Standard Definitions for Common Types of Pavement Cracking (2020)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Standard Definitions for Common Types of Pavement Cracking 47 C H A P T E R 6 Testing of the Proposed Cracking Definitions This chapter describes the effort and results involved in testing the applicability and effectiveness of the proposed cracking definitions for collecting, processing, and reporting cracking data in a systematic and rational approach. In the test, multiple data sets were collected on highways in Oklahoma to determine the feasibility of applying the proposed three levels of cracking definitions. The goals of this work were three-fold. The first was to gain an understanding and proficiency in using already developed software solutions for cracking detection based on the proposed definitions. The second was to verify that field data can be processed in a timely fashion and yield accurate and repeatable cracking results. The third was to gain additional insight and experience using the proposed definitions and to make any necessary recommendations for possible modifications and future research. The presented tests were overall exploratory, and the result may guide future determinations of applicable precision and bias levels for the proposed cracking definitions. Field Data Collection Data collection was conducted using the research team’s PaveVision3D laser imaging vehicular platform (Wang et al. 2015). To identify the desired pavement sections with different cracking types and severity levels, the PaveVision3D system collected over 100 lane miles of pavement image data around Stillwater, OK. Since pavement surfaces in Oklahoma are predominantly asphalt, it is relatively straightforward to select asphalt pavement sites with low, medium, and high severity levels of cracking. However, there are much less concrete pavement sections and most of them have few cracks. In addition, CRCP was not found during the field data collection. As a result, 12 sites were selected for the field testing, six asphalt and six JPCP pavement sections. Each test site was 0.2 miles (0.32 km) long with five repeating runs that were conducted in August 2019 to investigate the applicability and repeatability of the proposed cracking definitions. Site locations are listed in Table 11. Figure 18 illustrates the software interface for automatic pavement cracking data processing. Based on Deep-Learning (DL) algorithms, the ADA3D software is able to conduct automated cracking, rutting, roughness, and texture analyses (Zhang et al. 2017), with the following two capabilities to facilitate the implementation of the proposed cracking definitions:

Standard Definitions for Common Types of Pavement Cracking 48 Table 11 Summary of Field Testing Sites Site ID Pavement Type Beginning GPS Coordinate Visual Inspection Longitude Latitude 1 Asphalt -97.08737 36.1183 With many wide cracks. 2 -97.08447 36.126 With many wide cracks. 3 -97.11255 36.1158 With some cracks. 4 -97.10567 36.1161 With some cracks. 5 -97.13946 36.1159 With few small cracks. 6 -97.13002 36.1161 With few small cracks. 7 JPCP -97.42505 35.6147 With some cracks. 8 -97.42509 35.6102 With some cracks. 9 -97.42496 35.6336 With some cracks. 10 -97.42502 35.6207 With some cracks. 11 -97.34529 36.0846 With few small cracks. 12 -97.3453 36.0718 With few small cracks.  Automatic lane marking detection. Cracks on collected pavement images are detected only within areas between the identified lane markings.  Automated wheel path and grids generation. The wheel path and grids are necessary for the computation of the cracking indexes proposed in this project at various cracking extent and severity levels. Figure 18 Interface of ADA3D (courtesy of WayLink Systems Corporation). Automatic Lane Marking Detection A software module was integrated into the ADA3D software to automatically detect lane markings using the collected 2D pavement images (Zhang et al. 2018). As wheel paths are

Standard Definitions for Common Types of Pavement Cracking 49 defined based on the positions of lane markings, this feature is critical for Level 1 and Level 2 analysis. Examples of the software interface for lane marking detection is shown in Figure 19. To evaluate the efficiency of the proposed lane marking detection algorithms, six testing pavement sections with an approximate total length of 38,000 feet were surveyed and selected for the case study. The F-measure, a commonly used measure in machine learning, was calculated to evaluate the accuracy of the proposed algorithm in lane marking detection. It was demonstrated that the F-measure values were 96%, 94.2%, and 82% respectively for pavement lane markings in excellent, fair, and poor conditions. Figure 20 shows typical detection results of the lane markings. Image area within the two-lane markings is further used for cracking analysis. Figure 19 ADA3D for lane marking detection (courtesy of WayLink Systems Corporation). Figure 20 Illustration of lane marking detection (Zhang et al. 2018).

