Determining The Relative Detectability Of Ground Weapon Systems CSC 1984 SUBJECT AREA Operations DETERMINING THE RELATIVE DETECTABILITY OF GROUND WEAPON SYSTEMS SUBMITTED TO THE MARINE CORPS COMMAND AND STAFF COLLEGE QUANTICO, VIRGINIA IN PARTIAL FULFILLMENT OF REQUIREMENTS FOR WRITTEN COMMUNICATIONS MAJOR EDWIN T. CARLSON UNITED STATES ARMY APRIL 20, 1984 Introduction There is a common maxim of ground combat which states, "If you can see a target, you can hit it." The validity of this assertion is apparent since today's direct fire weapons are characterized by such high velocity trajector- ies that the projectile almost follows a line of sight from the weapon to the, target in a matter of seconds or fractions of seconds. To take full advantage of these weapons improved sighting systems, night optics, and laser range finders have been developed by both western and Warsaw Pact countries. These sighting devices are designed to improve target detection capabilities and to speed target engagements. Such advances in weapons technology have precipi- tated serious concerns about the survivability of ground weapon systems on future battlefields. Lower silhouettes, faster speeds and less exposure times during engagements are becoming primary weapon system design characteristics. An individual gunner or crew depends on his or their weapons to defeat the target, to operate reliably and safely, and to be employed with minimum exposure to enemy observation and fire. This latter attribute of the weapon is often the most difficult characteristic to determine objectively. It is relatively easy to conduct a test to confidently determine a weapon's accuracy and lethality against prospective targets. It is also relatively easy to confidently assess the weapon's reliability under different conditions and rates of fire. Operational weapons testing can provide the results of accuracy and reliability testing but usually provide only subjective esti- mates of a weapen's susceptability to detection by the enemy. The operational user of the weapon to be procured, the Army or Marine Corps, usually ask the correct questions about the survivability of a new ground weapons system. However, operational and developmental testing has usually only been sufficient to provide subjective answers where objective answers would be better. Objective answers provide the best means for fully informed procurement decisions. The focus of this paper is to establish a method of operational testing and analysis of ground anti-tank weapons systems that will provide an objective assessment of firing signature detecta- bility. Theory A ground weapon system is (the) most vulnerable to detection when it fires. The associated smoke, flash, and noise signals all who may be observing the battlefield disclosing the weapon's presence and location. The magnitude of this disclosure (the amount of smoke and flash, and the loudness), the speed with which it can be detected and engaged, and the accuracy of the enemy to pinpoint its location are the keys to objectively determining the detectabil- ity of amy ground weapon system. The parameters of interest to the analyst and ultimately to the decision maker would be the probability of detection, the times to detect and engage, and the accuracy with which the weapon position was pinpointed. The probability of detection of a weapon system as it fires can be esti- mated in operational testing by requiring independent observers to survey a simulated battlefield and record the number of times they detect a firing weapon. As the number of observers are increased and the weapons are fired under varying conditions such as light (day/night) and range (distance from firing location to observers), the frequency of detection provides a good estimate of the probability of detection; p = (# detections) / (# firing) x # observers. (THIS REPRESENTS DIVISON) In probability theory this function is the maximum likelihood estimator for a random variable taken from a Binomial distribution. The Binomial distribution best models the actual process of a string of detection opportunities which is of the form 0 or 1 or Yes/No. That is, only one of two possible outcomes will occur in any trial. This is much like the toss of a coin, heads or tails. Ultimately the actual probability of occurrence associated with such a Binomi- al distribution will be p (heads) = # heads/total # of tosses.1(THIS REPRESENTS DIVISON) It is expected that each observer will take a different length of time to detect the weapon. For each firing of a weapon the time begins when the weapon fires and is precisely stopped when an observer thinks he detects the weapon. The collection of these times for each weapon and for each observer provide a distribution of time increments. If all of these time increments are plotted in a cumulative distribution function the function will resemble that in Figure 1. Click here to view image This cumulative distribution function is of the exponential form F(t) = 1-e-Kt.2 One hundred percent of all firings would rarely all be detec- ted by all observers so the function only approaches the 100% line as a limit as time goes to infinity. The time corresponding to the 50% point on the curve is called the median time and is the time above and below which 50% of all detections occurred. The median is considered a better measure of central tendency of the time distribution as opposed to the mean or average time. Some observers never detect some weapons and their detection times on these opportunities are running into infinite time. Including these trials in the average would skew the mean and omitting them would be less