This technique is by far the most commonly used rainfall loss abstraction procedure for models of this nature. It demands an initial loss estimate that purports to simulate initial catchment wetting when no runoff is produced, followed by a constant continuing loss rate expressed in mm/h to account for infiltration once the catchment is saturated.
This procedure is at the very best crude and can in many instances cause greater errors in runoff estimates than all the other modelling considerations combined. Although it is a commonly adopted procedure, considerable caution should be exercised with its application.
The below table gives a guide to the variations in recommended loss rates generally stated in the literature, however, this data should be used with caution.
|Type of Catchment Surface||Initial Loss (mm)||Continuous Loss (mm/hr)|
Roofs of houses, factories and commercial buildings,
road surfaces, etc.
|5.0 - 20*|
10 - 25*
|2. Loam soils||5.0 - 20*||3.0 - 10*|
|3. Clays, dense structured soils||5.0 - 20*||.5 - 3.0*|
4. Clay subject to high shrinkage and in a cracked
state at the start of the rain
|25 - 35*||4.0 - 6.0*|
|5. ARR (1977)||0 - 50*||-|
* Values taken from an unpublished report by Aitken (1974) based on various textbook values.
The problem of loss estimation is complicated by the fact that the design storm approach in urban drainage design infers the use of rainfall bursts rather than complete storm events. Consequently design storm loss rates need to reflect the possibilities of pre-burst catchment wetting.
Depending on historical sequences of storms and the statistical interpretation of catchment parameters, the design storm loss rates could vary greatly from those associated with complete storm analyses.