Standard Definitions for Common Types of Pavement Cracking 50 Wheel Path and Grid Generation Once the lane markings are determined, wheel path and grids are generated to summarize cracking data according to the proposed definitions. Figure 21(a) is the original image with cracking results, while Figure 21(b) includes the wheel path locations and generated grids. The lane markings are highlighted with blue lines. Grids with a “red dot” represents high severity crack width, a “yellow dot” represents medium severity crack width, and a “green dot” represents low severity crack width. (a) Original image. (b) Image with wheel paths and grids. Figure 21 An example of images with wheel paths and grids. In Figure 21(b), only “yellow dot” and “green dot” are found on the image, which means the severity of cracking only includes low and medium levels. It is acknowledged that some roadways may not have lane markings. In that case, user-defined “lane width” would be used to

Standard Definitions for Common Types of Pavement Cracking 51 determine the image area for cracking detection. The shown software capabilities in Figures 19 to 21 are used herein for illustration purposes only. Such features are generally available today from multiple suppliers with variations in implementation. Pavement Cracking Data Analysis with Proposed Definitions Level 3 Cracking Results The cracking percent or Level 3 Index, which represents the overall cracking extent, is the primary cracking criterion of a pavement section. The number of grids with cracking divided by the total number of grids defines the cracking percent. Table 12 summarizes the descriptive statistics of Level 3 cracking data for each site, including the mean cracking percent, standard deviation (SD), and coefficient of variation (CV). The SD of the five replicate tests for each site is less than 2.0%, indicating excellent repeatability of the cracking results using the proposed definitions. Table 12 Statistical Summary of Level 3 Cracking Data Site ID Level 3 Cracking Data (%) Mean SD CV 1 34.9 0.4 1.1 2 41.7 0.8 1.8 3 56.5 1.7 3.0 4 26.7 1.7 6.2 5 3.6 0.1 1.5 6 4.5 0.3 7.4 7 2.0 0.3 17.0 8 1.6 0.2 13.0 9 1.3 0.0 3.4 10 8.6 0.3 3.7 11 4.2 0.1 2.4 12 2.9 0.3 10.2 Figures 22 illustrates the Level 3 cracking data for the selected asphalt and JPCP sites. The cracking percent of the same testing site shows good repeatability among the five testing runs. Therefore, with the automatic cracking detection, lane marking detection, and the generated wheel paths and grids, the proposed definitions can produce consistent Level 3 cracking data results for the testing sites over multiple image data collection events. This result demonstrates the applicability of the proposed cracking definitions in a fully automated environment. However, the traditionally “Ground-Truth” results were not used as the comparing reference in the study due to being out of scope in this limited study, which should be included in future implementation studies in terms of defining what “Ground-Truth” should be and how to obtain it.

Standard Definitions for Common Types of Pavement Cracking 52 (a) Six asphalt sites. (b) Six JPCP sites. Figure 22 Summary of Level 3 cracking data. Level 2 Cracking Results Per the proposed Level 2 cracking definitions, the cracking severities are evaluated according to the average cracking width within each grid. The Level 2 cracking percentages for low, medium, and high severity were computed and summarized. Figures 23 and 24 illustrate the Level 2 cracking data results for the five repeat runs on asphalt and concrete pavement sections. Table 13 provides the statistics of Level 2 cracking results for each site. Level 2 cracking data sheds light on the distribution of cracks over the low, medium, and high severity levels. For instance, the average cracking percent (Level 3) for site 5 is 3.6%, among which 56.1% is in low severity, 14.9% in medium severity, and 29.0% in high severity (Level 2). In other words, site 5 has a low extent of cracking (3.6%) and 56.1% of them are at low severity. 0 10 20 30 40 50 60 70 Site 1 Site 2 Site 3 Site 4 Site 5 Site 6 Cr ac ki ng P er ce nt (% ) Testing 1 Testing 2 Testing 3 Testing 4 Testing 5 0 10 20 30 40 50 60 70 Site 7 Site 8 Site 9 Site 10 Site 11 Site 12 Cr ac ki ng P er ce nt (% ) Testing 1 Testing 2 Testing 3 Testing 4 Testing 5