representative of what actually occurred. Therefore, the median detection time will be used as the parameter that best represents the expected time for a detection to occur. The accuracy with which an observer can locate or pinpoint the weapon due to its firing signature can be measured if both the weapon and observer loca- tions are precisely known and the observer can provide an azimuth and range estimation from his position to the target. By plotting weapon and observer locations, observer azimuth, and range estimation on a map the lateral error can be determined for each trial. Figure 2 depicts how this would look. Click here to view image The observer error is measured from the actual firing weapon location to the point determined by the observer's azimuth and range estimation. For better accuracy the positions of the observer and the firing weapon should be surveyed to at least eight-digit grid coordinates, providing an accuracy of plus or minus 10 meters. Ten-digit grid coordinates provide an accuracy of plus or minus one meter. The collection of these lateral observer errors form a distribution from which certain summary statistics such as the mean and standard deviation can be determined. The mean of this distribution is simply the average distance, expressed in meters, between the actual weapon location and the estimated weapon location. The relative magnitude of this average error objectively determines the accuracy with which observers could pinpoint the weapon's location based upon detection of its firing signature. Previous Testing Research reveals that only two opetational tests have been conducted in which the above method or variations of these methods have been used. In 1955 Project Pinpoint was conducted to provide information related to the problem of target acquisition by tanks in an overwatching role and to determine the effects of several factors on target acquisition, such as weapon type, light condition and range.3 In 1981 the Viper Light Antitank Assault weapon was tested during its Operational Test II (OT II) to determine its detectability relative to the M72 LAW, which the Viper was designed to replace.4 The unknown variables in each test consisted of the frequency of detec- tion, the time to detect and engage, and the accuracy with which the observers could pinpoint the firing weapon. The controlled variables common to both tests consisted of the type weapon, the range between observer and firing weapon, and the light condition (day or night). Project Pinpoint also varied the number of times the same weapon fired from the same location to determine the effect of multiple firings on target acquisition and detectability. Due to ammunition constraints the Viper and LAW only fired once from each position per trial. Whereas the Viper OT II tested the comparative detectability of the Viper and the LAW, Project Pinpoint tested the battalion antitank (BAT) rifle, the M48 90mm tank gun, and the 76mm towed antitank gun. Data collection requirements for both tests included the number of detec- tions, the time to detect and engage, the precise location of observers and firing sites, the azimuth determined by the observer from his position to where he thought the firing weapon to be, the observer's range estimate to the target, and the particular component of the firing signature which prompted detection of the firing weapon. Observers in both tests were mounted in tanks. Several reasons contrib- ute to this being an excellent consideration. First, the safety of the observers, even though they were outside the firing fans of the weapons, was enhanced. Second, observation of the battlefield by overwatching tanks was realistic. Third, the tank's fire control system and azimuth indicator mounted on the turret ring provided an excellent means of measuring the azimuth from the tank to the detected target. They also accurately measured the speed of engagement as the crew went through the drill of ranging, loading and simulated firing at the target. Fourth, the range finder of the tank gave the observer the means to determine more accurately the range from his position to where he thought the target was. The simplicity of this method of collecting data on ground weapon system detectability is appealling. There are no unique test site requirements other than those of a range facility appropriate to the size of the weapon being tested, a downrange observer tank area which provides visibility of the firing sites, and the ability to survey weapon and tank locations precisely. The observers need no other qualifications other than those expected in any compe- tent tank crew The primary observer would, naturally, be the tank commander for each tank. The gunners or crews of the ground weapons being tested would require training on that weapon, but no additional training, beyond sound employment techniques, would be needed for them to participate in such a detectability test as discussed here. The data collectors have to be capable of keeping the same time accurately at the firing site and at each observer tank. Questionnaires have to be issued to each tank commander to determine the particular firing signature component that cued his detection of the target. Multiple firing sites must be surveyed to prevent the observers from limiting their visual search area to the particular sites they may begin to learn as the test is conducted. The test must be conducted under operational conditions. The more artificiality induced in the conduct of the test will make the validity of results more questionable. Once the test is completed and the data is validated, correct data analy- sis will allow