There are two key definitions associated with location referencing:

• the location is the point on the road
• the address is a string of characters used in a management system which uniquely and unambiguously defines a location

There are two elements to location referencing: the location reference method and the location reference system.

The location reference method is a set of procedures used in the field to ensure that the proper address is used to describe a location and that the proper location can be found using its address. It needs to be well documented and designed to accommodate all situations. There are two common location reference methods:

• Spatial: gives an address consisting of a set of coordinates. This is commonly done using Global Positioning System (GPS) data
• Linear: gives an address consisting of a distance and direction from a known point, for example:

Kilometer point (e.g., 9.29)

Kilometer post (e.g., 9.29 with equations)

Reference point (e.g., xx + 0.29)

Reference post (e.g., xx + 0.29)

It needs to be appreciated that one location can have many addresses. This is illustrated in the figure below which shows that the same location could be described by five different addresses – all of which give the same end result.

A location referencing system is the application of location referencing methods. Often, several different methods are used simultaneously, for example to reference different types of data or by different groups within a single organisation. The location referencing system will:

• resolve different addresses for the same location
• relate one address to another
• relate one location to another

While spatial (i.e. GPS) data have been tried in a number of pilot studies, their actual day-to-day use in PMS is quite limited. It is mainly confined to presenting results in a graphical format, for example on maps.

Almost all road systems are therefore based on a linear reference system. These describe the location in terms of the offset or chainage from a location reference point (LRP).

The location referencing method used with ROMDAS is a linear one, that is, the survey starts at a given point and progresses along the road. Thus, the survey chainage increases when driving away from the start point and decreases when heading back along the same road towards the start point. The address can be expressed either relative to LRPs or as the distance from the start LRP (the user can select).

The best results are obtained by resetting the chainage to 0 at each LRP thereby expressing all data in terms of the offset from the last LRP, for example 32+150 or "Bridge 1310+450". As described below, this is not only useful for the field teams when they go out to repair the roads, but also for reconciling data between surveys.

The reason for regular LRPs is principally because of odometer calibration. No matter how well calibrated a distance/speed sensor is, there will be variations in the lengths recorded between different surveys on the same road. This will create problems in reconciling data.

Surveys are always done between a start and end point. The common practice is often to only record these two chainages, however, this is inadvisable. No matter how well calibrated a distance/speed sensor is, there will be variations in the lengths recorded between different surveys on the same road. This will create problems in reconciling data.

A better approach is to have regular LRPs along the road. These can be existing km stones, culverts, buildings, signs, or any physical feature which will not change between surveys. By selecting LRPs at regular intervals, generally 1 km, one minimises the errors between different surveys, particularly those conducted in successive years. This is done by resetting the chainage at each LRP thereby expressing all data in terms of the offset from the last LRP. To illustrate the importance of using LRPs and resetting the chainages at each LRP, consider the figure below. This consists of a road which has been accurately surveyed and has LRPs at 1000 m and 2000 m.

No matter how well calibrated the odometer is, it will never read exactly the same in two surveys of the same road. This applies not only to the ROMDAS, but to any distance measuring device. Proper calibration limits these effects, but they can never be eliminated. Thus, in each subsequent survey the sections will not be identical, with the chainage errors accumulating as one travels along the road.

In Case 1, the odometer is underestimating the distance. As a consequence, the actual sampling intervals are greater than the target 250 m. At the first LRP there is only a small difference, however, as one continues along the road the errors accumulate so one eventually has the target segments completely out of synchronisation with the actual segments.

However, were one using LRP resets the errors would not accumulate and would only pertain to the last sampling interval. This is illustrated in the figure under "Case 1 - Reset" where the final sampling interval before the LRP reset is shorter than the others. At each LRP the data are resynchronised thereby ensuring that the errors are confined to each section. It is also important to note that the data, for example the roughness, is calculated on the actual distance measured so the value will be representative of that actual segment, even though they are measured on a shorter segment.

Case 2 arises when the odometer is over-estimating the chainage. As in Case 1, the error accumulates so the sections are soon unsynchronised. In this instance, ROMDAS synchronises the sections depending upon the distance after the last sampling interval when the LRP is recorded. Two situations arise:

• the user can specify a distance after the LRP to increment to the next LRP. By default, this is set at 100 per cent of the sampling interval. If the LRP is recorded within this zone, the data will be stored as in "Case 2 - Reset A";
• if the LRP is not recorded within this zone, the data are recorded as "Case 2 - Reset B".

The importance of using LRPs and LRP resets cannot be overemphasized. Highway agencies which do not use these invariably have problems reconciling their data from year to year. Many have to resort to sophisticated (or not so sophisticated!) processing algorithms ("rubber banding"), while others simply give up in frustration. ROMDAS has been designed in such a way that you can easily avoid these problems. Because of that we STRONGLY recommend the use of LRP resets.

Digital photographs should always be taken of LRPs to provide a permanent record. This is particularly true for the start and end LRPs.

ROMDAS is designed to assist with recording digital photos. The user can enter a digital photo number at each LRP and this is stored with the LRP data. ROMDAS will automatically generate the same file name as the camera and this is also stored. If your camera is not included in the ROMDAS list the software will be updated to include it as an option.