Measuring stockpile volumes can seem redundant if the material is already being measured with weighing stations at the point of entry and exit. Although truck scales can be very accurate on a single measurement, as the number of measurements add up, so does the uncertainty they add to the total inventory level.
A good way of understanding why the error level in current inventory can be much higher than that of the measurement device, in this case, the weighing station, is by imagining the material and its error level as separate things. Every time a new load comes in, it carries in with it a certain amount of error or uncertainty of measurement. The amount of error depends on the level of precision of the scale, but for simplicity's sake we will imagine the error is 1%. When a load of inventory gets taken out, the uncertainty level does not leave with it, since there can't be negative uncertainty. If this in and out process carries on long enough, it is easy to imagine that the error accumulation becomes larger and larger as a percentage of the actual inventory present in the plant.
This error accumulation problem can be solved by taking direct rather than derived measurements. What this means is that value of the measurements does not depend on any previous measurements. Such independent measurements are uncorrelated and are therefore immune to error drift.