2. Technical Overview

This section provides an overview of the science and technology behind geophysical classificationThe process of making principled decisions, using data collected by geophysical sensors, to differentiate between buried items that are potentially hazardous and those that can be safely left in the ground during munitions response actions..

2.1 Introduction to Geophysical Classification

Geophysical classification as applied on a munitions response project consists of the following four steps:

  1. Detection. Detection involves the initial mapping and identification of buried metal objects on the site. Often referred to as digital geophysical mappingMapping data generated from a geophysical system that digitally records geophysical and positioning information to support initial mapping and identification of buried metal objects on a site. (DGM), this step can be accomplished with traditional or advanced geophysical sensors. The detection process is discussed in detail in Survey of Munitions Response Technologies (UXO-4).
  2. Cued data collection. In this step, a richer data set is collected by positioning an advanced EMI sensor over each buried metal object detected, and then collecting 60–120 seconds of data. As the technology develops, this step may be combined with the detection step, requiring the collection of fewer cued data and reducing data collection to one mobilization.
  3. Parameter extraction. Software analysis of the advanced sensor data estimates the intrinsic characteristics of each metal object. These estimated characteristics relate to the objects’ size, shape, symmetry, aspect ratio, wall thickness, and material composition.
  4. Classification. In this final step, the estimated characteristics are used to classify each buried metal object as either a potential munition that must be removed from the site or a fragment or piece of debris that can be safely left on the site.

2.1.1 Sensor Basics

Although successful geophysical classification of munitions using advanced EMI sensors has only recently been demonstrated, the underlying physics governing the building blocks of EMI sensors were explored and understood almost 200 years ago.

In the case of a basic metal detector used by treasure hunters and other hobbyists, an electrical current is sent through a circular coil, which is swept near to the ground by the operator. If a metal object is in the immediate vicinity of the coil, electrical currents are set up, or induced, in the metal object. The induced electrical currents in the metal decay over time (milliseconds [ms] time scale) and, in turn, generate secondary magnetic fields. The secondary magnetic fields induce currents to flow in the receive coil. The receive coil currents are measured, and the presence of the buried metal is indicated to the operator by a visual or audible signal. If there is no metal in the vicinity of the sensor coil, no secondary magnetic fields are generated (Figure 2-1) and no currents are induced in the receive coil.

Figure 2-1. Operation of a basic metal detector.

Advanced EMI sensors that are designed to classify munitions vary in size and design, as shown later in this section, but are fundamentally simple and share the common components of the basic metal detector described above. The primary difference between basic metal detectors and the advanced EMI sensors used in classification involves the sophistication of the coil design. Basic metal detectors use a limited number of single-axis coils in a simple design; historically, the single axis EM61 cart was the most commonly used sensor to collect geophysical data on a conventional munitions response project. Advanced EMI sensors, on the other hand, make use of many transmit and receive coils that are rigidly assembled in a fixed-array configuration. The advanced sensorsMunitions-classifying sensors that are designed with many transmit and receive coils rigidly assembled in a fixed-array configuration. The combination of multiple receive coils, large bandwidth electronics, and supporting sensor data results in the collection of significantly more data than can be collected with single-axis EM61 sensors.’ coils also use low-noise digital electronics with large bandwidths to receive and record the induced currents.

The importance of multiple coils is illustrated in Figures 2-3 and 2-4. With a single-coil (single-axis) sensor (Figure 2-3), the orientation of the exciting field is one direction only and multiple measurements are required to excite all axes of the object being interrogated. A multiaxis sensorAdvanced EMI sensor with excitation and receive coils arranged to interrogate a buried object along multiple axes from one measurement location. (Figure 2-4) can excite all axes of the object from a single measurement position without the need to stitch together the data from multiple measurements, each of which contains uncertainties in location and orientation.

Figure 2-3. Single-axis sensor.

Multiple measurements are required to completely interrogate an object with a single-axis sensor.

Figure 2-4. Multiaxis sensor.

A multiaxis sensor can completely characterize an object from a single measurement position.

The combination of the multiple receive coils, large bandwidth electronics, and supporting sensor data of the advanced EMI sensors results in the measurement of significantly more data than those collected with single-axis EM61 sensors. Figure 2-5 illustrates and compares the data volumes acquired at a single measurement point by both an advanced and traditional sensor.


Figure 2-5. Data points collected at a single location.

