Civilian Uses of Surveillance Satellites

Early DSP satellite being deployed from the Space Shuttle cargo bay. Following release from the Shuttle, an upper-stage rocket motor propelled the satellite to its operational geosynchronous altitude. Following the Shuttle Challenger accident, DSP satellites were subsequently boosted into orbit by unmanned rockets. (NASA)

Civilian Uses of Surveillance Satellites

Dee W. Pack, Carl J. Rice, Barbara J. Tressel, Carolyn J. Lee-Wagner, and Edgar M. Oshika

Every 10 seconds nearly the entire Earth's surface is scanned by Defense Support Program (DSP) infrared surveillance satellites looking for the telltale signs of hostile missile launches. The Aerospace Corporation has been investigating the feasibility of using this existing capability to detect natural disasters and other related environmental phenomena.

For the past 6 years, Aerospace researchers have pursued a systematic program to explore the possibilities of using DSP satellites to detect and study fires and volcanic activity. Case studies have included wildfires such as Southern California's Topanga-Malibu fire of 1993, biomass burning in the Southern African savannas, volcanic eruptions and the spread of ash clouds from Mount St. Helens in Washington and Columbia's Nevado El Ruiz, and recent activity at the volcano Popocatapétl in Mexico. These results are being used to assist in the development of the Hazard Support System, a new disaster detection and mitigation program recently established by the National Reconnaissance Office (NRO) and the United States Geological Survey (USGS).

The DSP Satellite System

First launched in 1970, DSP satellites are the space-based component of the nation's missile early-warning system. These geosynchronous satellites observe Earth, using a spinning array of infrared detectors that are sensitive to emissions from hot point sources at or near Earth's surface.

DSP's unique 10-second revisit rate is unprecedented. In contrast, the current GOES (Geostationary Operational Environmental Satellite) weather satellite images the United States every 15 minutes in its standard scanning mode and provides a full disk image of Earth only every 3 hours.

Polar orbiting meteorological satellites—such as the National Oceanic and Atmospheric Administration's (NOAA) Television Infrared Observation Satellite (TIROS) and DOD's Defense Meteorological Satellite Program (DMSP) satellites—have sensors that image midlatitude regions only twice a day, while Landsat Earth resource satellites can revisit a given location only every 16 days.

The globally deployed DSP constellation could supplement current civil agency environmental observation systems by providing infrared monitoring with continual, nearly full Earth coverage. In addition, the wide dynamic range of the DSP sensor allows the dynamic changes of intense fires and volcanic activity to be studied without sensor saturation.

Aerospace DSP Groundstation for Environmental Monitoring

In recent years, our environmental monitoring work was facilitated by the construction and operation of a DSP groundstation at Aerospace corporate offices in El Segundo, California. This groundstation, known as the A8 Research Center (named for the facility's location) or ARC, emulates the DSP architecture and Space Based Infrared System (SBIRS) technology path.

Aerospace researchers have determined that DSP satellites can detect wildfires soon after they break out. In remote areas where "911" is not available, this early detection capability can aid in the rapid deployment of firefighters to the scene. (U. S. Forest Service)

ARC, which integrates commercial, government, and Aerospace-developed hardware and software, was assembled at relatively low cost. It is well-suited to data processing and fusion, algorithm development, and proof-of-concept demonstrations for both military and civil applications. The use of a stand-alone research and development groundstation ensures that our research has no impact on primary DSP mission operations, and that we are not affected by configuration changes at the operational sites.

We successfully used DSP satellites that had exceeded their operational life and had been sent into higher, so-called supersynchronous, orbit. This would not have been possible without a dedicated direct downlink facility. We also processed real-time data from several other satellite programs (GOES, TIROS, DMSP) and carried out sensor performance comparisons and real-time data overlay studies. Our main focus, however, has been testing the scientific and civil mission utility of DSP.