the operational evaluator to begin to objectively answer cer- tain key questions prompted by the concerns of the operational user. These are: . What is the frequency of detection? . Which signature components are the most significant? . How quickly can the weapon be detected and engaged? . How accurately can the weapon's location be pinpointed? Methods of data analysis to provide answers to these questions are below using examples from the Viper OT II and Project Pinpoint. Frequency of Detection As discussed earlier, the frequency of detection is merely the total number of detections divided by the total number of detection opportunities and can be written as a percentage. Table 1 and Table 2 present the parcent- age of detection for Viper and LAW overall and differentiated by light condi- tion.5 The numbers in parenthesis are the number of detection opportunities that occurred in the test for any set of weapon and light conditions. In the Viper OT II thirty-two rounds of Viper and LAW were fired. Five observer tanks afforded the opportunity to attain a possible 160 detection events for each weapon system. However, a sample population of only 155 detection oppor- tunities for Viper and 158 for LAW were considered for analysis, since seven unresolved data base errors prevented all 320 data elements from being used.6 Click here to view image Overall the Viper was detected 73.5% (114/115 = .735 X 100 = 73.5%) and the LAW was detected 42.4% of the time. The Chi-Square test of a difference of two proportions was used to confirm statistically that the difference between these two results is significant.7 Signature Component In the Viper OT II the observers were questioned concerning the particu- lar component of a launch signature that was responsible or partly responsible for detection of the system. These data are presented in Table 3. As can be seen, the majority of detections of both systems were due to flash, smoke and the combination of flash and smoke.8 The correlating of signature components and frequency of detection as is done in Table 3 provides the analyst with not only the means to determine which signature components were significant, but also the means to statisti- cally determine to what degree each component or combination of components contributed to detection. Click here to view image Time to Detect and Engage Given that the observers in the tanks observed a weapon fire, the time to detect is determined on a time-line that includes exact time of firing and exact time of detection. The time to detect a given firing can then be deter- med as the magnitude, in seconds, of the interval between the two times. As an example, the cumulative distributions for both Viper and LAW detection times are presented in Figure 3.9 For such distributions the median time is preferred as the measure of central tendency, as mentioned earlier. Interest- ingly, in the Viper OT II the Viper and LAW detection time distributions were essentially the same and were found not to be significantly different. The median time to detect both LAW and Viper was found to be the same at 3 seconds.10 Click here to view image The time to engage the detected system was also determined on the same time-line so that the time to detect the system and the time for the tank crew to return fire with the main gun could be determined. Figure 4 presents the time to detect and return fire (counter fire time) on both LAW and Viper systems.11 As seen, the distribution of times for both systems is very simi- lar with a common median time to detect and return fire of 9 seconds. In Viper OT II neither LAW nor Viper were faster to detect and to simulate engage- ment by the observer tanks. Analysis of the Project Pinpoint recognition time data (equivalent to the counterfire time variable of Viper OT II) reveals the same phenomenon. The cumulative frequency distributions of detection times for the BAT, the tank, and the towed gun (see Figure 5) were found not to differ significantly with a median time of 8 seconds. It would be premature at this writing based on the results of Viper OT II and Project Pinpoint to suggest that the time to detect a ground antitank weapon system is independent of the type weappon. However, future evaluators should not be surprised to discover similar results in such tests of newer weapons. Click here to view image Pinpoint Accuracy The expression, lay of the gun, denotes the action of a tank crew bring- ing their main gun to bear on a prospective target, ranging to the target and selecting the appropriate ammunition. Laying the gun essentially equivocates to those tasks necessary for target engagement. The azimuth indicator, mount- ed on the turret ring, can be zeroed at an arbitrary point of reference and all azimuths of gun lay can be measured with respect to this point of refer- ence. This method was used in the Viper OT II and Project Pinpoint. The azimuth from the lay of the gun to the target area was compared to the true azimuth from each tank gun to the actual weapon position, which were accurate- ly surveyed for this very purpose. This angular difference, measured in mils, was converted to meters at the true range of the target system in order to determine the distance right or left of the true location at which a fired tank projectile would pass or impact (1 mil equates to 1 meter at 1,000 meters). Figure 6 graphically depicts the method of determining the ob- server's horizontal error. Click here to view image Figure 6. The technique of determining an observer's horizontal error. The distribution of all horizontal lay errors then can be analyzed to deter- mine certain summary statistics such