For each location at which data are collected, an advanced sensor collects vastly more data points (all of the data points that make up the graphs shown on left) than does a traditional EM61 sensor (the four points shown on right).

2.1.2 Parameter Extraction

The measured decays from an EMI sensor are a function of the intrinsic properties of the object being interrogated as well as its orientation and distance from the sensor. In other words, a single item can produce a wide variety of signatures, depending on the way it is buried (Figure 2‑6).


Figure 2-6. Measured decays from the TEMTADS 2x2 sensor based on location and orientation of the munition.

The decays resulting from the same object vary widely as the object is moved under the sensor.

To extract information that is intrinsic to the object, munitions classification using EMI data requires an analysis step that removes the effects of burial depth and object orientation and derives target response signatures that depend only on the object itself (intrinsic parametersIntrinsic characteristics of a buried metal object, including size, shape, symmetry, aspect ratio, wall thickness, and material composition.). This is accomplished by fitting the observed data to an EMI response model (a process geophysicists refer to as inversionFitting measured sensor data to an EMI response model (commonly the dipole model) to obtain the model parameters, including the object's location and depth, orientations of its principal axes, and its principal axis response functions.) to obtain the model parameters (the object's location and depth, the orientations of its principal axes, and its principal axis response functions) needed to reproduce the observed EMI data. This process is illustrated in Figure 2-7. The location, depth, and orientation parameters can be used to guide any subsequent excavation, while the principal axis polarizabilitiesThree principal axis responses returned by the inversion process, which relate directly to physical attributes of the object under investigation. Information inferred from the responses—including the object’s size, shape, and wall thickness—forms the basis for classification decisions. (sometimes referred to as the object’s EMI fingerprint) serve as the basis for the classification decision.

Figure 2-7. Principal axis polarizabilities.

The objective of the analysis process is reducing the sensor measurements, which are affected by the object’s burial depth and orientation (on the left), to a single set of time-dependent signatures, which are invariant to the target’s depth of burial and orientation (shown on the right). These reduced responses are referred to as the object’s principal axis polarizabilities; they can reveal information about the object’s size, shape, and wall thickness, and are used as the basis for subsequent classification decisions.

Because munitions are inherently axially symmetric, the polarizabilities of munitions have the distinct characteristic of one large principal axis response and two smaller but equal responses; metallic objects that do not possess axial symmetry have three nonequal response curves. An example of polarizability responses for a munition and a fragment is presented in Figure 2-9.

Figure 2-9. Polarizabilities of munition and fragment.

Polarizabilities for munitions are generally characterized by one large principal axis response and two smaller but equal responses (top pane). Polarizabilities for arbitrarily shaped metal items (bottom pane) do not possess this symmetry, and instead have three nonequal response coefficients.

The magnitude of the principal axis polarizabilities reflects the size of the object. Figure 2-10 shows that, while both of these objects are recognizable as symmetric munitions, the 75 millimeter (mm) projectile is larger than the 37 mm projectile by a factor of 6–8.


Figure 2-10. Polarizabilities as a function of size.

The principal axis polarizabilities scale with the size (volume) of the object.

2.1.3 Classification

The final step in this process involves making a classification decision for each detected metal object. This can be accomplished by comparing each object’s polarizabilities to previously measured munitions’ polarizabilities. Statistical classifier algorithms (using machine learning methods) and library-matching classification applications have progressed through the ESTCP Geophysical Classification Pilot Program (SERDP/ESTCP 2014a), and most practitioners now use a library-matching approach.

The library-matching process proceeds by quantifying the difference between the derived polarizabilities of each detected buried metal object with the polarizabilities of known munition items in a library. The objective is to specify how similar the polarizabilities of the unknown objects are to polarizabilities for known objects. There are multiple technical approaches for quantifying the similarities, but the essence of the classification process is to compare the amplitudes and shapes in a numerical and quantitative manner. In Figure 2-11, the polarizabilities of an unknown object (the colored lines) are compared to signatures of three different specific munitions items: 37 mm, 75 mm, and 105 mm projectiles (shown in gray). It is visually apparent that the signatures of the unknown buried metal object are similar to the 75 mm signatures. In this particular case, the unknown object was indeed later proved to be a 75 mm projectile. A screenshot of the library used for the comparison in Figure 2-11 is presented in Figure 2‑12.

