer control rack immediately at 500°C. During AIRDAS science missions, where highly aft of the head electronics. Aft of the control computer accurate temperature discrimination is required, a narrow- is the Litton LN-100G INS/GPS. The forward looking ing band-pass filter is employed for channel three. This video camera can be seen in the nose of the UAV, while filter narrows the channel to 3.95 to 4.05 (cid:12)m, allowing for the command and control communications antenna is accurate temperature discrimination, but restricting the located on the belly of the fuselage, aft of the payload energy transfer to the sandwiched detector channel 4. This bay and ports. The hatch cover for the payload bay (not reduces the signal strength at channel four and precludes visible) incorporates the NERA telemetry antennas on useful data in that thermal region. Integration of non-linear both sides of the airframe. TABLE2. AIRDASSYSTEMCONFIGURATION ANDCHARACTERISTICS FORPAYLOAD response pre-amplifiers for the near-infrared (channel 2) OPERATIONSABOARDGA-ASI ALTUS® II UAV and thermal-infrared (channel 3) allows for a greater range System Composition: Texas Instruments®RS-25 thermal of temperature discrimination than do standard linear cali- line-scanner optics; Non-linear brated pre-amplifiers. The pre-amplifiers allow for two dis- detector pre-amplifiers; Sixteen-bit tinct, non-linear temperature discrimination curves to be Digitizer; Dichroic filters for spectral developed. For low temperatures (30 to 100°C), discrimina- channel separation. tion is approximately 1.0°C/count while, in the higher tem- Control Computer: Pentium®I Pro 233 MHz system; perature portion of the non-linear curve (500 to 600°C), On-Board ETHERNET; SCSI; High- discrimination is approximately 0.2°C/per count in the speed serial ports running QNX OS; mid-infrared thermal channel 3. This design allows for dis- Kingston®18.0 GB removable hard crimination for low temperature earth ambient targets, and drive storage device; Integrated Motorola®Chassis-Mounted GPS also discrimination of discrete temperatures emitted by a Receiver; Crossbow®DMU-FOG-VG variably hot target. Fiber-Optic Gyro, 2-axis For the FiRE experiment, discrete temperature variations were not necessary to convey the fire properties; therefore, Weight and Power: Scan Head: 95 lbs.; Control computer images of relative fire intensity sufficed for data telemetry and peripherals: 70 lbs.; Requires 28V DC @ 20 amps to the disaster managers. Therefore, the narrowing filters were not employed for the FiRE experiment. The fire data Sensor Parameters: Quantization: 15-bit true (signed 16 bit) were portrayed as either single-channel grayscale images FOV: 108 degrees (channel 3 or 4) or as thermal-infrared-visible (RGB) color IFOV: 2.62 milliradians composites. Temperature variations were portrayed as vary- Scan Rate: 4–23 scans/second Digitized Swath Width: 720 pixels ing shades of color in these image files. Spatial Resolution: 26 ft. (8m) at 10K ft. Each of the specific AIRDASbands provides useful NE∆T: Band 2 and Band 3: (cid:6)0.5°C information for fire analysis. The visible band 1 is suitable @ 500°C Temperature for monitoring smoke plumes as well as distinguishing sur- Sensitivity: 1.0˚C/DN / 0.2˚C/DN (non- face cultural and vegetative features not obscured by smoke linear two-step pre-amp; below and or clouds. Band 2 is suitable for analysis of vegetative com- above breakpoint) position, as well as very hot fire fronts, while still penetrat- Thermal Calibration Temp: (cid:1)700˚C ing most associated smoke plumes. Band 2 is sensitive to Spectral Configuration: Channel 1: 0.64 - 0.71 mm fires and hot spots at temperatures above 573°K (300°C) 2: 1.57 - 1.70 mm (Riggan et al., 1993). Band 3 (mid-infrared thermal) is specif- 3: 3.75 - 4.05 mm (narrowing ically designed for estimating high temperature conditions. filter available) 4: 5.50 - 13.0 mm Band 4 is designed to collect thermal data on earth ambient temperatures and on the lower temperature soil heating con- 396 Month 2003 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING ditions behind fire fronts, as well as the minute temperature tennas were selected for integration simplicity, cost, and ad- differences in pre-heating conditions (Ambrosia et al., 1994). equate functionality. The M4 was designed primarily for the The AIRDAS contains an integrated Motorola® chassis- mobile remote satellite voice and data communications mar- mounted Global Positioning System (GPS) unit and a ket. The M4 functions as a “suitcase phone” with a modem, Crossbow DMU-FOG-VG (Fiber-Optic Gyro-Vertical Gyro)® phone handset, and folding, lightweight antennas (Figure 7a). two-axis gyro. The GPS data are integrated into the scanner The INMARSATsystem currently operates at 64Kbs, sufficient output and provide encoded location information on air- for