PUB in practice: case studies - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 277

Table 11.3. Observed and predicted long-term E/P for selected sub-basins. The sub-basin locations are
indicated in Figure 11.3

Predicted, Model 1: temporally
distributed, spatially lumped

Predicted, Model 2: temporally lumped,
spatially distributed, w ¼ 1.2 Observed

A: Ghataprabha 0.65 0.78 0.82

B: Palleru 0.78 0.73 0.80

Discussion
The Budyko analysis suggests that streams in southern
India belong to one of two major hydrogeographic
regimes. Catchments that have some fraction of their drain-
age area in the Western Ghats have significantly lower
evaporation coefficients (E/P) and larger runoff coeffi-
cients compared to catchments draining the central Deccan
Plateau and/or the Eastern Ghats, which is controlling for
climate. The reason for the higher runoff generation in the
Western Ghats is not demonstrated here, since neither
the temporally distributed, spatially lumped model nor
the temporally lumped, spatially distributed model was
able to predict the high runoff coefficients in catchments
draining the Western Ghats. The spatial heterogeneity in
precipitation, in particular the extremely high precipitation
in the Western Ghats, likely contributes to the high runoff
coefficients, although other catchment properties, including
soil type or land use, could contribute towards the regional
differences. While both temporally and spatially lumped
models yielded acceptable predictions of long-term E/P in
areas with relatively homogeneously distributed precipita-
tion on the Deccan Plateau, the lumped models did not yield
accurate predictions in areas with highly heterogeneous pre-
cipitation. In the Western Ghats, this heterogeneity becomes
particularly important, since the strong rain shadow creates a
juxtaposition of high and low rainfall areas, which yields a
low lumped precipitation value but high runoff.
The long-term average evaporation and runoff coeffi-
cients analysed here can serve as the starting point for
understanding the geographic sources of water in southern
India. Regionalisation of the kind presented here is neces-
sary for determining the spatial and temporal distribution
of water resources in ungauged basins, particularly given
the scarcity of gauging stations. Due to inter-state conflicts
over surface water allocation, this basic information on the
spatial distribution of surface water is not available for
many catchments in India. Quantitative estimates of runoff
by climatic and geographic region is an important step
forward in determining where water is generated in south-
ern India and where it is used, and could be a starting point
for encouraging further data-sharing among states. Inter-
annual variability in water resources is as important as
the long-term annual mean, so future efforts should focus
on the roles of climate, soil and land cover impacts on

inter-annual variations in the hydrology of the two differ-
ent runoff regimes identified here.

11.3 PREDICTING MEAN ANNUAL
RUNOFF ACROSS HUANGSHUI
BASIN, CHINA
shaofeng jia

The issue from societal and hydrological perspectives
The Huangshui Basin in China has great importance for
Qinghai province. Half the population of Qinghai live in
this river basin. Although the density of runoff gauges in
the Huangshui Basin (0.8 station/1000 km 2 ) is less than
what is common in more developed eastern China, it is still
the most densely gauged area in Qinghai. In west Qinghai,
the density of gauge stations is only 0.03 station/1000 km 2 .
Providing high-density spatial information of runoff depth
based on readily available information from a DEM is of
high societal value for water resources management in the
Huangshui Basin. Moreover, the methodology developed
in this case study to obtain runoff depth from a DEM,
vegetation index and other easily obtainable information
may be extended to other places in Qinghai province that
are more sparsely gauged.
This study aimed to map the runoff depth across the
mountainous region of the Huangshui catchment, making
use of relationships between runoff depth and a range of
geographic variables, including altitude, distance from
water vapour source, and vegetation index. Using inter-
polation methods (see Section 5.3.3 ) and multiple linear
regression techniques (see Section 5.3.1 ), we created a
model to disaggregate the mean annual runoff measured
at the gauging station to the entire basin area. It is thus a
statistical interpolation and regionalisation method that
makes use of readily available information to estimate
annual runoff in ungauged areas.

Description of the study area
The Huangshui catchment is located in the east of Qinghai
province. The Huangshui River is a first-order tributary of
the Yellow River, flowing from north-west to south-east.