Figure 2-11. Library comparison.

Comparison of an unknown buried metal object, shown in color, with library signatures for 37 mm, 75 mm, and 105 mm projectiles. In this case, it is visually apparent that the unknown object is a 75 mm projectile.

Figure 2-12. Collection of EMI signatures for various types of munitions.

As shown here, libraries often consist of not only the EMI signatures (shown in the middle portion of the figure), but also photographs and additional statistical attributes, such as decay rates, size estimates, and descriptions of the specific munitions items.

Comparisons can be made between the polarizabilities of each unknown object and those of the other unknown objects on the site, just as a numerical comparison can be made between the polarizabilities of the unknown objects and those in the project-specific library. This process, illustrated in Figure 2-13, identifies groups of objects (often termed clusters) with similar polarizabilities. A representative sample of objects in a cluster can be excavated to determine whether those objects are hazardous.

Munitions not in a project-specific library can still be identified as TOIs through either of the following means: (1) a large number of items have similar EMI characteristics; or (2) an item has characteristics that uniquely distinguish it as a TOI. The former, commonly referred to as a cluster, is identified by the analyst as a group of very similar sources for which there is no ground truth; these items cannot be confidently classified as non-TOIs. A small number of the cluster items must be excavated to reveal their identity, and if any are found to be TOIs, their EMI characteristics are added to the project-specific library and included in all future library-matching activities for that site. When the EMI characteristics of an unknown source suggest it is long and cylindrical (referred to as axially symmetric) and thick walled (even if there is only one such anomaly across the entire project site), that item should be identified for excavation because its characteristics are common to munitions. Although these characteristics are not measured directly by the EMI instrument, the intrinsic properties are determined through careful analysis of the EMI signature. As with cluster items, if a source having a munitions-like signature (in the EMI sense) is discovered to be a TOI, that signature is included in all future library-matching activities.

Figure 2-13. Cluster of unknown objects’ polarizabilities.

These objects were detected during the ESTCP demonstration at former Camp Beale. One of the objects was excavated and determined to be an expended fuse.

Because the library is used as the point of comparison for identifying TOIs, the integrity and content of the library is important to the ultimate success of the classification project. In other words, if a particular type of munition is anticipated or known to be present on a site, the library should contain reference signatures for it. Site historical records and prior munitions response information should be reviewed to identify munition types.

A baseline master library of munition signature responses is being compiled by ESTCP. The master library will reside on a DOD-hosted website, and periodic updates to the library will be coordinated by DOD. The government project manager will be responsible for ensuring that the current version of the master library is obtained at the beginning of the project. The site team will construct a site-specific library by adding any unique munitions known to be at the site and, in some cases, removing small munitions that are known to not be present. This site-specific library can be modified during the course of the project as new information about the munitions on the site is obtained.

2.2 Advanced EMI Technologies

The most widely available advanced EMI sensor is the MetalMapper, developed by Geometrics (Geometrics 2015). As shown in Figure 2-14, this system is composed of three orthogonal 1 meter (m) x 1 m transmitters for target illumination, and seven three-axis receivers for recording the EMI response decay. Its sampling is electronically programmable, but it typically measures the decay out to 8 ms after the transmitters are turned off. The sensor is normally deployed in a sled configuration, mounted to a tractor or all-terrain forklift, although other schemes are possible. Centimeter (cm)-level-accuracy global positioning system (GPS) equipment is used for navigation and geolocation, and an inertial measurement unit (IMU) is used to measure platform orientation.


Figure 2-14. MetalMapper.

The drawing on the left shows the three orthogonal transmit coils and the seven three-axis receive cubes. The photo on the right shows the MetalMapper in its standard deployment mode on the back of a tractor.

Survey mode—a data collection scheme in which the user scans the ground with a sensor to accomplish 100% coverage (also referred to as a reconnaissance survey, dynamic survey, or detection survey)

Cued mode—a data collection scheme in which the user positions the sensor at discrete XY locations previously identified by other means (also referred to as static or stationary measurement)

The MetalMapper system is designed to be used in both survey and cued detection modes. In survey modeA data collection scheme in which the user scans the ground with a sensor to accomplish 100% coverage (also referred to as a reconnaissance survey, dynamic survey, or detection survey)., only the z-axis transmitter is used for excitation, which results in a shorter measurement period. This allows higher survey speeds and more area coverage per hour. All three axes of all seven receive cubes are used to monitor the response decay. In cued modeA data collection scheme in which the user positions the sensor at discrete XY locations previously identified by other means (also referred to as static or stationary measurement)., the MetalMapper is positioned over each buried item on its target list and collects the full suite of three-axis transmit, three-axis receive data while stationary.