the FIREproject data telemetry activities. craft position to the header file for each flight segment The AIRDAS scene data and navigation file were sent (scan line). The two-axis gyro sends encoded information from the AIRDAS control computer to the NERA system on- on pitch and roll to the control system in order to allow for board the UAV. Data were telemetered through a pair of post-flight correction. A magnetic compass assists in deter- phased array antennas mounted into the skin of the ALTUS® mining heading, allowing for geometric correction. The II fuselage, pre-positioned (in elevation) to acquire signal barometric altimeter data are also incorporated in the lock with an appropriate INMARSAT geo-stationary commu- header. The system accommodates additional serial inter- nications satellite. A custom fairing was built to accommo- faces to integrate other avionics navigation systems on air- date the satellite communications antennas mounted on frames that acquire such information. both sides of the UAV airframe (Figure 7b). These opposing Significant software and hardware modifications were antennas allowed the aircraft to fly in either direction to made to the AIRDAS system to allow UAV remote payload attain proper intercept angles. Orientation of the aircraft control from the GCS. These modifications included stream- flight direction was optimized to attain an antenna inter- ing the system health data from the on-board computer to cept angle of 90 degrees. Proper intercept angles were the payload engineer’s computer on the ground. This allowed determined from known fixed locations of the nearest IN- for real-time monitoring of payload performance, image area MARSAT satellite in the constellation. For the FiRE demon- location, and spatial coverage. stration, both the Atlantic Operating Region–West satellite located at 54° West longitude over the equator and the Remote Operations of Payload Pacific Operating Region satellite at 178° East longitude at The AIRDAS payload engineer and all system controls were the equator were used. Given the fixed location of the satel- collocated in the GCS. Collocation enabled direct commu- lites, and the known location of the UAV, an appropriate nication with the pilots and flight engineer, and access to antenna intercept angle was derived and used to obtain the same forward- and downward-looking video data. A maximum bit rate throughput during data telemetry. downward-looking video camera, placed in the nose of the When signal strength was maximized (through an indica- ALTUS® II, provided real-time coverage information of the tion at the payload engineer’s workstation), a TCP/IPnetwork potential target areas. By incorporating this camera view connection was established and the image data were trans- with the streaming GPS and moving map display, the pay- mitted to either of the appropriate antenna panels. A switch- load engineer could monitor the UAV position and select ing mechanism was created to select the appropriate trans- the optimum flight profile (approach, altitude, and speed) mitting antenna remotely (depending on flight orientation for AIRDAS data capture. and best signal strength intercept angle). The engineer initiated AIRDAS image data gathering after ascertaining the position of the UAV in relation to the Ground Data Distribution Site target (fire) area. When the UAV passed over the approxi- AIRDAS data and the associated navigation data file were mate center of the fire target area (visible on the down- transmitted via a File Transfer Protocol (FTP) login from the ward-looking video display), the engineer saved that AIRDAS UAV through INMARSAT to a SUN SPARC workstation server image stream to a file. The AIRDAS image data were com- located at the NASA-Ames Research Center, Moffett Field, pressed from the signed 16-bit data to 8-bit data and California. The data were accessible immediately at the “remapped” employing a histogram equal probability dis- project website (http://geo.arc.nasa.gov/sge/UAVFiRE/). tribution stretch to portray the appropriate fire information from each band. The resultant file was then jpg-compressed to 720 by 640 pixels by three bands (approximately 100 Kb) to reduce unnecessary data volume prior to telemetry. An associated navigation file of approximately 9 Kbs, com- posed of sensor profile and attitude data (pitch, roll, yaw, time, heading, etc.), was also captured and telemetered to a ground computer server along with the image file. The navigation files were used in AIRDAS image geo-rectification. Following completion of a data collection pass, the resultant image and navigation file were saved for teleme- try. The ALTUS® II UAV was then aligned with a heading necessary to transfer data via the telemetry antenna on Figure 7. (a) NERA Telecommunications M4 WorldCom- either side of the fuselage. During that