Standard Definitions for Common Types of Pavement Cracking 53 Table 13 Statistical Summary of Level 2 Cracking Data Site ID Level 2 Cracking Data (%) Low Severity Medium Severity High Severity Mean SD CV Mean SD CV Mean SD CV 1 6.4 0.2 3.9 11.3 0.4 3.6 82.3 0.5 0.6 2 5.6 0.3 6.1 9.1 0.4 4.9 85.3 0.6 0.7 3 61.0 1.4 2.3 36.8 1.3 3.5 2.2 0.2 9.9 4 57.1 3.8 6.7 39.1 3.1 7.9 3.8 0.9 23.9 5 56.1 7.2 12.8 14.9 8.6 57.5 29.0 1.9 6.5 6 73.6 3.8 5.1 24.7 4.1 16.5 1.7 0.3 20.2 7 69.7 3.1 4.4 26.7 2.8 10.4 3.6 0.8 22.8 8 65.3 3.8 5.9 29.1 3.0 10.4 5.6 1.5 26.4 9 89.8 2.4 2.7 9.1 2.2 23.8 1.1 0.7 56.7 10 61.0 1.4 2.4 33.4 1.2 3.4 5.6 0.5 8.4 11 72.9 2.0 2.7 24.2 2.3 9.6 2.9 1.1 39.5 12 82.5 6.5 7.8 16.2 6.1 37.7 1.3 0.7 51.8 For asphalt pavements, sites 3 and 4 have largely low and medium severity level cracks, while sites 1 and 2 have predominantly high severity cracks. For JPCP sites, no site has cracking percent at high severity larger than 10%. Sites 9 and 10 show relatively higher cracking percent numbers at low and medium severity. The low SD values indicate the Level 2 cracking data automatically obtained per the proposed definitions are consistent and repeatable among different data collection events. Comparisons with “Ground-Truth” results are necessary and proposed to be conducted in future implementation studies.

Standard Definitions for Common Types of Pavement Cracking 54 (a) Low severity cracking percent. (b) Medium severity cracking percent. (c) High severity cracking percent. Figure 23 Level 2 cracking data for asphalt sites. 0 20 40 60 80 100 Site 1 Site 2 Site 3 Site 4 Site 5 Site 6 Cr ac ki ng P er ce nt (% ) Testing 1 Testing 2 Testing 3 Testing 4 Testing 5 0 20 40 60 80 100 Site 1 Site 2 Site 3 Site 4 Site 5 Site 6 Cr ac ki ng P er ce nt (% ) Testing 1 Testing 2 Testing 3 Testing 4 Testing 5 0 20 40 60 80 100 Site 1 Site 2 Site 3 Site 4 Site 5 Site 6 Cr ac ki ng P er ce nt (% ) Testing 1 Testing 2 Testing 3 Testing 4 Testing 5

Standard Definitions for Common Types of Pavement Cracking 55 (a) Low severity cracking percent. (b) Medium severity cracking percent. (c) High severity cracking percent. Figure 24 Level 2 cracking data for JPCP sites. 0 20 40 60 80 100 Site 7 Site 8 Site 9 Site 10 Site 11 Site 12 Cr ac ki ng P er ce nt (% ) Testing 1 Testing 2 Testing 3 Testing 4 Testing 5 0 20 40 60 80 100 Site 7 Site 8 Site 9 Site 10 Site 11 Site 12 Cr ac ki ng P er ce nt (% ) Testing 1 Testing 2 Testing 3 Testing 4 Testing 5 0 20 40 60 80 100 Site 7 Site 8 Site 9 Site 10 Site 11 Site 12 Cr ac ki ng P er ce nt (% ) Testing 1 Testing 2 Testing 3 Testing 4 Testing 5

