Airborne Polarimetric Microwave Imaging Radiometer

J.P. Bobak
Remote Sensing Division

D.J. Dowgiallo
Interferometrics, Inc.

N.R. McGlothlin, Jr.
Praxis, Inc.

The Airborne Polarimetric Microwave Imaging Radiometer (APMIR) is a passive microwave sensor being developed by the Remote Sensing Division. The instrument measures naturally occurring microwave radiation from the surface of the Earth and the atmosphere at several different frequency bands and at multiple polarizations. It will be a primary calibration tool for the Special Sensor Microwave Imager Sounder (SSMIS) and WindSat, which is currently under development at NRL. These satellite sensors will provide key data to Navy and Air Force weather forecasters. Since the geophysical parameters of interest have very small electromagnetic signatures, highly accurate sensors, such as APMIR, are needed for calibration of the satellite instruments. APMIR is carried aboard an NRL P3 Orion aircraft, and scans in azimuth and elevation. The movement of the sensor is measured and controlled to a high degree of accuracy. The first flight of the system occurred on December 19, 2001.


Despite unprecedented technological innovations over the past few decades, Navy ships and aircraft are still at the mercy of the weather. Storms can damage or sink ships, or at the very least, slow movements or prevent replenishment at sea. Clouds or fog can make positive identification of ground or sea targets impossible, and can hamper the use of munitions that use optical or infrared guidance. On a larger scale, the oceans and the atmosphere above them are key to the study of global climate. The oceans can store a tremendous amount of energy, and understanding the interaction between ocean and atmosphere is crucial to the global circulation models (GCMs) used to simulate climate dynamics on a planetary scale.

Meteorological forecasting models, whether they produce short-term weather forecasts or long-term climate forecasts, must be provided with inputs such as the amount of water vapor and the wind speed and direction in the atmosphere at various positions. In situ sensors on ships, buoys, or aircraft can only provide point measurements at a relatively small number of locations. These points are also bunched together around shipping lanes and coastal areas, so that vast areas of the oceans are not monitored at all by point sensors. To achieve the necessary coverage, and to do so in a timely manner requires sensors carried on satellites. A variety of different sensors are carried on meteorological satellites, including visible and infrared cameras, microwave radars, and microwave radiometers. Data from microwave radiometers can be used to estimate the amount of water vapor in a column of the atmosphere, the amount of condensed water in the clouds, or the temperature of the sea surface, among other quantities.


A radiometer is an extremely sensitive and stable receiver. As opposed to a radar, which transmits energy to illuminate objects, a radiometer relies on naturally occurring radiation from objects. This radiation is a noncoherent emission of energy at various wavelengths from individual atoms and molecules. All materials, including water and air, give off radiation over a broad frequency spectrum. The shape of the radiation spectrum and its level are determined by the characteristic properties of the material, its surroundings, and its physical temperature. In fact, there is a direct relationship between the physical temperature of the material and the amount of energy emitted. By measuring the radiation from a material at appropriate wavelengths, useful information about the material can be deduced. Measuring the relative amounts of radiation of different polarizations can sometimes provide additional information. The amount of electromagnetic energy given off by terrestrial sources is very small. This means that radiometers have very high gains and consequently must be extremely stable. Radiometers must be calibrated well and frequently during operation. Often a calibration takes place every few seconds. Even so, the instrument must be stable, with only slow, steady changes permissible between calibrations. The operation of the RF electronics must remain within tight constraints. If electromagnetic emissions or reflections from the scene are strongly dependent on viewing angle or polarization, the relative orientation of the radiometer's antenna must be known precisely. The condition of a ground-based instrument can be conclusively verified by subjecting the sensor to appropriate tests. However, access to satellite-borne systems ends at launch, so verifying their operation is more complicated. One of the primary means of checking performance is through intercomparison with other sensors. A variety of other sensors and other platforms are used, including similar instruments on other satellites, airborne sensors, and groundbased sensors. Each of the intercomparisons adds new information and provides a check on consistency. The Naval Research Laboratory has a heritage of performing this calibration and validation. This history includes the enormously successful series of Special Sensor Microwave/Imager (SSM/I) instruments of the Defense Meteorological Satellite Program (DMSP). A new generation of satellite-borne sensors with extremely tight requirements on accuracy will be launched in the next few years. The first of these is the Special Sensor Microwave Imager Sounder (SSMIS), DMSP's follow-on to SSM/I. SSMIS will provide information on such parameters as ocean surface wind speed, rain rate, sea ice, and land surface temperature. SSMIS is scheduled to be launched in October 2002. It will provide operational data to the military, and this information will go into weather forecasts that directly affect operations. Thus, there is a need for great accuracy in the data, which requires a careful calibration and validation program. A second new spaceborne sensor is WindSat, which is currently being developed at the Naval Research Laboratory. The Coriolis WindSat, a joint Navy, National Polar-orbiting Operational Environmental Satellite System (NPOESS), and Air Force mission, will be used to estimate wind direction over the ocean in addition to a variety of other geophysical parameters. It will do this by using fully polarimetric radiometers at several different frequencies. With the full polarization signature, estimation of the Stokes matrix (a measure of the complete polarization signature of the scene) will be possible. Having the full Stokes matrix allows a much better estimate of wind direction to be derived. Because the wind direction signature is smaller than many other radiometric signatures, the acceptable error levels for WindSat are lower than on many previous sensors. Because of the tight error budgets of these sensors, the calibration and validation process will be more difficult. The instruments used for intercomparison must have unprecedented levels of accuracy to achieve the goals set for these satellite radiometers.