as mean, median and standard deviation. Figure 7 presents cumulative frequency distribution of the horizontal errors for Viper and LAW.14 The average horizontal error obtained against the Viper was 22.7 meters and against the LAW was 23.9 meters.15 Confidence intervals can be constructed using this data at the level of confidence required to determine if a significant difference in accuracy exists between the two weapons being compared. In Viper OT II the results of this analysis indicate that the Viper was not significantly easier to locate or engage accurately than the LAW, with about 60 percent of the pinpoint positions of both systems less than 25 meters from the true location.16 Project Pinpoint results, shown in Figure 8, were that the average lay error was about 5 yards for each tested system, with 80 Click here to view image percent of all errors within the limits of 25 yard lay error.17 As in the Viper OT II analysis, Project Pinpoint results also saw no affect on pinpoint accuracy due to the weapon type. Another method of evaluating the accuracy with which detected ground weapon systems can be engaged is to determine the radial error involved in each engagement. By plotting the observer range estimation on the observer azimuth a point of impact can be determined. The distance from this point of impact to the true target location is the radial error for that particular trial (see Figure 2.). The distributions of these radial errors for each tested weapon will provide a comparison of the accuracy with which observers could bring indirect fire, mortars and artillery, to bear on the target.18 Click here to view image Discussion of Findings The methods of analysis discussed above and their results must be evalua- ted without forgetting the conditions under which the data were collected. This is why the proper operational setting and weapon employment procedures must be adhered to during the course of the test. If the tests were conducted under realistic operational conditions the evaluator can state his findings with some confidence that the test results approximate those results which may be achieved in combat. Attempting to answer the questions posed earlier, the Viper OT II and Project Pinpoint results parallel each other as follows:19 . The probability that an antitank weapon will be detected was dependent on the type of weapon. Viper was more detectable than LAW. . Flash and smoke, and the combination of flash and smoke, were responsible for the majority of detections. . If an antitank weapon were detected, the time to detect the target was not affected by the type of target weapon, the target angle, or the target range. The time to detect the Viper was not significantly different than the time to detect the LAW. . The horizontal lay error (pinpoint accuracy) with which a tank crew could engage a firing weapon system does not differ significantly between different weapon systems provided a detection is made. Eighty percent of all lay errors were less than 25 yards right or left of the true target location in Project Pinpoint. The accuracy of pinpointing the LAW or Viper system was not significantly different. Sixty percent of all lay errors were found to be less than 25 meters or each system. Payoff Ratio The issue of a weapon system's vulnerability to detection by enemy observers provides some insight into the broader issue of the survivability of the weapon and its' gunner or crew. If a weapon is rarely detected when it fires, it follows that it has a better chance of survival on the battlefield than a weapon that is frequently detected. The point at which the detectabi- lity of a weapon seriously endangers its survivability cannot be objectively determined. It is a subjective determination by the operational evaluator and decision maker. An aid to the evaluation of this question is the construction of a payoff ratio. The payoff ratio takes into-consideration the fact that a gunner may be willing to risk a higher probability of detection in order to achieve a higher probability of hit. This was the case in the Viper OT II. The Viper attained an overall probability of hit, Phv, of 0.51 in the test as compared to the LAW's probability of hit, PhL, of 0.24.20 As mentioned in the previous analysis the Viper's overall probability of detection was Pdv = 0.735 as opposed to the LAW's lower probability of detection PdL = 0.424 (see Table 1). Though a Viper gunner had a better chance of hitting his target, he also had a better chance of being detected once he fires. The overall comparison of Viper and LAW really amounts to an assessment of the trade-offs between each weapon's performance accuracy and survivability. If a ratio of these trade- offs is written, it could take the following form: (Phv)(PSv)/(PhL) (PSL) (THIS REPRESENTS DIVISION) where: PhL = probability of hit by LAW Phv = probability of hit by Viper PSL = probability of survival for LAW gunner Psv - probability of survival for Viper gunner Gunner survival can be essentially assured if the gunner is not detected by any observer upon firing, or, if detected by at least one observer, can reach cover in some time less than the observer can bring counterfire on his firing position. Pd is the probability that a gunner is detected by one observer upon firing, then (1 - Pd) is the probability that he is not detected. For n observers, the probability that he is not detected by any observer is (1 - Pd)n, assuming that all observers have the same probability of detection, Pd. The counterfire time distribution in Figure 4 is of the form F(t) = 1 - e-k(t-to) The probability of the observer firing in some time t, greater than