Planificación para Grado de Servicio

The heights of the most important gully heads were measured with tape whenever

possible. Table 8.2 gives the heights of the 10 highest vertical headcuts found in red loess and yellow loess, as well as field estimates of the amount of loose soil material

Table 8.1 Properties of different types of loess. Gansu loess data were derived from Derbyshire et al. (2000), Derbyshire & Meng (2000) and Dijkstra et al. (2000)

Gansu loess This study

Wucheng/ Malan Red Yellow

Parameter Lishi loess loess loess loess

D50 (mu) 10-18 30-40 20 42

Clay (% below 2 mu) 18-25 8-14 10 6 Silt (% 2 - 64 mu) 70-77 63-69 82 73

Sand (% above 64 mu) 5 23 8 21

Cohesion (kN/m²)^a 75-100 50-75 100^b 45^b Bulk density (kg/m³) 1520-1810 1380-1440

Dry bulk density (kg/m³) 1570 1270

Wet bulk density (kg/m³) 2018 1777

a assuming a condition of structural strength, i.e., without any fissures or cracks

b derived through back-analysis by assuming that measured headcut heights are maximum possible headcut heights.

Table 8.2 Measured and estimated (Italic) headcut heights (m) of red loess and yellow loess gullies and estimated volume of loose soil material below the headcut, May 1999. The position of the headcuts is shown in figure 8.1

Red loess gullies Yellow loess gullies

Gully Gully Soil fall Gully Gully Soil fall

No. Height volume No. height volume

(m) (m³) (m) (m³)

1 28.6 1 1 14.3 0

2 28.6 10 2 12.0 0.5

3 27.0 * 3 11.8 0

4 24.9 5 4 10.6 0.5

5 24.1 1 5 10.5 3

6 23.9 1 6 10.3 0.5

7 22.1 0.5 7 10.0 0.5

8 20.0 0 8 10.0 0.5

9 19.3 0.5 9 10.0 0

10 17.3 * 10 9.9 0.5

Mean 23.6 2.4 10.9 0.6

* volume of loose soil material could not be estimated as the gully bottom was not visible

accumulated below the headcut. Figure 8.1 shows the position of the gullies included in table 8.2. Figure 8.2 shows pictures of some of the gullies listed in the table, indicating a clear difference in size and morphology between red loess and yellow loess gullies. Both table 8.2 and figure 8.2 show that red loess gullies are much larger. They are also more

directly linked to the stream network, as they are usually found at lower elevation than the yellow loess gullies. In all, more than 50 gullies with headcut heights of over 7 m were measured. The total number of gullies in the catchment is much higher. Table 8.2 also shows that loose sediment volumes were highly irregular; some gullies had large quantities of loose sediment, but most had little or none. Red loess gullies appeared to have more loose sediment than yellow loess gullies, but this is probably caused by the fact that they are larger and that one gully (No. 2) had a very large amount.

Figure 8.1 Lithological map of the Danangou catchment. The position of the gully headcuts from table 8.2 is also shown in the map; red loess gully headcuts are shown with a 4-point star, yellow

loess gullies with a 5-point star

1 2

3

45

6

7 8

9

10

1 2

43 5

6 7

8

9 10

Figure 8.2 Left: red Loess gully complex containing red loess gully heads 2, 4 and 5 of table 8.2.

The position of gully heads 2 and 4 is indicated with arrows; gully head 5 is not included in the picture. Note the person in lower middle portion for scale. Picture taken by R.Vergouwe. Right:

yellow loess gully 4 of table 8.2 (indicated with arrow)

In principle, the difference in headcut height between red loess gullies and yellow loess gullies should reflect a difference in cohesion. It must, however, be realised that the yellow loess thickness is generally not much more than 10 m, so that the difference suggested by table 8.2 is in fact greater than would be the case if the yellow loess thickness did not limit the size of the gullies.

As described in chapter 4.5, over 200 erosion pins were installed in 12 gullies distributed evenly over the Danangou catchment. Pins were classified according to their position on headcut tops, headcut walls, sidewall tops, sidewall walls and gully bottoms. Pin lengths were measured 5 times between October 1998 and September 2000. Pins that had been disturbed (e.g. by children or goats) were not used in the analysis. The measuring results over the 2-year period are shown in table 8.3. The data suggest that there is erosion on gully walls, no change around the gully edge and deposition on gully bottoms. The large standard deviations are caused by individual pins being affected by soil fall. There are some uncertainties with regard to the measurements (e.g. the degree of disturbance), but the data nevertheless suggest that average gully headcut retreat rates are small and that soil fall on gully headcuts is an important process. It should be noted that the

measurement period had below average rainfall.

