Distress Data Consolidation Final Report

The authors were able to conduct a more detailed review of 372 of the manual surveys that appeared to be discrepant. Preliminary indicators (i.e., DIM interpretation, summarization errors, seasonal effects, and effects due to differences in data collection methodology) were established for the reviewers to denote possible explanations for these discrepancies. However, the preliminary reviews only considered the data that were entered in the database.

The immediate concern is how much of these data can be salvaged with some additional review of the actual maps. To explore this possibility, the 372 discrepant surveys (along with their unquestioned counterparts) were reevaluated to identify opportunities for resolving the discrepant data. As the data from these test sections comprise half of the total data set, this effort should be indicative of prospects for success for review of the remainder of the discrepant data.

As stated previously, manual distress surveys of the LTPP test sections were not conducted on a routine basis prior to 1995, as the photographic distress surveys were considered the primary mechanism for collection distress data. In 1995, LTPP data collection personnel were instructed to conduct manual surveys every two years on each test section, and review of prior surveys, as part of the audit process for data collection, was suggested. Review of prior surveys as part of the survey preparation process has been required since 1997. Almost two-thirds of the discrepant data from this reduced data set were conducted prior to 1995, as shown in table 18. The data shown in table 18 suggest that, with time, the number of discrepant surveys should continue to decline.

Table 18. Summary of discrepant surveys by year.

Percentages are shown in parentheses for each type of pavement.

CLASSIFICATION OF DISCREPANCIES

In reviewing the maps, explanations were categorized, as shown in table 19.

Table 19. Discrepancy categories.

Processing Errors

Processing errors are probably the easiest to identify and resolve. There were several instances in which a typographical error was made when the data were entered into the database. These are not extremely common, but they do occur. There also were instances in which the maps were incorrectly summarized (i.e., there were errors in determining the total quantity of distress). Some math errors were identified from 1994 when the surveys were converted from English units to metric units. There were some instances for which it was later determined that the data were acceptable, contrary to the original reviewer's concerns. The processing errors are summarized in table 20.

Table 20. Processing errors.

Distress Identification Concerns

Distress identification is still the most likely explanation for discrepant surveys. However, some of these can be rectified (e.g., patches noted with less than one-tenth of a square meter in area). Some of the concerns, such as fatigue versus longitudinal cracking in the wheel path, would require changing old surveys to reflect newer interpretations of the DIM or overriding observations made by previous surveyors. Surveys were not changed unless an obvious error in interpretation could be documented (e.g., patch sizes). The distress identification concerns are summarized in table 21.

Table 21. Distress identification concerns.

Other Concerns

The "other" discrepancy category consisted of the largest number of discrepancies (46 percent), though some do not require resolution. Those distresses with minimal quantities and nonlinear growth rates were considered acceptable, as discussed in chapter 3. Note that the automated review was designed to recognize these as acceptable surveys, as described in chapter 4.

Some sections were identified for which a maintenance or rehabilitation activity was noted by the original reviewer but not recorded in the database, or an activity was recorded in the database but overlooked by the original reviewer. The automated review process was written to examine the data by construction event; therefore, sections in which a maintenance or rehabilitation event has occurred and was appropriately recorded in the data would not be labeled as discrepant by the automated review process.

From this more in-depth review, several PADIAS surveys were identified as causes of the discrepant label. Because PADIAS maps were not available for this review, manual surveys that indicated discrepant PADIAS surveys could not be pursued.

The remaining miscellaneous concerns are summarized in table 22.

Table 22. Miscellaneous concerns

SUMMARY

In summary, the critical issue is how much of the data can be included in the consolidated database if some additional review is conducted using the manual distress survey maps. Table 23 lists the discrepancies by the categories of those that can be easily addressed and those that require additional review. Note that 372 questionable surveys were identified, some of which have more than one issue requiring review.

Because this process took roughly 40 hours to complete (equating to roughly 6 minutes per discrepant survey), it would appear that the opportunity to rectify 50 percent of these discrepant surveys would warrant a comparable review in each region. The review and resolution of these discrepancies could add about 15 percent more data to the consolidated data sets.

Table 23. Resolvable discrepancies from manual surveys.

*Percentages are in parentheses.