Agricultural Water Management 65 (2004) 211–224

Runoff characteristics of artificial catchment materials for rainwater harvesting in the semiarid regions of China Xiao-Yan Li a,∗ , Zhong-Kui Xie a , Xiang-Kui Yan b,c a

Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences, Lanzhou 730000, PR China b Department of Resources & Environment Sciences, Gansu Agricultural University, 730070, Lanzhou, China c National Institute for Agro-Environmental Sciences, 3-1-3 Kannondai, Tsukuba, Ibaraki 305-8604, Japan Accepted 29 September 2003

Abstract Rainwater harvesting is being promoted to solve water problems for agricultural and domestic uses in the semiarid loess regions of China. In recent years, however, the current rainwater harvesting practices are still confined to rural family units to supply household water needs and for very limited supplemental irrigation purposes due to low runoff efficiencies of the limited catchment types and consequently low amount of collected water. The current runoff catchments mainly include rooftops, courtyards, earth and asphalt-paved roads. It is necessary to test artificial catchment treatments to select optimum treatments for large-scale use in the region. This study evaluated runoff characteristics of six surface treatments relative to rainfall amount and intensity and antecedent rainfall during naturally occurring rainfall events in the semi-arid loess regions of northwest China. The surface treatments included two basic types, i.e., earthen (natural loess slope and cleared loess slope) and barrier-type surface treatments (concrete, asphalt--fiberglass, plastic film, and gravel covered plastic film). The results of the study indicated that runoff and runoff efficiency of the earthen surface treatments were closely related to the rain intensity, while runoff from the asphalt fiberglass, plastic film, gravel-covered plastic film, and concrete surface treatments was more governed by the amount of the rainfall. Asphalt fiberglass had the highest average annual runoff efficiency of 74–81%, followed in decreasing order by the plastic film (57–76%), gravel-covered plastic film (56–77%), concrete (46–69%), cleared loess slope (12–13%), and natural loess slope (9–11%). Antecedent rainfall had an obvious effect on the runoff yield for the cleared loess slope, natural loess slope, and concrete.

∗ Corresponding author. Present address: Estacion Experimental de Zonas Aridas, Consejo Superior de Investigaciones Cientificas (CSIC), General Segura 1, 04001, Almeria, Spain. Tel.: +34-950281045; fax: +34-950277100. E-mail addresses: [email protected], [email protected] (X.-Y. Li).

0378-3774/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.agwat.2003.09.003

212

X.-Y. Li et al. / Agricultural Water Management 65 (2004) 211–224

The threshold rainfall was 8.5, 8.0, and 1.5 mm for the natural loess slope, cleared loess slope, and concrete treatments, respectively, without antecedent rainfall effects and 6.0, 5.0, and 1.2 mm, respectively, with antecedent rainfall effects. Due to the impermeable surface, antecedent rainfall had little effect on the runoff yield for the asphalt fiberglass, plastic film, and gravel-covered plastic film treatments, which had a threshold rainfall of 0.1, 0.2, and 0.9 mm respectively. © 2003 Elsevier B.V. All rights reserved. Keywords: Rainwater harvesting; Surface treatments; Runoff; Threshold rainfall; Semiarid; China

1. Introduction Water harvesting is a method of collecting surface runoff from a catchment area and storing it in surface reservoirs, or in the root zone of a cultivated area. It can be a source of water for a variety of purposes in arid and semiarid regions when common sources, such as streams, springs, or wells, fail (Frasier, 1980). In addition to supplying drinking water for people, livestock, and wildlife, water harvesting systems can provide supplemental water for growing food and fiber crops. Water harvesting is an ancient art practiced in the past in many parts of North America, Middle East, North Africa, China, and India (Oweis et al., 1999). Evenari et al. (1961) documented that water harvesting was used for growing crops in the Negev Desert 4000 years ago; American Indians used similar systems 700 to 900 years ago in the Southwestern United States (Myers, 1975). Chinese farmers used a flood diversion technique called “Warping” (harvesting water as well as sediment) 2700 years ago in China’s loess areas (Li et al., 2000b). Each water harvesting system should have the following four components: (a) runoff producing catchment, (b) runoff collection scheme, (c) runoff storage facility, and (d) cultivated or cropped area (Oweis et al., 1999). There is a general agreement that the first two components are found in all water harvesting systems. Runoff producing catchment is the most important component in a water harvesting system, which is responsible for quantity and quality of water from runoff collection. Since different surface treatments to increase runoff have different runoff response, and consequently affect the required sizes of both the catchment and storage facility (Frasier, 1980), many surface treatments have been proposed and tested throughout the arid and semiarid regions of the world (USDA, 1975; Dutt et al., 1981; Evett and Dutt, 1985). These treatments include mechanical treatments (smoothing and compacting), colloidal dispersion methods (slaking), hydrophobic applications (water repellents), surface binding materials (cementing and sealing) as well as surface covering (asphalt, rubber and plastic) (Tadmor and Shanan, 1969). However, no surface treatment is suitable for all applications. It depends on local rainfall characteristics (amount, intensity and distribution), construction materials, site conditions, installation methods, and labor cost. Therefore, water harvesting techniques do not always transfer well from one set of conditions to another (Ojasvi et al., 1999), and most of these used trial-and-error for the design of water harvesting catchments (Suleman et al., 1995). In the case of semiarid regions of China, rainwater harvesting is being promoted to solve water problem for agricultural and domestic uses in recent years (Li et al., 2000a; Li and Gong, 2002). The current popular catchments for rainwater harvesting are lands, or other