Another commonly used sensor system is the “TEMTADS 2x2,” which was developed by the Naval Research Laboratory. This sensor is composed of four individual EMI transmitters with three-axis receivers, arranged in a 2 x 2 array, as shown in Figure 2-15. The center-to-center distance between the individual sensors is 40 centimeters (cm), yielding an 80 x 80 cm array. The data acquisition (DAQ) computer is mounted on a backpack worn by one of the operators. A second operator controls the data collection using a tablet computer that wirelessly communicates with the DAQ computer. The second operator also manages field notes and team orienteering functions.


Figure 2-15. TEMTADS 2x2.

The drawing on the left shows the four individual sensors in the array, each comprising a z-axis transmit coil and a three-axis receive cube. The photograph on the right shows the TEMTADS 2x2 in survey mode.

In survey mode, each of the four TEMTADS sensors is energized sequentially, and the decay data from all 12 receive coils are recorded to 2.7 ms with minimal averaging. This allows for reasonable survey speeds. For cued measurements with the array static, the four transmitters are energized sequentially and the response is recorded for 25 ms after transmitter turnoff, resulting in 48 (4 x 4 x 3) transmit/receive pairs.

The man-portable vector (MPV), a handheld EMI sensor, has been designed to extend the classification performance of the latest vehicle-based geophysical platforms at the numerous MRSs where forest vegetation or challenging terrain limit access to these platforms. Because of its small size, the MPV has a lower areal coverage rate than larger units. It is composed of a single circular transmit coil and an array of five three-axis receive cubes, as shown in Figure 2-16. The MPV system is still under development and not yet commercialized.

Figure 2-16. Man-portable vector.

The drawing on the left shows the MPV’s transmit coil and the five three-axis receive cubes. The photograph on the right shows the MPV in survey mode.

The characteristics of these three sensors are summarized in Table 2-2.

Table 2-2. Summary of the characteristics of three advanced EMI sensors




Implementation Issues



  • 1 m cube
  • Three-axis transmitter
  • Seven three-axis receive cubes
  • Sample to 8 ms after transmitter turnoff
  • Near-perfect classification at demonstration sites
  • Good depth performance—large transmit moment
  • Both survey and cued modes
  • Vehicular-borne, so some sites precluded
  • Requires GPS

Commercially available


  • Mounted on a small cart, overall dimension 80 cm2
  • Backpack is 25 pounds (lbs.)
  • Four single-axis transmitters
  • Four three-axis receive cubes
  • Sample to 25 ms after transmitter turnoff
  • Near-perfect classification at demonstration sites
  • Less depth capability—smaller transmit moment; best to 50 cm
  • Both survey and cued modes
  • Cart-based deployment
  • Operation with or without GPS

Soon to be commercially available (five arrays currently available)


  • Hand carried on wand, 12 lbs.
  • 50 cm single-axis transmitter
  • Five three-axis receive cubes
  • Sample to 8 ms after transmitter turnoff
  • Can be manipulated in three dimensions to obtain multiple looks at the target
  • Near-perfect classification at demonstration sites
  • Less depth capability—smaller transmit moment; best to 50 cm
  • Both survey and cued mode
  • Small and maneuverable for applications in wooded areas
  • Survey mode with GPS; cued mode with local beacon positioning

Developmental sensor

2.3 Benefits

Geophysical classification provides several benefits that make it worthy of consideration over current standard technology, as discussed below.

2.3.1 Decision Making

With standard technology, the geophysical sensor placed over a piece of metal produces a reading that simply indicates the presence of a metallic object somewhere beneath that sensor. Many things can affect the magnitude of this reading, such as the size and depth of the metallic object. Because traditional sensors only produce a reading in one plane, the only information known is the size of the reading, which could vary due to a wide variety of factors. For instance, the same reading could be the result of a large item buried deeply, a small item buried just below the ground surface, or even multiple small items buried close to one another; with traditional sensors, there is no way to know the difference. Thus, when single-axis EMI sensor technology is used, a dig/no-dig decision is typically derived from comparing the magnitude of an object’s reading (usually in millivolts) to a threshold millivolt level defined through the Data Quality Objective (DQO) process.