the time the gunner can take cover is 1-F(t) = e-k(t-to) The probability, P, that the gunner will be detected by at least one obser- ver, when n observers are present is [1 - (1 - Pd)n] = P (This is the probability of at least one detection from n observers.) Thus, the payoff ratio of Viper to LAW can be written as Click here to view image where to is the earliest time counterfire can be returned. Since the counter- fire time distributions for Viper and LAW in Figure 4 are essentially the same k/v = k/L = 1/t-t/0 (THIS REPRESENTS DIVISON) The probability of survival for each weapon appears complicated but is merely the probability that the LAW or Viper gunner is never detected or, if detected, can find cover faster than at least one observer can engage him. Using the data attained in the Viper OT II where the average engagement time was 9 seconds. Click here to view image and n is the number of observers. Table 4 presents the values of this payoff ratio for increasing time to take cover and for one, two and three observers.21 It is evident that the Viper payoff decreases as the gunner remains exposed longer after firing and the number of potential observers increase. If the gunner can reach cover immediately (3.5 seconds or less), the payoff ratio merely becomes the ratio of the probilities of hit, since the probabilities of survival are essentially one. (It is assumed for simplicity that once a gunner can reach a covered position, his probability of survival is unity.) Any payoff ratio value greater than one indicates that the Viper offers the greater payoff. Payoff ratio values less then one indicate that the LAW is the most advantageous system. Click here to view image The payoff ratio shows that the Viper can offer a viable advantage over the LAW in spite of its increased detectability. It is also surmised that the Viper gunner Would be ill-advised to fire the Viper more than once from the same position, since the results of Project Pinpoint indicate that the proba- bility of detection increases with successive rounds fired from the same position.22 Conclusions Operational testing of ground weapons systems can objectively address the issue of the weapon's detectability in order to provide better answers to the user's issue of survivability. The methods of testing, data collection, and analysis advocated herein present some guidance as to how this may be accom- lished. The demands upon the operational tester are minimal. There is a need for tank-mounted observers in safe down-range locations, firing sites and tank sites must be accurately surveyed to ten digit coordinates, and accurate time keeping capabilities must be present at all locations. A detectability sub- test could be easily incorporated into the weapon accuracy portion of the operational test, and need not be conducted for its own sake. Of special importance is the need to retain operational realism in the employment of the tested weapons and the observer tanks. The data attained can be analyzed using the very simple statistical techniques presented above. More sophisiticated analysis can be used but are beyond the scope of this paper. Determination of a payoff ratio should aid in providing a good assessment of the trade-offs between accuracy and detectabil- ity. In addition to determining the extent of the payoffs for each system tested, certain insights can be gleaned from this analysis concerning proper weapon employment techniques that can take full advantage of weapon accuracy and still reduce weapon detectability to an acceptable level. REFERENCES 1Morris H. DeGroot, Probability and Statistics (Reading: Addison-Wesley Publishing Company, 1975), p. 285. 2Leonard Kleinrock, Queuing Systems (New York: John Wiley and Sons, Inc., 1975), p. 66. 3John P. Young, Andrew T. Ackles, et. al., Project Pinpoint; Disclosure of Antitank Weapons to Overwatching Tanks (Chevy Chase: John Hopkins Uni- versity, 1958), p. 1. 4U.S. Army Operational Test and Evaluation Agency, Independent Evaluation of the Viper Light Antitank/Assault Weapon, XM 132, Operational Test II (Falls Church, 1981), p.6. 5Ibid., p. 29. 6Ibid., p. 28. 7Ibid. 8Ibid., p. 30. 9Ibid., p. 32. 10Ibid., p. 31. 11Ibid., p. 33. 12Ibid., p. 31. 13John P. Young, Andrew T. Ackles, et. al., Project Pinpoint; Disclosure of Antitank Weapons to Overwatching Tanks (Chevy Chase: John Hopkins Uni- versity, 1958), p. 57. 14U. S. Army Operational Test and Evaluation Agency, Independent Evaluation of the Viper Light Antitank/Assault Weapon, XM 132, Operational Test II (Falls Church, 1981), p. 34. 15Ibid. 16Ibid., p. 31. 17John P. Young, Andrew T. Ackles, et. al., Project Pinpoint; Disclosure of Antitank Weapons to Overwatching Tanks (Chevy Chase: John Hopkins Uni- versity, 1958), p. 51. 18U.S. Army Operational Test and Evaluation Agency, Independent Evaluation of the Viper Light Antitank/Assault Weapon, XM 132, Operational Test II (Falls Church, 1981), p. 35. 19Ibid., pp. 35-36. 20Ibid., p. 39. 21Ibid., p. 40. 22Ibid., p. 39. BIBLIOGRAPHY 1. DeGroot, Morris H., Probability and Satistics. Reading; Addison-Wesley Publishing Company, 1975. 2. Independent Evaluation of the Viper Light Antitank/Assault Weapon, XM 132, Operational Test II. U. S. Army Operational Test and Evalution Agency, 1981, IER-)T-279. 3. Kleinrock, Leonard. Queueing Systems. New York: John Wiley and Sons, Inc., 1975. 4. Viper Operational Test II Test Report. U.S. Army Operational Test and Evaluation Agency and U.S. Army Infantry Board, 1981, FTR-OT-279. 5. Young, John P., and Ackles, Andrew T., et. al., Project Pinpoint; Disclosure of Antitank Weapons to Overwatching Tanks. Chevy Chase: John Hopkins University, 1958.