X.-Y. Li et al. / Agricultural Water Management 65 (2004) 211–224

213

surfaces with naturally low infiltration rates that are used for other purposes at the same time. They include rooftops, courtyards, earth, and asphalt-paved roads. Due to their low runoff collection efficiencies and low amounts of collected water, these rainwater harvesting practices in China are still confined to rural family units to supply household water needs and for very limited supplemental irrigation purpose. Therefore, we tested some possible artificial surface treatments to increase the efficiency of runoff collection. Runoff behavior is very important for the successful design of the rainwater harvesting system; early water harvesting researchers often reported results as an “annual runoff percentage" which was described as the percent of annual rainfall that ran off (Evett and Dutt, 1985). Shanan and Tadmor (1979) warned against the use of annual runoff percentage in design of microcatchment systems. Hollick (1982) reported that annual runoff percent suffered from the disadvantage that it gives no indication of the relationship between runoff and rainfall intensity and duration so that extrapolation to new areas or drought years is difficult. Efficient design usually requires information on how rainfall amount and intensity and antecedent rainfall affect runoff from catchments with different surface treatments. The objective of our work was to evaluate runoff characteristics of six artificial surface treatments during naturally occurring rainfall events in the semi-arid loess regions of northwest China.

2. Materials and methods 2.1. Climate This study was conducted during the rainy period from May 1998 to October 1999 at the Gaolan Research Station of Ecology and Agriculture, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences. The station is located in the transitional zone between arid and semiarid regions (Gaolan County, Lanzhou, Gansu Province, 36◦ 13 N, 103◦ 47 E) at an altitude of approximately 1780 m. Mean annual precipitation is 263 mm, with nearly 70% falling between May and September. Mean annual temperature is 8.4◦ C with a maximum temperature of 20.7◦ C (July) and a minimum of −9.1◦ C (January). Average annual pan evaporation is 1785 mm. The soil is a sandy loam (sand: 12.3%; silt: 66.9%; clay: 20.8%) of loess origin, which belongs to Haplic Orthic Aridisols (Li et al., 2000b). 2.2. Experimental design and treatments Six surface treatments were evaluated: concrete, plastic film, gravel covered plastic film, asphalt fiberglass, cleared loess slope, and natural loess slope (Fig. 1). There were three replications for each treatment and the treatment plots were laid out as a randomized design. The plot size was 3.3 × 6.0 m with the longer sides parallel to the slope (14%). Cement block borders, 30 cm high, were installed around each plot to define the catchment areas and to improve the accuracy of runoff measurements. Concrete treatments were constructed using mortar mix of 5:1 (sand:cement) and consisted of several 1 × 2 m concrete slabs with a thickness of 5 cm. Fissures between slabs were sealed with asphalt emul-

214

X.-Y. Li et al. / Agricultural Water Management 65 (2004) 211–224

Fig. 1. A photograph showing experimental plot layout.