Our investigation of the application of DSP infrared remote-sensing capabilities to Earth observation has had two main thrusts: natural hazard detection and environmental observation. The missions associated with the two main thrusts are timely hazard warning and longer term monitoring to study the scale of global fire and volcanic activity. The former usually involves rapid detection and identification, while the latter employs continual synoptic DSP observations over long periods of time. The satellite capabilities investigated fall into several categories:

Fire detection
Fire monitoring, including observations of widespread biomass burning
Volcano eruption detection, including lava flows and explosive eruptions
Volcanic activity monitoring
Volcanic ash cloud detection and tracking for improved aviation safety

Wildfire Detection

We studied the use of DSP for wildfire detection, both at urban-wildland interfaces and in more remote regions. The goal is to quantify the utility of the 10-second revisit rate offered by DSP's geosynchronous Earth scanning satellites.

We established that these satellites possess a useful capability under dry atmospheric conditions without significant cloud cover. These conditions correspond to the severe fire danger conditions typical, for example, of the "Santa Ana" weather phenomenon that occurs during the Southern California fire season. Santa Ana winds are the dry hot winds that often accompany a high pressure system over the Great Basin in the Western United States.

A case study of such a fire was the observation of the Topanga-Malibu fire of 1993. During late October through early November 1993, more than 21 major wildfires burned over 200,000 acres in Southern California. At the height of the activity, more than 10,000 firefighters were deployed. The fires caused four deaths and more than 150 injuries, destroyed more than 1,200 structures, and caused over a billion dollars in property damage.

Six southern California counties were declared disaster areas. Approximately half of the fires are believed to have been caused by arson. The fires started during two main periods, October 26–27 and November 2, during Santa Ana conditions characterized by low humidity, high temperatures, and dry, gusty, northeasterly winds.

"This fire is headed to the coast!" Radio message received by Los Angeles Fire Department officials from firefighters battling the Topanga-Malibu blaze on November 2, 1993.

Several features are apparent in the intensity vs. time plot for the Topanga-Malibu fire that, based on our subsequent experience, appear to be typical of the signature of such wildland and urban/wildland interface fires. First, a plateau of linear intensity growth occurs following the start of the fire. Then a rapid exponential increase in intensity occurs due to a sudden increase in temperature and fire area. This transition to exponential intensity growth corresponds to the point at which the fire erupts suddenly into a firestorm. It was at this point that firefighters reported 30- to 40-foot flames raging in an area of chaparral that had not burned since 1926. The grim assessment: "This fire is headed to the coast," was radioed to officials.

A prominent intensity dip occurs later in the data. Such intensity changes are also typical of these severe fires. The dip could be due either to obscuration by smoke or to a temporary fuel shortage caused by the burning out of one topographical area before the fire's wind-blown spread to another unburned region—for example, when cresting a hill. The latter explanation is believed to be more likely in this particular case. The early plateau, rapid rise, and the up-and-down intensity structure are characteristic of all the serious brush fires studied.

To quantify the potential utility of DSP, we first needed to determine how quickly the fires became visible, so the days on which the fires started were our highest priority. It was also important to identify those characteristics of the signals that could be used to develop automated or semiautomated detection and identification procedures. In addition, it was of interest to track the growth of the fires as they spread and to create as complete a record as possible.

To determine how rapidly individual fires became detectable, we examined the DSP data for the geographical areas around the individual fires before and after their start. The Topanga-Malibu fire was reported to 911 operators at 10:45:10 a.m. As with all 911 calls in Los Angeles County, the call and the time were automatically recorded. This fire started within sight of a number of homes and is believed by arson investigators to have been reported almost immediately.

Infrared data points showing a fire in this location first appear on a DSP space sensor return at 10:48. Automated detection is triggered by the continuous presence of hot cells within the fire area and could have occurred as early as 10:50 a.m.

Analysis of DSP data for numerous fall 1993 fires shows that the first detectable infrared signals were detected from 0 to 14 minutes after the 911 call reporting a fire had been received, with automated detection lagging ground reports by 3–15 minutes. In densely populated areas such as Los Angeles County, where such fires can occur terrifyingly often and people are alert to their danger, the 911 system serves as a lifeline to the fire department. But in more remote areas where 911 is not available, DSP's capability for rapid automated fire detection makes a strong case for its use.

The data recorded by DSP of the Topanga-Malibu blaze provide a full record of space-based infrared intensity observations of a brush fire as it evolved into a firestorm. To our knowledge no comparable continuous observations of a severe wildfire's radiant intensity growth have ever been recorded by other space- or aircraft-based platforms. This illustrates the unique capability of a geosynchronous space platform to provide global coverage for disaster monitoring and characterization.