Airborne sensors that can underfly the satellites are a primary source of intercomparison data. An aircraft instrument provides data that are most similar to that from the satellite sensor itself. The airborne sensor can view the same area of the ocean as the satellite, at about the same time. The airborne sensor also views the ocean through most of the same atmosphere as the satellite sensor.

For intercomparison to be useful, the airborne sensor must also meet a strict error budget. Beyond the absolute accuracy requirements of the new calibration and validation tasks, an effective sensor must provide information to determine the full Stokes matrix, and must have certain imaging (two-dimensional mapping) capabilities. Scanning over a large range of azimuth angles is essential for viewing the largest amount of surface area possible in a given amount of time. The footprint of each satellite radiometer is tens of kilometers on a side. The aircraft sensor has a much smaller footprint (around 2 km). Taking data only over a strip an order of magnitude smaller than the satellite sensor footprint might give a misleading indication of the signature of the whole footprint. Also, as many independent satellite footprints as possible should be viewed before prevailing conditions change significantly (in the absence of fronts, an hour or two is a reasonable period), so flying multiple passes over the satellite footprint along different bearings is not feasible. Scanning in azimuth gives a much better indication of conditions over the whole satellite footprint. Additionally, having a full 360-deg azimuth capability assures that the spaceborne and airborne sensors have the same azimuth angle view of the scene for at least some of the samples. This is an important consideration because of the direction-dependent signature from the ocean surface. The ability to take data at variable incidence angles is needed to match the various incidence angles of the satellite sensors. There is a strong dependence of signature on view incidence angle, and even small differences have a large effect.

Based on these needs and a survey of instruments that existed at the time, the construction of a new radiometer system, specifically designed to meet the requirements of these satellite sensors, was the best solution for the calibration/validation needs. The new sensor was designed from an error budget that allotted a maximum amount of uncertainty in antenna pointing, a maximum rate of temperature change for the electronics, and a minimum linearity for the detectors, along with a number of other parameters. The design took advantage of much that had been learned from the design of the satellite systems.

Because of the challenging nature of the accuracy requirements, a truly interdisciplinary effort was needed. The electronics were carefully chosen to provide high, stable gain and low power consumption. The mechanical structures need to support precise and repeatable antenna pointing. Tight temperature control is critical to keep the radiometer electronics from drifting. The software is required to position the instruments precisely, control the operation of numerous switches and digitally-controlled settings, and process the large amount of data coming from an entire suite of sensors. The motion control system must be robust enough to ensure proper pointing of the sensors even while they are being buffeted by the wind stream. The electrical layout and design of the system has to be done within tight packaging constraints.


APMIR includes channels from 5.8 to 37 GHz (C- to Ka-band). This range of frequencies covers the lower bands on SSMIS (19.35, 22.23, and 37 GHz) and all of the bands on WindSat (6.8, 10.7, 18.7, 23.8, and 37 GHz). Each of these channels provides vertical and horizontal polarization data, and several are fully polarimetric. Table 1 shows the specific frequencies and capabilities of each channel.

Table1 Image

Both internal and external calibration is used. External calibration is by means of two blackbody targets. Each target is a collection of metal pyramids completely covering an octagon about 50 cm across. These pyramids are coated with a material that absorbs microwave energy. The dimensions of the pyramids and thickness of coating is chosen such that at the frequencies of interest, very little reflection occurs from the surface of the target. The target radiates energy at a rate approximating a blackbody at the same physical temperature. One of these targets will be heated to 313 K, while the other will be at approximately the temperature of outside air, which can be 233 K or colder at altitude. The two targets will provide a two-point calibration of each complete radiometer chain.

Internal calibration is accomplished by switching between the antenna and internal calibration sources at the input to the RF gain chains. The internal calibration sources produce a known amount of noise that can be related to an equivalent scene temperature. On APMIR, noise diodes and matched loads are used as the internal calibration loads. The internal calibration cannot account for changes in the antenna (since the calibrated noise is injected into the system downstream of the antenna). However, it can account for changes in the active portion of the radiometers (amplifiers, analog-to-digital converters, etc.). These components are more prone to changes in performance than the passive components in the front end. Calibrating internally requires only switching a latching circulator or PIN diode switch rather than physically moving the radiometer antenna from the scene to a blackbody target. Therefore, less scene data are lost, and internal calibration can be done more frequently than external calibration.

