Prediction of annual runoff in ungauged basins - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 71

Figure 5.1. Annual runoff in two US catchments: (top) Williams River in West Virginia (332 km 2 ) and (bottom) Santa Ysabel Creek in
California (290 km 2 ). Photos: C. Clark, M. B Stowe.

(see Chapter 7 ) and the flood frequency curve (see Chapter
9 ). Understanding the drivers and causes of annual runoff
variability can improve our ability to predict runoff vari-
ability at all time scales, including the complete runoff
hydrograph (see Chapter 10 ). These signatures can assist
in the development and parameterisation of rainfall

vegetation for the two catchments. The catchment in West
Virginia ( Figure 5.1 , top row) has much higher mean annual
runoff (close to 1000 mm/yr) with moderate inter-annual
variability (range of about ±300 mm). The catchment in
California ( Figure 5.1 , bottom row) has instead a very low
mean annual runoff (below 50 mm/yr) but high variability
between years (close to zero and/or exceeding three times
the mean). It is interesting to explore why there is much less
runoff but with greater variability in the Californian river
compared to that in West Virginia. Predicting annual runoff
in ungauged basins is the starting point for predicting all
other runoff signatures in this topic. Therefore, insight into
the causal processes leading to the long-term mean and
variability, and how similarity and dissimilarity between
catchments can be defined is essential.
This chapter begins by discussing the process controls
on the nature and extent of annual runoff variability, and
how these are governed by the combined effects of climate,
soils and vegetation (including land cover change). Under-
standing of these catchment physiographic and process
controls is used to formulate a list of similarity indices that

runoff
models (Farmer et al., 2003 ; Bárdossy, 2007 ). Mean
annual runoff is also often used as an index to normalise
other signatures as part of regionalisation studies. For
example, it is common to estimate normalised flow dur-
ation curves for ungauged catchments, where the normal-
isation is with respect to the mean runoff (McMahon and
Adeloye, 2005 ) or the median runoff (Best et al., 2003 ).

-

5.2 Annual runoff: processes and similarity
Figure 5.1 presents examples of mean annual runoff and
annual runoff variability from two catchments of similar
size located in the USA, but in two contrasting climates: the
relatively wet West Virginia and the dry Southern Califor-
nia. The pictures are representative of the landscape and