procedure, the municator® portable satellite terminal phased array payload engineer determined the appropriate INMARSAT antenna. The antennas are approximately 340 mm by satellite for data relay, and initiated a communications 774 mm by 12 mm (depth). The antenna communicates connection. with the INMARSAT series of communications satellites at 1626.5 to 1660.5 MHz. Data throughput is 64 Kbs with Airborne Data Telemetry System near future upgrades to +300 Kbs. (b) The antennas The UAVdata telemetry system was derived from a wireless mobile handheld telecommunications system that communi- (one on each side of the UAV fuselage at the FiRE logo) were installed at an appropriate intercept angle for cates through the INMARSATseries of geo-stationary satel- satellite communications. Data were relayed from the lites. The NERA World Communicator M4 is a portable satel- lite terminal offering ISDN functionality and a pure digital AIRDAS system through this antenna setup, through interface (NERA website URL: http://www.nera.no/ INMARSAT, and to a dialup server site at NASA-Ames. index.html, last accessed 07 August 2002). The NERA an- PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING Month 2003 397 Automated website page refresh updates were performed 0.68 (cid:6)m) to compose the three-band color image (Plates 1 whenever a new image file set was received, updating infor- and 2). Five data collection passes were made over the fire mation at the user’s monitor. Labels for the uncorrected site with data telemetry occurring on the completion of images were provided automatically from the navigation file each pass. (such as date, time, etc). The uncorrected jpg file allowed Real-time quicklook imagery was viewed on the FiRE the disaster manager an opportunity to view the imagery in website by the fire managers immediately following teleme- near real time (within two minutes of telemetry initiation). try. Concurrently, the image and navigational file were acquired from the server by the image-processing operator and integrated with the DACS software to create a geo- Image Geo-Rectification Process rectified image file. Image quality control was performed The image geo-rectification processing was performed on a on the geo-rectified data sets to ensure data integrity prior Dual-Pentium III, 1.0 GHz computer. The image data and to distribution to the fire managers and public. The image navigation file, housed at the NASA-Ames server, were ac- files were formatted to tif and associated tfw files and cessed by the PC via FTP. The two files were then used to transferred via FTP to the NASA-Ames server and website. create the geo-rectified data sets using the Terra Mar Data The DACS processing took approximately six minutes before Acquisition Control System (DACS) software package. The data were sent to the FiRE website. DACS system was designed to receive ephemeris streams The ALTUS® II completed the FiRE data collection mis- from airborne platforms and to rectify the associated image sion and landed at 0930 (PDT) (Figure 8). Within the one file in real time. The DACS system requires platform posi- hour allocated for the FiRE demonstration, the ALTUS® II tion (GPS), pitch, roll, heading, and altitude information in UAV had been launched, attained altitude, and made five order to automatically project and geo-reference the passes over a controlled burn; and the AIRDAS payload data imagery. The terrain-corrected, geo-rectified image files had been telemetered from the UAV to INMARSAT and back were generated using the navigational file (scan line acqui- to the ground, and the data were geo-rectified and were sition time, aircraft altitude, GPS location, pitch, roll, yaw distributed through the World Wide Web where they were angle, and track), sensor attribute and configuration infor- made available to disaster managers around the globe. mation (sensor scan angle and scan rate), and terrain data (USGS DEM data). The navigation data were used with the photogrammetric projective transformation equations Results and Discussion (Wong, 1980) to project the imaged pixels to a ground loca- The UAV FiRE demonstration was highly successful, ex- tion. The aircraft orientation data provided the parameters ceeding many of the objectives of the program, yet the au- necessary to compute the transformation coefficient matrix, thors also note some technology upgrades that could have which then was iteratively corrected for the ground eleva- improved the demonstration. The successes and techno- tion when a DEM was available, or was adjusted to a user- logical barriers that arose are discussed in the following supplied average elevation (Buechel et al., 2001). Symbol- sections. ogy such as fire perimeter delineation, annotation notes, or geo-referencing tick marks may be added to the image as well. A spatial database, Terra-Mar’s