The radiometers are housed in an aluminum sphere with a diameter of approximately 90 cm. The weight of the sphere, including the internal electronics, is 140 kg. The radiometers are mounted in a rack made of extruded aluminum components. The rack was designed to be extremely rigid to minimize antenna movements resulting from flexure of structural components in the wind stream. Figure 1 shows the internal configuration. There are three radiometer enclosures. Each box contains one or more radiometers, including the antenna, RF components, video circuitry, analog-to-digital converter, and microprocessor unit. One box contains the 37 GHz radiometer. A second enclosure houses the 18.7/19.35 GHz and 22.235/23.8 GHz radiometers (these radiometers share a horn antenna). The 6.8 GHz (including the 5.8 and 7.2 GHz systems) and 10.7 GHz systems are in the last box (also sharing a single horn). A fourth enclosure contains DC-to-DC power converters that accept power at 28 V DC from the cabin of the aircraft. This power may be contaminated by noise as it crosses the slip rings. Inside this box, the power is filtered and converted to a variety of different voltage levels needed by the radiometers. Thermal blankets have been made for each radiometer to help maintain a constant temperature. The antennas are recessed into the sphere. To maintain the spherical profile in the wind stream, the openings in the sphere are filled with a closed-cell, low-loss foam. This material also provides thermal insulation for the antennas. The foam is very strong, and causes little distortion to the radiometric signals.

Fig1 Image

Exploded view of the sphere and internal components. The radiometers are the blue boxes in the rack, and one of the foam inserts is shown in green.

The sphere is mounted in a yoke that gives it two degrees of freedom. The system scans through a full 360 deg in azimuth, and the elevation control allows views from nadir to inside the bomb bay of the aircraft for external calibration. This results in unobstructed view incident angles from 0 to 60 deg. The yoke, in turn, is mounted on a pallet in the bomb bay of a Lockheed P3 Orion.

The aircraft will typically fly at an altitude of 7.6 km (25000 ft) or higher and at a speed of 140 m/s (270 kts). Under these flight parameters, the sphere will rotate in the azimuth direction at around 10 rpm for imaging. Every few minutes, the sphere will rotate in the elevation direction to look at the two external calibration targets, which are attached to the portion of the yoke inside the bomb bay of the aircraft. The sphere can continue its azimuth motion while it looks at the external targets, as these rotate with the sphere. Figure 2 shows the sphere in the pallet.

Fig2 Image
The yoke as it is mounted in the bomb bay pallet. The coverings of the sphere and yoke are removed to show internal detail.

A geared brushless motor moves the sphere in elevation via an electronic clutch and a worm gear. A custom brake is used to hold the elevation position of the sphere during data taking. Azimuth motion is accomplished by another geared brushless motor that drives a chain and sprocket system. The position of the sphere in each of the degrees of freedom is determined by resolvers. The resolver provides positional information to a few hundredths of a degree and is rugged enough to withstand the operational environment.

Inside the cabin of the aircraft are two 19-in. electronics racks that contain the system computer, the motion control electronics, linear power supplies that supply the sphere with 28 V power, and aircraft position and attitude indicators. The position and attitude information will be measured by a GPS-based system that is described in greater detail below. GPS antennas will be mounted at two places on the fuselage and on each wing for determing attitude.


Simultaneously meeting the requirements of the error budget, the need for a rugged, dependable system that could provide good data under a variety of extreme environments, and the difficulties of building an instrument that would be approved for flight on a Navy aircraft (to say nothing of budget and schedule constraints), led to a number of innovative solutions. A small sample of these are described below.


One major challenge was finding a material that could be used to maintain the profile of the sphere in front of the antennas. The antennas are recessed into the sphere due, in part, to the need for them to be parallel to each other. Finding a material that would not severely degrade the performance of the antennas, while being strong enough to survive the rigors of flight was a difficult task. Adding the requirements of availability and price made it even more difficult. After an extensive search, the mechanical team found a suitable material. It is a low-loss, closed-cell foam. The foam was chosen for several of its properties that include: low dielectric loss, high heat transfer resistance, chemical stability, closed-cell structure, and relative ease of machining.