sion. Plastic film treatments were established by covering the compacted soil surface with 0.08-mm-thick plastic film. Compaction was achieved with a roller: the bulk density of topsoil reached 1.6 g cm−3 . The only difference between the gravel-covered plastic and plastic film treatment was that the former was covered by a layer of pebbles (7.5–14.0 cm in the long dimension and 5.0–9.0 cm in the short dimension). The purpose of surface pebble mulching was to reduce light radiation to extend the life span of the plastic film. Asphalt fiberglass treatments were installed by spreading asphalt fiberglass on the compacted soil surface. The natural loess slope treatments were the originally undisturbed loess surfaces characterized by microbiotic crust and sparse vegetation. The vegetation consists of Stipa purpurea, Artemisia scoparia, and Waldst. Et Kitag. (Virgate Sagebrush). The vegetation cover varied between 5–10% in drought seasons and 20–60% in rainy seasons. The difference between the cleared loess and natural loess surface treatments was that all plant materials and debris was removed from the soil surface for the cleared loess surface treatments. Debris was cleaned by hand and plants were manually clipped to the ground surface. Runoff from each plot was collected in a 200-l calibrated barrel. The barrel was covered with a plastic sheet to prevent catching precipitation and to prevent evaporation of the collected runoff water. The runoff was measured after each rainstorm, or twice daily during continuous rainfall events. A standard rain gauge and recording rain gauge were used to obtain the amount and intensity of the rainfall. The rainfall-runoff efficiencies (percent of the rainfall) and threshold rainfall amounts (minimum rainfall required to produce runoff) for moist and dry soil surface were determined using linear regression analysis of the separate rainfall events as described by Frasier (1975b) and Diskin (1970). Regression analysis was done using the SPSS (Statistic Package for Social Science) procedure.

X.-Y. Li et al. / Agricultural Water Management 65 (2004) 211–224

215

Table 1 Frequency distribution of rainfall amount and intensity in various class intervals during the experimental period Rainfall intensity I10 (mm h−1 )

Rainfall amount (mm)

Percentage (%)

20

20

60

18

12

10

66

16

13

5

3. Results and discussions 3.1. Rainfall characteristics During the experimental period, there were 91 rainfall events. Annual rainfall was 253 mm in 1998 and 344 mm in 1999; 1999 was a wet year, compared to the average of rainfall of 263 mm. Rainfall was erratic in both years: 40% of the rainfall was concentrated in August during 1998 and in July during 1999. The distribution of daily rainfall and rain intensity is presented in Table 1 and Fig. 2. About 60% of the rainfall events were of less than 5 mm and 5 mm h−1 (I10 ). The data analysis also indicated that about 60% of the total amount of the annual rainfall resulted from rainstorms with over 10 mm of rainfall. Rainfall intensity was generally higher during rainstorms with large amount of rainfall: the correlation coefficient

Daily rainfall (mm)

40 30 20 10 0 May 5 Jun 19 Aug 3 Sep 17 Nov 1 Dec 16 Jan 30 Mar 16 Apr 30 Jun 14 Jul 29 Sep 12

-1

Rain intensity (mm h )

60 50 40 30 20 10 0 May 5 Jun 19 Aug 3 Sep 17 Nov 1 Dec 16 Jan 30 Mar 16 Apr 30 Jun 14 Jul 29 Sep 12

1998

1999

Date Fig. 2. Daily rainfall and rain intensity (I10 ) distribution during the experimental period.

216

X.-Y. Li et al. / Agricultural Water Management 65 (2004) 211–224

Table 2 Monthly runoff (mm) and runoff efficiency (%) and the annual average for the surface treatments in 1998 (The values in parentheses are runoff efficiency calculated as percent of rainfall) Treatment

May

June

July

August

September

October

Annual

Natural loess slope Cleared loess slope Concrete Plastic film Gravel-covered plastic film Asphalt fiberglass

8.1 (15) 9.0 (17) 27.7 (51) 46.3 (86) –

2.2 (10) 2.2 (10) 10.5 (46) 18.5 (82) –

5.4 (10) 6.1 (11) 22.6 (40) 49.1 (87) 44.4 (79)

10.5 (11) 13.1 (14) 43.9 (48) 28.4 (31) 50.1 (55)

0.2 (1) 0.5 (3) 4.2 (31) 0.9 (7) 7.9 (57)

1.3 (9) 2.1 (15) 8.0 (57) 0.6 (4) 8.5 (60)

27.7 (11) 32.9 (13) 116.9 (46) 143.9 (57) 110.9 (56)

43.7 (81)

17.2 (76)

49.5 (88)

54.9 (60)

10.6 (77)

12.0 (85)

