A method for low-flow estimation at ungauged sites: a case study in Wallonia (Belgium)

Well-integrated water management can notably require
estimating low flows at any point of a river. Depending
on the management practice, it can be needed for various return
periods. This is seldom addressed in the literature. This
paper shows the development of a full analysis chain including
quality analysis of gauging stations, low-flow frequency
analysis, and building of a global model to assess low-flow
indices on the basis of catchment physical parameters.
The most common distributions that fit low-flow data in
Wallonia were two-parameter lognormal and gamma. The
recession coefficient and percolation were the most explanatory
variables, regardless of the return period. The determination
coefficients of the models ranged from 0.51 to 0.67 for
calibration and from 0.61 to 0.80 for validation. The regression
coefficients were found to be linked to the return period.
This was used to design a complete equation that gives the
low-flow index based on physical parameters and the desired
return period (in a 5 to 50 yr range).
The interest of regionalisation and the development of regional
models are also discussed. Four homogeneous regions
are identified, but to date the global model remains more robust
due to the limited number of 20-yr-long gauging stations.
This should be reconsidered in the future when enough
data will be available.