The low dielectric loss allowed the material to be placed between the radiometer antennas and the scene with an acceptable distortion of the signal. This was verified during tests in an anechoic chamber. The high heat transfer resistance (synonymous with thermal insulation) allowed the foam to be used as an insulator in several locations. The foam provides insulation in the front of the calibration loads. Ultimately, liquid-nitrogen-cooled loads are planned for APMIR. To prevent condensation from forming on the loads, which would lead to bad calibration data, the temperature on the surface of the load should be above the local dew point. This will be accomplished by insulating it sufficiently. The outside surface of the insulation will be close to ambient, but the microwaves will pass through to the target itself, which is much colder. The foam is also the material sewn into the insulating blankets for the individual radiometer enclosures.

The chemical stability and closed-cell structure provide an effective environmental seal and protection for the radiometer antennas. This is very important for an instrument mounted on a maritime aircraft.

The durability of the material allowed it to be shaped without disintegrating (as many of the materials tested did). During flight, the foam is subjected to a several-hundred-knot air stream, so it must be tough, but able to be machined to a complex shape. Foam slabs were first cut into circles that were glued together into cylinders. These were profiled by hand to match the external profile of the spherical radome by using multiple processes including cutting, sanding, milling, and carving.


State-of-the-art polarimetry involving aircraft platforms requires highly accurate attitude and position information for data processing and error correction. The angular pointing of the antennas must be known to within a few hundredths of a degree. Measuring angles to this level of precision on a system that may be moving in two degrees-of-freedom and is slung beneath an aircraft, which itself is rolling, pitching, and yawing, is not a trivial task. The solution for APMIR is a multilayer system that measures the position and attitude of the aircraft with respect to the Earth, measures the position of the sphere with respect to the aircraft, and measures the position of each antenna with respect to the sphere.

For the aircraft, the inertial navigation system (INS) currently in place is not sufficiently accurate for APMIR. New systems are very expensive, and gyro systems in general are subject to drift errors that can accumulate during extended operational maneuvers. A GPS system that uses four antennas to determine attitude was identified. The cost is a fraction of the price of popular gyroscope-based systems, but with comparable attitude accuracy and no drift problems. Additionally, GPS time-tagging and velocity information is available from the system.

The position of the sphere with respect to the aircraft is measured by a pair of two-speed resolvers. These are similar to encoders, but are magnetic rather than optical. This makes them more robust, which is an important consideration for airborne instrumentation.

Fig3 Image
Part of the laser alignment involving the 6 and 10 GHz antenna.

The final link is accomplished via an inclinometer cube. A cube is mounted to each radiometer antenna. The cube consists of three inexpensive, but highly accurate inclinometers that have been positioned to measure key angular parameters: Earth incidence angle (pitch), polarization rotation angle (roll), and scan azimuth angle (yaw). When incorporated into the APMIR radiometer sensors, the cube is positioned directly to the orthomode transducer (OMT) for alignment with the electrical axes of the sensor's horn. Once affixed to the horns, the inclinometer cubes are calibrated by laser alignment (Fig. 3). Horn pattern measurements are made with the inclinometers in place so that the pattern can be exactly related to the scene via the inclinometer readings.


Temperature control is a crucial element for stable and repeatable radiometry. The performance of amplifiers in particular is strongly dependent on physical temperature. A specification of 0.02C maximum change over a period of 8 s was set to allow these gain changes to be calibrated out. In addition to thermal stability, a circuit that contributes little noise with small power changes is preferred. Often the turning on and off of a temperature controller is readily apparent in the data.

In an effort to maintain the highest quality data, a custom temperature control circuit was developed for APMIR. A "soft" heater control circuit was designed and built by NRL's Passive Microwave Section. The circuit is considered a "soft" controller in the sense that voltages fed to the heater elements are gradually increased or decreased in an analog fashion. The circuit functions by comparing a setpoint temperature value against a thermistor, and amplifying the difference with a power transistor. The power transistor drives a series of custom rod heaters that have been embedded into the component base plate for even heat distribution. A thermal switch is added in parallel with the power transistor so that the set point temperature can be achieved quickly by driving the heaters with full power to within a few degrees of the set point. The quick ramp-up allows the radiometers to stabilize in 45 min and enter a low power draw mode for maintaining temperature. A mockup of a radiometer enclosure was tested in a thermal chamber. Even with a difference of 80C between chamber air and the temperature of the inside of the enclosure, the temperature stability easily surpassed the desired specification.

Fig4 Image
APMIR installed on Flight Support Detachment P3 before first flight, December 19, 2001.


APMIR flew for the first time on December 19, 2001 (Fig. 4). The flight was a complete success. The 19, 22, and 37 GHz radiometers operated in a dual-polarized mode. First underflights of SSMIS are scheduled for November 2002.

In 2002, the 6 and 10 GHz radiometers will be built, and the polarimetric capability of the radiometers will be fully implemented. Additionally, the final components of the pointing determination system will be put into place on the aircraft.

[Sponsored by ONR, DMSP, and NPOESS]

Link to Highlights of NRL's First 75 Years

Sign Up for NRL News