2

4

4

Table 8.3 Pin length change over a 2-year period (October 1998 – September 2000). Negative sign indicates a decrease in pin length

Pin Position Number of pins Average change Standard

Pin length deviation

(cm) (cm)

Gully bottom 28 -0.95 2.08

Headcut top 16 0.16 2.34

Headcut wall 37 0.89 2.62

Sidewall top 40 -0.00 1.08

Sidewall wall 80 0.41 1.45

Repeated observations (before and after runoff events) at gullies without erosion pins also showed that the gully heads in the Danangou catchment do not change shape perceptibly during runoff events. Nevertheless, gullies do produce sediment during runoff events.

Since many gullies are very large (table 8.2), even a small retreat rate can still produce large amounts of sediment. Some flow erosion and wall collapse may occur during

events, but much material coming from the gully during events was probably produced by soil falls in between and just after events. Field observations showed that in many gullies, soil falls occur from time to time in between events (chapter 3). The loess shows many almost vertical cracks, which may have formed by tectonic forces, stress release or desiccation. Because of the presence of these cracks, soil falls are mostly of a slab-like or column-like form. Most soil falls are fairly small, with heights and widths of a few metres and a thickness of about 0.2 – 0.4 m.

Vandekerkhove et al. (2001) measured headcut retreat rates of gullies in southern Spain and also found that soil fall as a result of tension crack development was one of the major causes of gully headcut retreat. Their average headcut retreat rates were much larger (10 cm per year), but their results were also dominated by soil falls in certain individual gullies. Another study in semi-arid southern Spain, Collison (2001), likewise identified tension crack development as a major cause of gully head instability. Oostwoud Wijdenes

& Bryan (2001) studied gully headcut retreat in silt-loams near Lake Baringo, Kenya.

They found that tension and desiccation cracks developed in between storms and that the depth of these cracks was a function of headcut height, soil properties and the length of the dry period since the last storm.

The material produced by these soil falls accumulates on the bottoms of the gullies, ready to be transported during the next runoff event. Figure 8.3 shows one of the yellow loess gullies in the catchment before and after a runoff event that occurred on July 20^th, 1999.

As can be seen from the picture on the left, there was a great deal of loose soil material available on the gully bottom before the event. Most of it was produced by a small soil fall on the left bank of the gully (to the right in the pictures). The picture (Figure 8.3) reveals that all this loose soil material was removed during the runoff event. Some erosion of the right bank (left in the pictures) probably occurred during the runoff event.

The runoff discharge passing through this gully has been estimated at several hundred litres per second, based on flow width and depth estimated after the event. The estimated flow width was about 1.3 m, while the estimated depth was 0.3 m. Surface velocity of the runoff was measured in the main stream during the event and was found to be about 2 m/s. A flow velocity of, e.g., 0.75 m/s for this much smaller stream is therefore reasonable. This gave a discharge of close to 300 l/s. The yellow loess gully shown in figure 8.3 was not a typical gully for this area (its position and activity were influenced by a large mass movement), but it clearly showed the process of soil accumulation between events and removal during events.

Figure 8.3 Loose soil material in one of the gullies of the Danangou catchment before (left) and after (right) the runoff event of July 20^th, 1999

As noted in chapter 2 the LISEM model is a storm-based erosion model. Since gully headcut retreat as a result of storms is negligible in the Danangou catchment, it can be ignored in storm-based modelling. Gullies can, however, produce major sediment volumes during runoff events because of the removal of loose soil material accumulated on the gully bottom due to soil fall. The process of soil fall is, however, one that cannot be modelled on a storm basis. Instead, it should be modelled on a daily basis. As the number of gullies is large, it is not easy to simulate soil falls using process based stability models. Such models would require detailed information about soil cohesion and shear strength, as well as about soil moisture content. In addition, the location of cracks should be known. All this information would be required for all gullies in order to perform a thorough analysis. The only possibility is to use a pragmatic approach by assuming certain values for the soil physical parameters. Such a daily-based model should produce a map showing the locations and amounts of loose soil material available in the gullies at the time of a particular event. The LISEM model could then be adapted to incorporate

such a map in the calculations for the event. Since the accumulated soil fall material is loose, it will show very little cohesion, and the transport capacity will be the only factor controlling sediment removal.