The classification sensors provide substantially more data (both raw and derived) about the buried items, allowing for more accurate, defensible decision making. The sensors used for geophysical classification measure EMI responses at multiple points simultaneously, which provides a more complete sampling of the spatial field in three dimensions. EMI sensors provide a data set of responses that the classifier software can then use to essentially derive (invert) the attributes (for example, size, shape, wall thickness) and location of the item that created it. The software can then compare that set of attributes to a library of known items to find a match. Rather than using a threshold millivolt level to make dig/no-dig decisions, classification data allow for fine-tuning of the criteria to include or exclude items from the dig list. The additional data help in explaining to stakeholders or in legal proceedings why the particular selection criteria were used. They also provide information for project archives to inform future interest in the site. Further, each time an excavated item matches that predicted by the classifier, confidence in the system increases, and, conversely, instances of items not matching can indicate that the system is not working properly and requires reevaluation.

2.3.2 Economics

For many sites, geophysical classification provides a significant economic advantage over the standard detection process. The initial classification survey often costs more than traditional methods due to the required collection of higher quality data to precisely identify the locations of buried items (for later classification-grade data collection), as well as to the added cost of collecting and processing cued data over the reacquired targets. However, in most cases, the cost saving of reducing the number of buried items to be intrusively investigated exceeds the added initial expenses, thus lowering the overall cost of remediation per acre. This cost savings means that individual sites can be remediated less expensively, while maintaining or exceeding existing quality levels, and that a fixed budget can accommodate the surveying of many more sites.

An example of the shifting allocation of costs and potential overall cost reduction is presented in Figure 2-17, which is based on cost data from an ESTCP demonstration project. In current practice, most munitions response project costs are associated with excavating clutter, while excavating munitions (the main goal) accounts for less than 5% of each project. By comparison, with geophysical classification, data collection (detection survey and cued data) accounts for nearly half of the overall munitions response project cost, and only about a third of the project cost is spent on excavating clutter. For the ESTCP demonstration project, the classification process was able to reduce the amount of clutter to be excavated by approximately 80%, thus resulting in a total cost savings of 45%. Although additional spending for QC is required for classification to provide confidence in project results, the cost savings are large.


Figure 2-17. Cost allocation and savings.

2.3.3 Explosives Safety and Evacuation

In accordance with DOD Explosives Safety Board requirements, during intrusive excavation activities at areas known or suspected to contain munitions and explosives of concern (MEC), all nonessential personnel are prohibited from entering the area immediately surrounding the excavation (DOD 2010). The no-entry area surrounding the excavation is known as an exclusion zone (EZ). The size of an EZ is calculated based on the blast overpressure distance, or the fragmentation distance of the largest MEC item expected to be encountered.

Although an EZ surrounding manual operations is smaller in diameter than an EZ surrounding mechanized operations such as heavy equipment excavation, it could still obstruct residences, businesses, or public traffic routes. In such cases, buildings within EZs must be evacuated and, similarly, public traffic routes must be barricaded and drivers asked to stop during intrusive operations. If any nonessential personnel refuses to comply with these requirements, the excavation operation must cease.

The use of geophysical classification can significantly reduce impacts to a community surrounding an excavation site because the need to excavate fewer items should reduce the frequency and duration of EZ enforcement. Additionally, geophysical classification can reduce the time needed to complete the remediation of all identified sites, thereby reducing explosive risks to communities.

2.3.4 Cultural and Environmental Conservation

In cases where MEC contamination is present in culturally or environmentally sensitive areas, excavating fewer holes results in less disturbance because of the reduction in soil and vegetation disturbance and because fewer people are in the area, and for a shorter time. The compressed time to complete field work also makes it easier to work around important seasons for protected animals, such as mating or migratory periods. It can also more easily accommodate property owner conflicts, such as during hunting or farming seasons.

2.4 Limitations

Geophysical classification is not applicable in every situation. Some limitations for this approach involve the technology itself. Others involve site-specific characteristics that impose site access limitations, such as areas of dense vegetation; extremely rough, unstable, or steep terrain; or areas subject to electromagnetic interference. Further information on site-specific conditions that can impact or prevent geophysical data collection and classification is provided in Section 3.2.

