As described in Flood Estimation Handbook Volume 3 (Reed et al., 1999), urbanisation can have a significant impact on the hydrological regime of a catchment.
In general, urbanisation results in: - Faster runoff due to improved drainage; - Increased runoff as urban surfaces are usually less permeable than rural surfaces; - Reduced sensitivity to antecedent catchment wetness as urban surfaces wet-up quickly.
Urban effects tend to be more pronounced in response to short-duration rainfall events, such as are typical of convective storms. As these storms more typically occur in summer it can results in urban catchments having an altered flood regime, with a greater tendency for all-year or summer flooding and increased frequency of floods.
The WINFAP urban adjustment procedure is used to account for the influence of catchment urbanisation in the flood frequency estimation procedures making maximum use of local data, including at-site data. The urban adjustment procedures in WINFAP are based on those published by Kjeldsen, 2010. The procedures adjust 'rural' estimates of QMED and the L-moments (L-CV and L-SKEW) to reflect the impact of urbanisation in a catchment. This can be applied in reverse to provide 'deurbanised' estimates. The adjustment to QMED and L-moments are incorporated through the new descriptor URBEXT2015 (based on LCM2015), as derived within Vesuviano and Griffin, 2025.
These procedures have been modified for use in WINFAP as summarised in the WINFAP 5.3 urban technical note.
The degree of urbanisation can be edited in 'Edit Urban Parameters'.
WINFAP allows adjustments of the degree of urbanisation within a catchment through the use of readily available datasets such as the 1:50k OS map (URBAN extent), or OS Open Built Up Areas (built up extents proportion) dataset, hereafter referred to as OS Open Built Up Extents proportion. By establishing a relationship between these with the URBEXT2015 index and updating these proportions to reflect changes over time an updated URBEXT index can be calculated.
The linear relationship between URBEXT2015 and URBAN is as follows: $$URBEXT2015 = 0.7806 URBAN $$
The relationship between URBEXT2015 and the OS Open Built Up Extents proportion is as follows:
The URBEXT index can also be updated directly using later versions of the Landcover map (Rowland et al., 2025.
Urban Adjustment to QMED
As described above, the impact of urbanisation can lead to faster response times of the runoff and a higher percentage of runoff. These processes are captured within the urban adjustment of the QMED.
Where no at-site data is available the urban-adjusted estimate (QMEDURBAN), is a product of the 'as rural' estimate (QMEDRURAL) and an estimate of the Urban Adjustment Factor (UAF):
The UAF equation is presented below. The first term represents the increase in QMED due to the impact of urbanisation on runoff response times and the second, based on the PRUAF (percentage runoff urban adjustment factor), the changes to the percentage runoff.
$$UAF = (1 + URBEXT2015)^{1.8838} (PRUAF)^{3.52} $$ The PRUAF captures the differences between percentage runoff within a rural and urban catchment, reflecting the fact that these differences will be larger for permeable (high BFIHOST19scaled) catchments.
For highly urbanised permeable catchments the UAF can be large. Based on analysis of observed datasets (the NRFA Peak Flow Dataset) and model parameterisation this is set to a maximum of 10. A warning will be provided where this maximum has been applied.
Where at-site data is available, the 'observed' QMED is effectively the urban QMED and this is deurbanised to estimate the 'as rural' QMED.
Urban adjustment of L-moments for use in derivation of the growth curve
The processes affecting the response to rainfall will change at a greater rate within rural catchments than those within urban catchments. Within rural catchments, as events become larger, they are often preceded by wetter antecedent conditions which will result in increases in the percentage runoff. In addition, as events becomes larger there can also be a related expansion of the stream network producing quicker response times. In an urban catchment however, the percentage runoff and drainage routes will remain fairly static. Hence, for the most extreme rainfall events (higher return periods) the difference between rural and urban catchments are likely to be smaller than for lower return period events.
As described above, the adjustment to the L-moments is based on the perception that urbanisation has the greatest effect on short return period floods and less impact on very long return period floods. The urban growth curve is therefore always flatter than the as-rural growth curve.
Equations for the L-CVURBAN and L-SKEWURBAN are presented below. Note that L-SKEW is not adjusted:
For each station in the pooling group the L-moments can be deurbanised. This deurbanisation of L-moments allows consistency as the 'rural' growth curve is subsequently urbanised for the Site of Interest.
Selection of 'deurbanisation' of the L-moments in the pooling group and at-site data
Where at-site data is selected for use you can decide whether to deurbanise the L-moments from the at-site data prior to application of the enhanced single site (ESS) methodology to produce the rural growth curve. You can also select to deurbanise the L-moments for the pooling groups stations.
The aim of the pooling group (or ESS) assessment is to produce a best estimate of the 'rural' growth curve. When defining a pooling group you select an URBEXT2015 threshold for stations to be included. In the context of urbanisation in WINFAP this represents the threshold above which you do not have sufficient confidence in the generalised procedures for durbanising L-moments.
Whilst WINFAP aims to provide you with flexibility in the selection of whether to deurbanise the pooling group and/or at-site L-moments, the aforementioned assumptions result in the following 'rules' which govern the defaults and any constraints related to deurbanisation in WINFAP:
-
By default the pooling group and at-site data, where available and 'suitable for pooling', is deurbanised.
-
The decision as to whether to deurbanise at-site data defaults to follow that of the pooling group. e.g. if you select to not deurbanise the pooling group then the at-site data will not be deurbanised.
-
If the URBEXT2015 value for the Site of Interest is less than the URBEXT2015 threshold, then you are not able to treat the at-site L-moments differently from those in the pooling group. This is unlikely to make a significant difference to the results and ensures consistency.
-
If the URBEXT2015 value of the Site of Interest is greater than the URBEXT2015 threshold, then you are able to treat the at-site L-moments differently to the pooling group. The catchment is likely to be heavily urbanised with site specific influences hence this flexibility allows you to apply case specific settings which may not be consistent. Warnings are shown for inconsistencies.
Urbanisation procedures and urban adjustment equations used in the 2008 method can be found in the Legacy Method