Since the fires of 1993, extensive additional fire detection testing involving DSP and civil satellite systems has been carried out by USGS, NOAA, U.S. Forest Service, Bureau of Land Management, state of Georgia, County of Los Angeles, and university and industry participants. These tests involved both prescribed fires and blind tests of the system's capability to detect naturally occurring fires.

Biomass Burning in Africa

Global biomass burning is a significant contributor to the emission of greenhouse gases such as carbon dioxide that may have an impact on Earth's climate. In the African savannas, significant burning occurs every year during the dry seasons. The grasslands are deliberately burned by the populace for forage improvement or other agricultural purposes, such as clearing land. Recent studies using polar orbiting satellites have suggested that global biomass burning, and activity in the African savannas in particular, is much more widespread than previously realized, and hence a more significant contributor to global greenhouse gas emissions.

GMT = 2:17	GMT = 6:58	GMT = 8:58

GMT = 10:58	GMT = 13:00	GMT = 15:00

GMT = 17:08	GMT = 20:08
		Images of infrared intensity recorded on July 3, 1992, during the course of a 24-hour cycle of burning across the whole of Southern Africa. Time is given in Greenwich Mean Time (GMT). Numbers on axes refer to degrees of latitude and longitude.

DSP has observed seasonal fire activity in Africa similar in nature to that observed by the infrared sensors on the polar orbiting TIROS and DMSP satellites. The northern savannas show peak activity during January and February, while the southern savannas exhibit peak fire activity during July and August. This peak activity corresponds to the dry seasons of these two regions. While the polar orbiting satellites can map out regional fire activity and discern seasonal patterns, they are unable to monitor the diurnal (day/night) cycle of fire since they pass overhead only twice a day and measure only a strip of Earth's surface. DSP's fast revisit rate and nearly complete continual hemispheric view were used to unique advantage to reveal this diurnal cycle of continental-scale burning.

To study diurnal variation in fire intensity, sensor noise was first suppressed through intensity thresholding. The noise-reduced data were then binned in equal area grids and the intensities of active cells within each grid were summed for a 3-minute period (18 scans) every half hour for a 24-hour cycle.

The total integrated radiant intensity across the whole of the southern African savannas can be plotted as a function of time, yielding the complete diurnal cycle of biomass burning. To create these curves, data for a given sampling time was summed across the full swath of the African continent for the region from 5 degrees South to 20 degrees South (note that this excludes the burn activity in South Africa and the island of Madagascar). The total observed infrared intensity changes by a factor of approximately 2500 over the course of the day. This is the first time this behavior has been accurately measured over the African continent, though such a variation has been hypothesized. The results show a diurnal variation of Gaussian appearance. This daily rise and fall in fire activity is believed to be caused by two factors: (1) human fire activity and work patterns, and (2) increasing air temperature and wind activity in the midafternoon, which lead to more intense burning. Differences between the two sensors reflect different effective thresholds; the basic shape of the curve is the same for both.

The images of infrared intensity shown at the bottom of pages 4–6 were recorded on July 3, 1992, during the course of a 24-hour cycle of burning activity across the whole of Southern Africa. Initially the central grasslands are quiescent, but as the day progresses quite widespread infrared emissions become evident. These peak in the afternoon (approximately 2 p.m. local time) and fall off distinctly as the day ends.

The spatial distribution of fire activity in the peak burning periods shows close agreement with results from polar orbiting satellites. In particular the curved boundary of fire activity northeast of Angola (approximately 7.5 degrees South, 22 degrees East) is noticeable. This marks the northern edge of areas with extensive grassland burning.

A New Tool for Volcanologists

Volcanic activity observed by DSP sensors has included explosive eruptions, ash emission and ash cloud drift, and caldera and lava flow activity. These data demonstrate the possibility of exploiting DSP satellites to improve eruption detection, to improve global volcano monitoring, and to augment civil system capabilities for ash cloud detection and tracking. This last application is important since volcanic ash clouds pose a hazard to aviation traffic.