Global Data Catalog automatically logged the image information, coverage, and original metadata for ease in referencing and retrieving in- cident imagery. The rectified output file was saved in a tif format with an associated tfw file. Spatial resolution of the geo-corrected data was set at 2.5 meters, matching that attained by the AIRDAS at the platform altitude of 945 meters (3100 ft) AGL. The tfw file was used to describe the geographic position, and contained a listing of variables such as the coordinates of the upper-left image corner, rotational characteristics, and pixel resolution. The common format of the tif and tfw files allowed import into compatible GIS or image process- ing systems. The total time required for these procedures (from sensor data acquisition, satellite uplink/downlink, geo-correction, and distribution via the web to the user community) was under 10 minutes. The geo-corrected AIRDAS tif and tfw data were placed at a central web server and were accessed by the on-site fire managers or by the general public. Data can then be over- lain with other digital data (DRG or DEM) and used in a GIS. Data Collection and Distribution During the FiRE mission, AIRDAS single-band and multiple- Plate 1. AIRDAS data collected from the GA-ASI ALTUS® band color composite images were collected and relayed II UAV flying at (cid:1)945 meters ((cid:1)3100 ft) AGL over the FiRE from the UAV through the telemetry link. The black-and- controlled burn at El Mirage, California on 06 September white images were composed of single-band data sets of 2001 at 0847 (PDT). The scene is composed of AIRDAS either the mid-IR thermal channel (band 3: 3.60 to 5.50 (cid:6)m) bands 4, 2, and 1 (R-G-B). The small fire can be seen in or the long-wavelength thermal channel (band 4: 5.50– white and bright yellow in the cleared area adjacent to 13.0 (cid:6)m). Color composites were also collected and relayed the runway. This scene was the first data set teleme- via the telemetry link. The color composites were com- tered to the ground station and is not geo-rectified. Total posed of thermal, mid-IR, and visible channels of the AIR- collection, compression, and web distribution time was DAS scanner. Either of the two AIRDAS thermal bands were approximately three minutes. combined with channels 2 (1.57 to 1.70 (cid:6)m) and 1 (0.61 to 398 Month 2003 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING Figure 8. The ALTUS® II UAV containing the NASA-Ames AIRDAS imaging payload on low pass prior to landing at the GA-ASI Flight Operations Facility. Flames and smoke from the controlled burn, ignited earlier, can be seen below the aircraft. of remotely piloted aircraft facilitates hazardous operations during critical data gathering conditions. With the ability to switch out pilots easily, the aircraft can remain on- station over a disaster event such as a fire for extended periods of time. This becomes critical in long-duration missions, or where monotonous, long-term data collection Plate 2. AIRDAS data collected from the GA-ASI ALTUS® II missions would stress an aircraft pilot. A UAV operates UAV flying at (cid:1)945 meters ((cid:1)3100 ft) AGL over the FiRE effectively under hazardous conditions (unstable air, large controlled burn at El Mirage, California on 06 September obscuring smoke plumes, and possible rugged terrain 2001 at 0847 (PDT). The small fire can be seen in the conditions) that pose risks to flight crews. circled area adjacent to the runway. This scene is a spa- Integration of the AIRDAS scanning instrument into the tially and geometrically rectified dataset of Plate 1. ALTUS® II was facilitated by the configuration and capacity Lat/Long tick marks (visible around the edges of the of the airframe payload and sharing of electronic and image) were automatically inserted for orientation pur- system command and control links for remote payload poses. The total collection, data compression, telemetry, operation. The AIRDAS, having been optimized for fire and geo-rectification, and data web distribution time was disaster-related imaging, was enhanced further with remote under 15 minutes. operations capabilities. These enhancements will enable integration on other UAVs in the future, simplifying the task of remote operations of the payload. The NERA M4/INMARSAT data telemetry communica- Successes tions link was highly successful in this demonstration and From a UAV operating at an altitude of about 945 meters will continue to prove itself as a useful telemetry system (~ 3100 ft) AGL, five fully geo-rectified fire status images given planned bit rate upgrades and evolution of an im- from an imaging payload were collected, uplinked via a proved and cost-effective airborne system. The antennas satellite telemetry system, and delivered to an Incident that were configured into the UAV payload fairing were