187.8 (74)

between rainfall and rain intensity was 0.587 (F(1,67) = 37.72, P < 0.0001) (Li and Gong, 2002). The results suggest that most storms are of small size with low intensity, and that the total amount of annual rainfall mainly depends on a few larger storms. The latter typically occurred during the monsoon period in the region. 3.2. Runoff and runoff efficiency characteristics Tables 2 and 3 present monthly runoff and runoff efficiency and the annual average for the surface treatments. The average monthly or annual runoff efficiency was calculated by dividing the monthly or yearly total volume of runoff by the corresponding total volume of rainfall. Runoff was generally higher for all the treatments in the months between May and August than in April, September, and October in accordance with rainfall distribution. Monthly runoff efficiency was higher and varied between 31% and 76% for the concrete and 60% and 83% for the asphalt fiberglass as compared to 3% and 94% for the plastic film and 4% and 79% for the gravel-covered plastic film. Monthly runoff efficiency was low for the cleared loess slope (3–23%) and the natural loess slope catchments (1–30%). The large magnitude of runoff efficiency variation for the plastic film and the gravel-covered plastic film was attributed to the fact that plastic film deteriorated by weathering after 5 months. Gravel cover extended the longevity of the plastic film for 1–2 months. Average annual runoff efficiency was consistently higher in 1999 than in 1998 for all the treatments (Tables 2 and 3). This was attributed to more rainfall events with high amount Table 3 Monthly runoff (mm) and runoff efficiency (%) and the annual average for the surface treatments in 1999 (The values in parentheses are runoff efficiency calculated as percent of rainfall) Treatment

April

May

June

July

August

September October Annual

Natural loess slope Cleared loess slope Concrete Plastic film Gravel-covered plastic film Asphalt fiberglass

4.6 (30) 3.5 (23) 6.6 (45) 10.3 (69) 7.2 (48)

6.1 (9) 5.0 (8) 47.0 (72) 57.0 (87) 55.6 (85)

3.7 (6) 5.0 (8) 46.7 (76) 57.9 (94) 54.4 (88)

15.0 (11) 24.3 (17) 105.5 (74) 126.2 (89) 123.1 (87)

0.1 (0) 0.8 (3) 13.8 (57) 7.7 (32) 16.5 (68)

0.2 (1) 1.2 (5) 12.9 (48) 2.4 (9) 7.0 (26)

9.5 (64)

52.6 (80) 51.0 (83) 117.6 (83) 19.5 (81) 22.2 (82)

0 (0) 0 (0) 4.0 (45) 0.3 (3) 0.4 (4)

29.6 (9) 39.8 (12) 236.5 (69) 261.8 (76) 264.2 (77)

6.5 (74) 279.0 (81)

X.-Y. Li et al. / Agricultural Water Management 65 (2004) 211–224

217

and intensity occurred in 1999 than in 1998 (Fig. 2). Similar trends occurred for monthly runoff efficiency: runoff efficiency was high in some months when more rainfall events with high amount and intensity occurred (Tables 2 and 3, Fig. 2). The results of this experiment indicated that the asphalt fiberglass catchment had the highest average annual runoff efficiency of 74–81%, followed in decreasing order by plastic film (57–76%), gravel-covered plastic film (56–77%), concrete (46–69%), cleared loess slope (12–13%), and natural loess slope catchment (9–11%). Runoff observations for the six surface treatments were also undertaken at intervals of about 20 min during the rainfall events of 1 July and 2 July 1999. There had been no rain for 20 days before July 1, so the runoff-producing process during the rainy days of July 2 and 1 would represent runoff with and without the effects of an antecedent rainfall. Fig. 3 shows cumulative rainfall and runoff from the six treatments measured during the rain of July 1. This rain lasted 5.66 h and the total amount was 18.2 mm. Runoff produced in the natural loess slope and the cleared loess slope plots occurred within 40 min of the onset of rainfall with a high intensity of 7.8 mm h−1 . In contrast, continuous runoff production occurred in the asphalt fiberglass, plastic film, gravel-covered plastic film, and concrete plots with an intensity that did not exceed 1.2 mm h−1 . The order of runoff generation for the six treatments was similar to that of annual runoff and runoff efficiency distribution (Tables 2 and 3, Fig. 3). Similar trends occurred during the 26.5 mm rainfall with a duration of 11.50 h on July 2 (Fig. 4), but greater runoff occurred due to less initial rainfall interception. The runoff efficiencies were 87, 94, 89, 81, 11, and 2% for the asphalt fiberglass, plastic film, gravel-covered plastic film, concrete, cleared loess slope, and natural loess slope treatments on 2 July,

Cumulative rainfall and runoff (mm)

20 15

Rainfall Concrete cleared loess slope plastic film Asphalt fiberglass Gravel covered plastic film natural loess slope

10 5

0.4

0.2

July 1 0.0 05:00

06:00

07:00

08:00

09:00

10:00

11:00

12:00

Time Fig. 3. Cumulative rainfall and runoff from the six surface treatments on 1 July 1999.