2.4.1 Technological Limitations

Multiaxis EMI sensors do not consistently detect deeply buried, smaller munitions, and they do not consistently differentiate munitions in highly cluttered target areas (Section 3.2.4). While larger towed units have a depth range similar to standard EMI sensors, handheld advanced EMI sensors are lighter weight and less powerful; although they can sometimes detect deeper items, they are primarily useful in collecting advanced classification data on objects in the upper 1–2 feet of the subsurface. However, because 80%–90% of clutter is detected in the upper 2 feet, portable units should be sufficient to classify a buried item as a TOI (most likely a munition) or a non-TOI, or to determine that the item cannot be classified and thus must be added to the excavation list. Further, the limited depth of detection of the smaller units has not caused a difference in results with ESTCP demonstrations.

While these instruments are designed to be used outdoors, the lack of extreme ruggedization can limit performance in harsh environments, and they are subject to breakdowns. The frequency of such breakdowns depends on design and site conditions. The limited availability of parts for timely on-site repairs can be an issue. However, design improvements are ongoing, and as more instruments are put into use, the availability and distribution of parts should improve. In addition, advanced EMI sensors are not designed to work in extreme weather conditions and are not currently used on airborne or underwater platforms.

Although recent ESTCP demonstrations have shown success in classifying multiple overlapping objects, high-density overlapping objects can be difficult to differentiate. However, the greater the knowledge and experience with the software that analyzes overlapping signatures, the greater the success in classifying individual targets.

Even when target data are clear, a wide range of unknown items or various conditions of munitions (such as damaged or bent rounds) must still be added to the classification library of munitions. The library of EMI responses from various munitions continues to expand with each survey and has been used to detect a wide range of munitions types, including some in various states of damage. In addition, commonly occurring nonmunition items—such as horseshoes, mufflers, and gas cylinders—that possess consistent polarizabilities can also be readily recognized. ESTCP is replacing the library used by research developers with one that has carefully defined procedures for measuring responses and consistent metadata.

2.4.2 Site Limitations

Commercially available advanced EMI sensors are typically mounted on platforms that can be pushed or pulled across an area. This approach tends to preclude their use under difficult site conditions such as thick vegetation, rockiness, and extreme terrain, and in highly muddy areas or those covered by water (Section 3.2.3). Also, as with all EMI sensors, certain geologic conditions can interfere (for instance, in areas with primarily mafic or ultramafic rocks such as basalt). Sites where electromagnetic interference is an issue (such as sites near electrical substations or transmission equipment) or those adjacent to large aboveground or belowground metallic structures may not be conducive to EMI technologies.

2.4.3 Cost-effectiveness

As mentioned in Section 2.3.2, geophysical classification is cost-effective when the additional costs to perform enhanced geophysical investigation are offset by a reduction in the number of intrusive investigations. At most sites, the cost associated with the number of excavations that can be avoided by employing geophysical classification exceeds the extra cost of performing a better initial survey and cued interrogations.

For cued interrogations, typical production rates vary from 175 to over 300 cued measurements per day. Difficult terrain increases the difficulty of maneuvering the equipment, resulting in lower production rates. Higher production rates may be achieved when the terrain is not difficult and anomalies are of high amplitude and easier to locate.

Geophysically noisy sites, TOIs that are smaller than a large portion of the non-TOIs (clutter items), and high-anomaly densities all make classification of individual objects more difficult; consequently, such conditions increase the number of non-TOIs that must be excavated to ensure that all TOIs are removed.

If the ratio of TOIs to non-TOIs across a site is much higher than typical, the number of excavations avoided by using geophysical classification may not justify the additional cost of employing the process. This situation was observed on an air-to-ground gunnery range at New Boston Air Force Station, New Hampshire, but it is not commonly encountered at most MRSs.

Additionally, if a site is relatively small, the cost of acquiring and mobilizing advanced sensors would likely outweigh any cost savings that could be realized through a reduction in excavations.

Publication Date: August 2015

Permission is granted to refer to or quote from this publication with the customary acknowledgment of the source (see suggested citation and disclaimer).


This web site is owned by ITRC.

1250 H Street, NW • Suite 850 • Washington, DC 20005

(202) 266-4933 • Email: [email protected]

Terms of Service, Privacy Policy, and Usage Policy


ITRC is sponsored by the Environmental Council of the States.