Command Center and over the World Wide Web. Our designed originally for remote satellite telephony and data objective to deliver accurate geo-rectified data employing communications and were modified for this demonstration. this methodology in less than one hour was achieved. Further refinements, including an inexpensive tracking an- Overall collection, telemetry, geo-processing, and delivery tenna, are warranted and forthcoming. Data usage costs for were achieved in less than ten minutes once the UAV was the INMARSAT system are reasonable, with anticipated de- on station at the fire site. In a “real” fire event, time con- clining costs over the next several years. Satellite signal sideration for the aircraft to arrive at the site must be ac- “lock-on” was achieved within specifications, although the counted for, although there are no significant speed differ- use of the UAV platform necessitated designing moveable ences in a UAV versus a manned, propeller-driven airframe. antenna mountings and aligning the platform for specific As UAVs attain higher speeds, they will compete with jet satellite signal intercept angles. This is another reason for aircraft for on-fire travel time, but will also have the advan- exploring tracking antenna opportunities to use aboard tage of increased linger and on-station time over fires versus aerial platforms. those possible by manned jet or propeller aircraft. The five Currently, INMARSAT communications are rated at images were obtained over a one-hour flight profile while 64 Kbs, although the actual throughput was approximately the ALTUS®II UAV“lingered” in the airspace above the con- 45 Kbs, due to NERA file information being sent with the trolled burn site. data stream. Larger bandwidth throughput upgrades (greater The GA-ASI ALTUS® II UAV proved to be a capable than 300 Kbs service by 2003/04) to the INMARSAT satellite platform for this disaster support mission. The uniqueness system will enable greater data volumes and minimize file PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING Month 2003 399 compression. The high bit rate, combined with tracking an- AIRDAS data. These enhancements include on-board data tenna configurations, should greatly enhance the ability to compression (from the “raw” signed 16-bit data to 8-bit telemeter data from airborne platforms. data) to efficiently portray the thermal and spectral range The NERA/INMARSAT telemetry system accommodated of the data. Other upgrades include development of various high data rates from the AIRDAS payload. This proved criti- band combinations, and color enhancement procedures. cal in providing multichannel color composite images of The authors are working closely with personnel from the various spectral band combinations of AIRDAS. The multi- disaster community to improve customizable products for channel color image data readily displayed the fire parame- disaster managers that would enhance their value for real- ters, such as fire front, perimeter, and hot spots, allowing time decision making. more rapid decision making on the deployment of resources During the FiRE demonstration, data were delivered to to combat the fire. An increased data throughput (color a NASA server and website for global distribution. This is multi-band images) will increase interpretability of fire not an optimum location for data storage or dissemination. (or other disaster) parameters by the disaster manager or The optimal solution would involve the establishment of a the GIS integrator. central “clearing house” for data from individual fire events. Distribution to multiple users at various Incident Command Barriers Centers located at the fire base camps or other locations Technological and procedural barriers were noted in the can be coordinated by an appropriate disaster agency. This FiRE demonstration and are addressed here. The shortcom- would allow the base camp to be mobile, and allow access ings were primarily in the available technology at the time to the data provided by wireless remote modem connection of the demonstration or involved cost constraints. to the central data server. This functionality would decrease A current barrier to the operational use of UAV technol- the need for full on-site (fire camp) data collection and ogy is the Federal Aviation Administration (FAA) lack of a processing and centralize these processes at an appropriate regulatory framework for UAV operations and certifications. location. These regulatory uncertainties precluded operating the UAV over a larger scale controlled burn in the nearby National Final Remarks and Future Focus Forest or on other State or Federal lands. The UAV industry is working closely with the FAA to develop the regulatory The FiRE project successfully demonstrated the potential framework to allow UAVs unrestricted access in national for utilizing UAVs