Standard Definitions for Common Types of Pavement Cracking 56 In addition, cracking information in each zone can provide pavement engineers with a better understanding of the pavement conditions relevant to traffic loading. For instance, if the wheel paths (Zones 2 and 4) exhibit higher cracking percent, it indicates the pavement structure has experienced load related deterioration. On the other hand, if the non-wheel path zones (Zones 1, 3, and 5) have higher cracking percent, it may imply that those cracks are mainly non-load related. Level 2 cracking data within each Zone can provide guidelines in determining proper pavement preventive maintenance, rehabilitation, and other pavement management activities. In this Level, besides cracking percent results for each site, cracking data for each zone were also calculated for asphalt pavement sections. For example, for the 1st data collection on site 1, the overall cracking percent (Level 3) numbers are 11.9%, 33.4%, 30.1%, 48.6%, and 20.6% for Zones 1 to 5, respectively (Table 14). Particularly in Zone 1, the cracking percent at low, medium, and high severity levels (Level 2) are 9.0%, 18.1%, and 72.9% as shown in Table 15. This site has higher cracking percent (Level 3) in Zones 2, 3, and 4 (Figure 25(a)), while high severity level cracks (Level 2) are dominant for all the five Zones (Figure 25(b)). Table 14 Summary of Level 3 Cracking Data for Site 1 Testing ID Cracking Percent in Each Zone (%) Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 1 11.9 33.4 30.1 48.6 20.6 2 11.9 33.5 31.0 49.1 21.9 3 8.9 33.8 30.9 48.8 17.4 4 9.6 34.4 31.2 49.7 22.5 5 10.0 33.4 30.6 49.4 19.8 Mean 10.5 33.7 30.8 49.1 20.4 SD 1.4 0.4 0.4 0.4 2.0 CV 13.1 1.2 1.4 0.9 9.7 Table 15 Summary of Level 2 Cracking Data for Site 1 Testing ID Cracking Percent in Each Zone (%) Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 L M H L M H L M H L M H L M H 1 9.0 18.1 72.9 6.0 10.5 83.6 7.0 11.0 82.0 5.4 10.9 83.6 6.0 14.8 79.2 2 8.3 24.3 67.4 7.8 10.2 82.0 5.4 14.4 80.2 5.8 9.1 85.1 7.2 14.7 78.1 3 8.3 29.6 62.0 6.2 11.3 82.5 7.0 11.1 81.9 6.3 10.1 83.5 10.4 13.3 76.3 4 7.7 20.7 71.5 6.5 10.8 82.7 7.1 13.4 79.5 6.2 11.0 82.9 8.5 18.0 73.6 5 18.2 20.6 61.2 6.6 11.6 81.8 6.6 11.1 82.3 5.4 9.7 84.9 8.3 11.3 80.4 Mean 10.3 22.7 67.0 6.6 10.9 82.5 6.6 12.2 81.2 5.8 10.2 84.0 8.1 14.4 77.5 SD 4.4 4.5 5.3 0.7 0.6 0.7 0.7 1.6 1.2 0.4 0.8 1.0 1.6 2.5 2.7 CV 43.1 19.7 8.0 10.8 5.3 0.8 10.7 13.0 1.5 7.2 7.9 1.1 20.4 17.2 3.5

Standard Definitions for Common Types of Pavement Cracking 57 (a) Level 2 cracking percentage at each zone (b) Level 2 cracking percent for each severity level at each zone Figure 25 Summary of Level 2 cracking data at site 1. The detailed cracking data analysis results for the other five asphalt sites are included in Appendix C: Summary of Level 3 and Level 2 Cracking Data of Each Zone. For each site, figures and tables are provided to report Level 3 and Level 2 cracking data of each Zone. Level 1 Cracking Results The summary of Level 1 cracking data for the selected testing sections is described below for each pavement type. Asphalt Pavements Longitudinal, transverse, and wheel path cracking were summarized per the proposed definitions. The lengths for the three cracking types and crack percentages in the wheel paths are listed in Table 16. Site 2 has the highest number of longitudinal cracks, site 3 has the highest numbers of transverse and wheel path cracking, while sites 5 and 6 have the lowest number in terms of crack length in these three categories. 0 10 20 30 40 50 60 70 80 Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Cr ac ki ng  P er ce nt  (% ) Testing 1 Testing 2 Testing 3 Testing 4 Testing 5 0 20 40 60 80 100 L M H L M H L M H L M H L M H Cr ac ki ng P er ce nt (% ) Testing 1 Testing 2 Testing 3 Testing 4 Testing 5 Zone 1 Zone 2 Zone 3 Zone 4 Zone 5