The Colombian volcano Nevado El Ruiz erupted explosively at night under heavy cloud cover on November 13, 1985. The eruption melted an estimated 10% of the volcano's ice cap. Subsequent lahars (mud and ash flows) killed an estimated 25,000 people 2 1/2 hours later when the town of Armero 74 kilometers away was inundated. Data saved by Aerojet Corporation analyst Charlotte Decker yielded dual satellite observations when examined later. These space observations are in close agreement with ground observer reports and arguably provide the most accurate initiation time of the eruption.

Currently only 10 percent of Earth's active volcanoes are well monitored. Thus the fast sampling infrared measurements taken by DSP surveillance satellites provide an exciting new volcanology tool. Such measurements have been used for the detection of an explosive eruption at the Colombian volcano Nevado El Ruiz, continual monitoring of Mexican volcano Popocatapétl during a period of moderate eruptive activity, and the rapid detection and tracking of the Mount St. Helens ash cloud.

Space-based observations of explosive volcanic activity offer an opportunity to improve global monitoring of dangerous volcanoes. DSP sensors offer another tool to augment seismic and other monitoring methods.

Recently, our research has focused on the well-instrumented volcano Popocatapétl, which is 5452 meters high and only 60 kilometers southeast of Mexico City. Because of the volcano's proximity to population centers, the Mexican government's Centro Nacional de Prevención de Desastres (CENAPRED) has implemented a vigilant warning program to inform the nearby citizens of volcano danger levels and threat status. The volcano also poses an ash cloud threat to local aviation, so ash warnings are issued both by Mexican officials and NOAA.

Among the sensors are remote video cameras, tilt meters, seismometers, and lahar (mud and ash flows) detection devices. Continual monitoring of the volcano's infrared activity from DSP satellites started December 2, 1998. The process of comparing our measurements with CENAPRED data has begun, initially using data available on the Internet.

A 24-hour period of moderate volcanic activity at Popocatapétl, Mexico. Repeated short infrared transient signals are periodically observed. These are believed to be moderate volcanic explosions that spew out incandescent rock, steam, hot gases and ash. During daylight hours these transient signals from hot gas and ejecta are also accompanied by longer-lasting lower intensity signals from solar scattering off the lofted ash cloud. The image at left shows a view of the summit taken by CENAPRED's remote video camera 7 minutes after the 18:07 eruption. (Video image courtesy of Universidad Autonama de Mexico)

Comparison of DSP remote sensing data with ground observations of Popocatapétl is being pursued to test system capabilities for continual monitoring of active volcanoes. Interesting possibilities are indexing eruption intensity from the calibrated infrared data, gathering explosion activity statistics, and examining the utility of these data (in combination with seismic data) for predicting serious eruptions.

DSP observations of the eruption of Mount St. Helens have been reported in the scientific literature. DSP's rapid revisit capabilities enabled researchers to identify multiple explosions within the initial eruption event, as well as to trace the rapid expansion of the ground cloud that overwhelmed the area north of the volcano.

Continual monitoring of the high altitude Plinean eruption column permitted a detailed evaluation of propagation of the massive ash cloud that resulted, including delineation of multiple plumes. The additional capability that DSP provides to current ash cloud detection capabilities is the timely initial observation and tracking of daylit eruptive ash clouds.

Tracking the rapid growth of the eruptive ash cloud during the first 40 minutes of the May 18, 1980, eruption of Mount St. Helens. Vectors show the velocity of winds aloft at three altitudes. The region in red shows the extent of blast deposits. (From R.S.J. Sparks, J.G. Moore and C.J. Rice, "The Initial Giant Umbrella Cloud of the May 18th, 1980 Explosive Eruption of Mount St. Helens," Journal of Volcanology and Geothermal Research, Vol. 28, page 257, 1986).

Outlook

DSP satellites provide a powerful infrared remote sensing resource for detecting and monitoring global fire and volcanic activity. Their unique capability for rapid continual sampling, the full Earth coverage, and the global deployment of DSP sensors can complement the capabilities of civil satellite systems.

Our research has demonstrated significant potential both for rapid detection of infrared emissions associated with various events such as fires and volcanic eruptions as well as longer time scale infrared monitoring for scientific research purposes. Exploring the scientific and disaster warning utility of this type of timely, fast-sampling, wide dynamic range infrared sensor is also useful for assessing possible applications of new-generation military sensors such as SBIRS or proposed sophisticated future civil systems.