for real-time remote sensing data gather- airspace. These issues should be resolved in the near future, ing to support disaster management strategies. The capabil- allowing expanded use of UAVs for commercial and ity of UAVs to safely acquire and disseminate data during disaster support activities. hazardous conditions could significantly enhance the dis- Integration of the M4 antennas into the UAV fairing aster management community’s ability to monitor and miti- restricted the aircraft to distinct flight alignment vectors to gate a broad range of disasters. Development and refine- maximize signal strength. In a large fire condition with ment of remote payload operations enhances the use of numerous fire support aircraft in constrained airspace over those payloads on UAVs and other platforms where weight the fire, vectoring the UAV to specific flight paths may or safety precludes the use of an on-board systems engineer. prove to be an unreasonable proposition. Valuable time Rapid development of satellite communications (telemetry) was also lost between the image acquisition time and trans- played a significant role in the FiRE demonstration. The mission of the data. A tracking antenna would allow con- ability to telemeter accurate multispectral imagery to any tinuous collection, communication, and data telemetry location in the world frees the disaster manager from being with INMARSAT without specific UAV flight orientations. “on-site” at the event. Data distribution can therefore be This communications antenna system shows great promise handled at an appropriate disaster facility, allowing multi- as a telemetry link in the near future, and is being exam- ple events throughout the continent to be flown, collected, ined for integration on future UAV missions. and distributed without the need for multiple ground sta- The use of an onboard differential GPS would increase tions and support crew to retrieve the data. Individual fire the accuracy of the platform position, particularly the event data can be distributed to the responsible on-site man- aircraft/sensor altitude calculation. Currently, altitude is agement team. With rapidly increasing data-bit rates, large, determined from the barometric pressure altitude instru- multiband data streams can easily be telemetered from any ment on the aircraft. This information is not of sufficient aircraft to anywhere on the globe. Our success at integrat- accuracy to allow refined geometric/terrain correction of ing telemetry equipment on the ALTUS® II UAV platform, image data. With improved GPS and aircraft pointing vector and the potential to integrate new tracking antenna sys- information, very reliable geo-location ((cid:1)1.0 pixel RMS) can tems, will greatly advance the field of rapid data delivery. be achieved. During this demonstration, geo-location RMS With an increasing bit-rate/decreasing-cost ratio, the next errors were approximately 3 to 4 pixels, sufficient to challenge will not be in data delivery, but in the ability to determine fire location. interpret all the information relayed to the manager and to Image geo-rectification error resides in the quality, de- make rapid, informed decisions during a major disaster tail, and rate at which the navigation information was such as a wildfire. Improvements to the image quality of saved and telemetered. Most of the navigation variables the delivered product will be foremost in the next phase of and sensor/aircraft positional information were collected these activities. Disaster managers and GIS specialists will on a one-hertz (1.0-Hz; 1.0/sec) cycle, while the AIRDAS have to play a key role in defining the variables they need scanning configuration is 4 to 23 scans per second. Al- to make more informed decisions during a disaster event. though navigation data are recorded for every scan line of Although the AIRDAS system has been used extensively sensor data, only every fifth scan line of navigation infor- for data collection over fires, this was the first integration mation was telemetered with the associated image. This of the system on an unmanned platform. Integration of this was done to reduce the size of the navigation file. A or any other imaging instrument was easily accomplished, planned improvement to this procedure is to transmit the given the design purpose of the UAV as a reconnaissance navigation information for every image scan line, thereby platform. If an operational UAV/fire-imaging payload were increasing the geo-rectification precision. required by disaster agencies, integration time and ability Further image utility upgrades are being developed to to rapidly deploy would be similar or faster than those portray fire information more efficiently on the collected same constraints posed by manned aircraft. 400 Month 2003 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING