PUB in practice: case studies - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 279

Table 11.4. Average annual runoff depth of drainage zones in Huangshui catchment

Drainage area (km 2 ) Discharge (10 9 m 3 /yr)

Drainage zone Runoff depth (mm/yr)

Xinachuan 809 1.633 201

Xiamen-Qiaotou 1466 2.696 184

Baliqiao 464 0.952 205

Xiamen 1308 3.617 277

Haiyan 715 0.473 66

Haiyan-Shiyazhuang 1732 1.786 103

Dongjiazhuang 636 0.840 132

Ledu-Minhe 1853 2.133 115

Fujiazhai 1112 2.056 185

Xining-Ledu 2521 2.506 99

Wangjiazhuang 370 0.437 118

Jijiabao 192 0.331 172

Shiyazhuang-Xining 2356 2.147 91

Based on the spline contours, we revised the runoff
depth contour by using DEM data. To revise the run-
off depth contour we used the criterion that a runoff
depth contour line should be nearly parallel

to the

elevation contour. This results in more detail in the

interpolation.
Finally, we used vegetation depth as a proxy for annual
runoff depth. Vegetation is a very important part of the
environment. The vegetation can usually reflect the local
landform, soil, geology, climate and elevation. Generally,
vegetation cover is thick in places where precipitation is
abundant, and is sparse in areas lacking precipitation. So
we can assume that more precipitation is expected in an
area where the vegetation cover is thick. And we can
qualitatively determine that in an area where the vegetation
cover is thick, more precipitation is received and runoff is
also abundant, and vice versa. Based on the vegetation
information on Landsat TM images, we revised the runoff
depth contour further.

50 km

Figure 11.7. Revision based on DEM. White lines: after revision;
white thin lines: before revision.

Results
After the contour lines had been revised using the DEM,
the interpolation showed more detail (see Figure 11.7 ).
Also the residuals became much smaller than those in the
spline interpolation model. The standard deviation of rela-
tive residuals dropped from 13.4% to 12.1%, though the
standard deviation of residuals increased a little (from
17.6 mm/yr to 18.0 mm/yr). The precision of the esti-
mated values of the small runoff depths improved, appar-
ently at the cost of a little less precision of the large runoff
depths.
After the indicator for vegetation depth had been used,
the contours became consistent with the vegetation cover-
age and we obtained revised runoff depth contours, as
shown in Figure 11.8 . The validation results showed that

We used a period of 42 years, between 1958 and 2000, to
calculate the average annual runoff depth of every drainage
zone ( Table 11.4 ). In addition, a digital elevation model
(DEM, 50 m × 50 m) and a Landsat TM image (summer
2000) of Huangshui catchment were used in this study.

Method
First we assumed that the runoff depth of the centre point
of the 13 drainage zones equals the average runoff depth of
the drainage zone and mapped the runoff depth by using a
normal interpolation method (spline). This resulted in the
white lines in Figure 11.7 .