Standard Definitions for Common Types of Pavement Cracking 58 Table 16 Statistical Summary of Level 1 Cracking Data for Asphalt Sites Site ID Statistical Result Longitudinal Crack (ft) Transverse Crack (ft) Wheel Path Crack (ft) Wheel Path Crack Percentage (%) 1 Mean 417.8 503.0 3767.1 15.2 SD 19.3 10.0 86.3 0.2 CV 4.6 2.0 2.3 1.5 2 Mean 1393.4 650.8 3114.1 14.6 SD 21.0 21.8 25.3 0.1 CV 1.5 3.4 0.8 0.9 3 Mean 734.2 1084.2 5982.5 17.2 SD 45.9 35.5 122.3 0.5 CV 6.2 3.3 2.0 2.9 4 Mean 523.5 377.0 2614.8 7.4 SD 111.3 11.6 124.7 0.5 CV 21.3 3.1 4.8 6.4 5 Mean 104.2 155.8 306.8 0.5 SD 81.7 2.8 14.4 0.1 CV 78.4 1.8 4.7 10.1 6 Mean 313.2 146.7 378.8 0.8 SD 79.0 8.9 10.6 0.1 CV 25.2 6.1 2.8 7.2 Further, in Level 1, the severity level for each type of cracking was evaluated based on the average crack width. The lengths of cracks at low, medium, and high severity levels are listed in Table 17. The majority of cracks on sites 1 and 2 falls into the high severity level for the three categories, most cracks on sites 3 and 4 belong to low and medium severities, while most cracks on sites 5 and 6 are in low severity. This result follows the trend observed from the Level 2 cracking data, as shown in Figure 23 and Table 13. Per the statistical results shown in Tables 12 to 17, Level 1 crack results from the five repeating tests have higher variations in terms of the CV numbers. Three possible causes are identified: (1) automated detection errors were amplified for Level 1 details due to its higher level of complexity; (2) crack classification was only used for Level 1 analysis, and its accuracy directly impacts data variations; (3) the accuracy of the lane marking detection algorithm could result in slightly different lane widths among the five data collections, and thus the widths of the five zones will differ slightly. Further research is therefore needed to quantify those variations and propose solutions for field implementation of the proposed cracking definitions at Level 1. For the cracking data at Level 1, the wheel path crack percent has the potential to be used for HPMS and MAP-21 reporting, and MEPDG local calibration. The detailed crack length and severity levels for longitudinal crack, transverse crack, and wheel path crack can also be customized by SHAs for PMS decision-making. Due to the fact that SHAs have relied on

Standard Definitions for Common Types of Pavement Cracking 59 traditional manual based cracking surveys for decades, further research is highly recommended to validate these potential applications using the proposed cracking data definitions. Table 17 Statistical Summary of Level 1 Cracking Data with Severity Levels for Asphalt Sites Site ID Statistical Results Longitudinal Crack (ft) Transverse Crack (ft) Wheel Path Crack (ft) Low Medium High Low Medium High Low Medium High 1 Mean 26.8 47.1 343.9 32.3 56.7 414.0 241.8 425.0 3100.6 SD 1.4 2.6 16.4 1.2 2.0 9.5 12.8 21.6 60.9 CV 5.3 5.5 4.8 3.6 3.5 2.3 5.3 5.1 2.0 2 Mean 77.7 127.3 1188.5 36.3 59.4 555.1 173.7 284.4 2656.2 SD 4.1 5.4 24.1 2.0 2.9 20.6 9.8 11.8 39.3 CV 5.2 4.3 2.0 5.5 4.9 3.7 5.6 4.1 1.5 3 Mean 448.4 269.6 16.1 661.8 398.4 24.0 3651.7 2198.4 132.1 SD 35.4 13.6 0.8 29.0 15.3 2.8 136.0 62.3 12.8 CV 7.9 5.1 5.2 4.4 3.8 11.6 3.7 2.8 9.7 4 Mean 300.1 204.4 18.9 215.4 147.5 14.0 1493.6 1023.3 97.6 SD 72.8 41.9 2.7 17.1 11.5 3.3 125.4 91.7 23.4 CV 24.3 20.5 14.3 7.9 7.8 23.2 8.4 9.0 24.0 5 Mean 60.8 13.4 30.1 87.4 23.2 45.3 171.9 45.9 89.1 SD 50.4 8.9 23.5 11.1 13.4 3.6 22.9 26.8 5.9 CV 82.9 66.3 77.9 12.7 57.7 7.9 13.3 58.3 6.6 6 Mean 229.9 78.3 4.8 108.1 36.2 2.3 278.8 94.0 5.8 SD 54.1 27.9 1.4 9.7 5.4 0.6 11.2 17.4 1.0 CV 23.5 35.6 28.9 9.0 14.9 25.1 4.0 18.5 17.3 Concrete Pavements For concrete pavement sections, the length of longitudinal and transverse cracking was calculated per the proposed definitions. In addition, the percent of cracked slabs and percent of slabs with transverse cracks were summarized for each site. The detailed data are listed in Table 18. Site 10 has the highest amount of longitudinal crack and transverse crack, sites 11 and 12 have significant amounts of longitudinal crack, while site 10 has the largest percentage of slabs with cracks. The largest percentage of slabs with transverse cracks was observed at site 2. Furthermore, the severity levels of the longitudinal and transverse crack data were evaluated based on the average crack width (as shown in Table 18). Most of the cracks on these JPCP sites fell into the low and medium severity levels, which agrees with the results in Level 2 (Figure 24 and Table 13). The percentage of slabs with transverse cracks could be used as the cracking percent for HPMS and MAP-21 reporting, and MEPDG local calibration. The detailed crack length and severity levels for longitudinal and transverse cracking can be customized by SHAs for PMS activities.

