A data acquisition framework for runoff prediction in ungauged basins - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 31

qualitative description for a-priori model selection in the
context of PUB. What data are acquired specifically
depends on the system under study, e.g., whether it is
located in an arid or in a humid region, and on the purpose
of the predictions. For example, if we are interested in
predicting low flow characteristics then taking a few
selected runoff measurements during low flow conditions
might be most helpful, while we might be interested in
indicators of historical flood levels such as flood marks for
inundation mapping. The following four-point discussion
provides an example of hierarchical data acquisition,
moving from globally available to locally specific data.

Dedicated
measurements

Local field
visits

National
hydrological network

Global
data sets

Spatial scale of data

Figure 3.2. Hierarchy of data acquisition: dedicated measurements
provide detailed information at high costs over small spatial scales;
global data sets provide more generalised information at lower costs
to the individual user.

3.2.1 Assessment based on global data sets
A catchment located anywhere in the world will have an
annual and seasonal climatology characterised by a given
precipitation regime and a basic energy balance. This broad-
scale context is often discussed in terms of the annual water
balance and a climatic index. The well-known Budyko
( 1974 ) diagram represents this as the ratio of mean annual
evaporation to mean annual precipitation versus the ratio of
mean annual potential evaporation to mean annual precipi-
tation, thereby relating a metric of the mean annual water
balance to a climatic aridity or dryness index ( Figure 3.3 ).
The location of a given catchment on this general relation-
ship or curve represents the relative degree of water versus
energy limitation and can inform the coarse interpretation of
the controls on catchment runoff. While valuable, this type
of broad-scale assessment does not include internal catch-
ment characteristics that can influence runoff dynamics nor
shorter-term climate and weather forcing that lead to
dynamic hydrology and storm runoff, but rather provides a
starting point for more localised assessment.

A hydrologist has a number of options to approach the
problem if runoff is to be predicted in a particular
ungauged catchment. Typically, the choice of data acqui-
sition depends on time and other resources available
( Figure 3.2 ). Data sets at the global scale provide the
context and bounds on hydrological behaviour and runoff
potential via basic climatology. Numerous global data sets
that are of relevance to predicting runoff in ungauged
basins exist that can be downloaded at no or little cost to
the user. In many instances, this broad-scale information
will not suffice to predict runoff with the required accuracy
or the required spatial and temporal resolution, so the
hydrologist will acquire hydrological data from any hydro-
logical network that is operated by the national or state
authorities. This usually involves more effort of data qual-
ity checking and predictive methods than when only using
the global data sets. If more time and resources are avail-
able the hydrologist will make a field visit to assess the
hydrological landscape based on his/her expert knowledge.
Local characteristics of climate forcing and internal catch-
ment characteristics provide insights into likely catchment
water storage, surface partitioning and internal redistribu-
tion, and release of water to runoff and evaporation.
Finally, if time and financial resources are even larger,
one could clearly collect some short-term measurements
or even install a stream gauge and other hydrological
equipment to get a better understanding of the catchment
response. This means that acquiring information for esti-
mating runoff from ungauged catchments can follow a
hierarchical approach, depending on resource availability.
The prediction methods presented in Chapters 5

3.2.2 Assessment based on national hydrological
network and national surveys
Every country will have some type of national hydro-
logical network, even though the spatial coverage of such
gauging networks might vary widely ( Figure 3.4 ). The
density of the stream gauge network in the regional vicin-
ity around the basin of interest at least partially defines
what approach to PUB can be utilised. The denser the
network, the more likely it is that a statistical approach to
transferring hydrological information will be successful.
The need for a more process-based modelling strategy
increases with the distance between measurement points.
One would generally seek any runoff and meteorological
data available locally or regionally. The available database
is then analysed with respect to annual water balance,
seasonality, storm behaviour, variability etc., which will

10 that
follow make use of the data acquired at a range of scales,
from proxy data at the local scale to global-scale data sets.
The remotely and locally observed catchment characteris-
tics can be combined for catchment regionalisation and/or

-