218

X.-Y. Li et al. / Agricultural Water Management 65 (2004) 211–224 35

Cumulative rainfall and runoff (mm)

30 25 20

Rainfall Concrete Cleared loess slope Plastic film Asphalt fiberglass Gravel-covered plastic film Natural loess slope

15 10 5 2.5 2.0 1.5 1.0 0.5

July 2

0.0 09:36

12:00

14:24

16:48

19:12

Time Fig. 4. Cumulative rainfall and runoff from the six surface treatments on 2 July 1999.

respectively, as compared to 78, 86, 83, 68, 3, and 1% on 1 July, respectively. This suggests that antecedent rainfall can affect runoff generation process. 3.3. Effects of rainfall parameters on runoff and runoff efficiency for various surface treatments Runoff from all the treatments increased with the increase in the amount and intensity of the rainfall (Table 4). Runoff and the amount and intensity of the rainfall were significantly correlated: the correlation coefficient of the runoff and the amount of the rainfall was about 0.60 for the cleared loess slope and natural loess slope and 0.81–0.99 for the asphalt fiberglass, plastic film, gravel-covered plastic film, and concrete treatments. However, the correlation coefficient of runoff and the rainfall intensity was over 0.90 for the cleared loess slope and natural loess slope and 0.46–0.58 for the asphalt fiberglass, plastic film, gravel-covered plastic film, and concrete treatments. Multiple regression equations of runoff with the rainfall amount and intensity indicated that runoff increased 52–87% for the asphalt fiberglass, plastic film, gravel-covered plastic film and concrete treatments with an increase in the rainfall and about 5–7% for the cleared loess slope and natural loess slope treatments. Increase in the intensity resulted in increase in the runoff by 12–17% of the total rainfall intensity for the cleared loess slope and natural loess slope treatments and 0.3–7% for the asphalt fiberglass, plastic film, gravel-covered plastic film, and concrete treatments. This suggests that runoff during a particular rain spell is more governed by the amount of the rainstorm for the asphalt fiberglass, plastic film, gravel-covered plastic film, and concrete than for the cleared loess slope and natural loess slope. Rain intensity is important for

X.-Y. Li et al. / Agricultural Water Management 65 (2004) 211–224

219

Table 4 Regression equationsa and coefficient values between runoff and the amount and intensity of rainfall Treatment

Equation

r

F value

Significance level

Natural loess slope Cleared loess slope Concrete Plastic film Gravel-covered plastic film Asphalt fiberglass Natural loess slope Cleared loess slope Concrete Plastic film Gravel-covered plastic film Asphalt fiberglass Natural loess slope Cleared loess slope Concrete Plastic film Gravel-covered plastic film Asphalt fiberglass

R = −0.27 + 0.14 P R = −0.43 + 0.19 P R = −0.46 + 0.60 P R = −0.0050 + 0.88 P R = −0.30 + 0.66 P R = −0.28 + 0.85 P R = −0.43 + 0.12 I R = −0.60 + 0.19 I R = 1.34 + 0.28 I R = 4.15 + 0.36 I R = 1.33 + 0.36 I R = 2.57 + 0.29 I R = −0.62 + 0.046 P + 0.12 I R = −0.87 + 0.068 P + 0.17 I R = −0.77 + 0.52 P + 0.0072 I R = −0.093 + 0.87 P + 0.0094 I R = −0.54 + 0.55 P + 0.071 I R = −0.33 + 0.82 P + 0.0027 I

0.57 0.60 0.83 0.99 0.81 0.99 0.91 0.92 0.58 0.46 0.61 0.57 0.93 0.94 0.86 0.99 0.84 0.99

19.90 23.17 96.96 3948.63 75.45 2623.84 208.04 263.85 21.41 13.97 23.29 19.43 127.38 156.96 57.38 1994.81 44.11 1475.68