Standard Definitions for Common Types of Pavement Cracking 60 Similarly, further research is recommended to validate these potential applications of cracking data for JPCP pavements. Summary The data collection and analysis in this chapter address the feasibility of applying the proposed cracking definitions using state-of-the-art 3D laser imaging and Deep-Learning based solutions. Even though the presented data are preliminary in nature due to limitations of scope and resources, it is clear from the tests that consistency and repeatability are obtainable through the applications. It should be noted that “Ground-Truth” cracking information on the testing pavements sections was not obtained to validate the precision and bias levels. This is a critical process to ensure the acceptance of the proposed cracking definitions, and it is highly recommended that an elaborate implementation and validation project be launched to address any possible weaknesses of the proposed definitions, and quantitatively validate the precision and bias levels of the proposed definitions in multiple SHAs in various geographical regions. For both data collection and processing technologies presented in this chapter, multiple suppliers in the market offer somewhat similar capabilities with certain proprietary advantages from vendor to vendor.

Standard Definitions for Common Types of Pavement Cracking 61 Table 18 Statistical Summary of Level 1 Cracking Data for JPCP Sites Site ID Statistical Results Longitudinal Crack (ft) Transverse Crack (ft) Slabs with Crack (%) Slabs with Transverse Crack (%) Longitudinal Crack (ft) Transverse Crack (ft) Low Medium High Low Medium High 7 Mean 149.9 45.8 29.9 7.9 104.4 40.0 5.7 32.0 12.1 1.7 SD 11.6 7.4 2.9 0.8 8.6 4.9 1.6 6.2 1.3 0.4 CV 7.8 16.1 9.5 9.9 8.2 12.3 27.6 19.3 10.7 25.8 8 Mean 133.5 4.3 21.2 2.7 87.4 38.8 7.6 2.8 1.2 0.3 SD 6.4 0.3 1.3 0.0 8.8 3.1 1.9 0.2 0.1 0.1 CV 4.8 6.3 6.0 1.6 10.0 7.9 25.2 13.8 14.5 46.6 9 Mean 27.4 83.6 15.5 11.0 24.7 2.4 0.4 75.1 7.6 1.1 SD 4.6 1.7 1.5 0.7 4.8 0.4 0.2 2.0 1.9 0.6 CV 16.8 2.0 9.4 6.6 19.3 14.7 50.5 9.5 26.8 47.8 10 Mean 391.6 389.3 63.8 51.6 238.9 130.7 22.0 237.4 129.9 21.9 SD 21.2 5.1 1.6 1.8 15.2 8.0 1.5 7.0 4.3 1.9 CV 5.4 1.3 2.5 3.6 6.3 6.1 6.9 9.5 9.6 11.7 11 Mean 396.9 1.6 47.1 1.5 289.4 96.1 11.3 1.1 0.4 0.0 SD 5.5 0.4 1.0 0.0 9.4 9.6 4.2 0.3 0.1 0.0 CV 1.4 23.8 2.2 3.0 3.2 9.9 36.9 23.7 22.4 NA 12 Mean 256.9 8.9 32.2 1.7 212.3 41.4 3.2 7.3 1.4 0.1 SD 8.6 0.8 1.2 0.6 21.4 14.7 1.7 0.8 0.5 0.1 CV 3.3 8.8 3.8 37.3 10.1 35.6 53.7 12.8 36.0 64.4