Content uploaded by Seth Cook
Author content
All content in this area was uploaded by Seth Cook on Mar 31, 2015
Content may be subject to copyright.
477Ambio Vol. 29 No. 8, Dec. 2000 © Royal Swedish Academy of Sciences 2000
http://www.ambio.kva.se
INTRODUCTION
Rainwater harvesting implies collection and storage of the rainy
season precipitation that would have seeped into soil or run off
into stream channels. It is an old technique that was probably
developed as long ago as 4500 B.C. (1, 2). Although earlier rain-
water harvesting systems were designed primarily to meet do-
mestic needs for water, in recent decades, scientists in many
countries such as Sub-Saharan Africa, the Middle East and
Southeast Asia, and especially India, have made efforts to de-
sign and develop a wide variety of techniques to collect, store,
and use natural precipitation for agricultural purposes (3–13). In
some countries, development of rainwater harvesting systems is
being promoted by the authorities as an alternative to the high-
cost large dams and water development projects (14). China has
a long history of rainwater harvesting techniques in many wa-
ter-deficient areas such as the Loess Plateau. This dates back to
the Qing and Han Dynasties about 2000 years ago (15). While
enabling people to survive in drought-prone environments, the
water yield from early rainwater harvesting techniques is not suf-
ficient for modern agricultural purposes (15). Since the mid-
1980s, scientists in Gansu province have worked to design mod-
ern forms of rainwater harvesting, which will be suitable for ag-
ricultural irrigation. An effectively designed system for water
management on rainfed cultivated land was first developed over
a decade ago. This system is termed rainwater harvesting agri-
culture (RHA). During the last 10 years, RHA has been success-
Article Fengrui Li, Seth Cook, Gordon T. Geballe and William R. Burch Jr
Rainwater Harvesting Agriculture:
An Integrated System for Water Management on
Rainfed Land in China’s Semiarid Areas
fully applied by many household farmers in semiarid areas of
Gansu and other provinces in northwest China. The purpose of
this paper is to give a general picture of the principles and meth-
ods of RHA as well as its recent development in China’s semi-
arid areas, especially in Gansu province.
WATER CONSTRAINTS TO RAINFED FARMING IN
CHINA’S SEMIARID AREAS
China’s semiarid areas extend across almost the whole country
(Fig.1). It includes much of northwestern China, particularly
Gansu, Qinghai, Ningxia, Shanxi, Shannxi and Inner Mongolia
provinces with a total area of 1.7 mill. km2, about 18% of the
nation’s territory. Annual rainfall throughout the region ranges
from 250 to 600 mm. Annual accumulated temperature, ≥10°
ranges from 1500 to 4500°C. Annual frost-free period varies
from 120 and 190 days. The major soil categories in the culti-
vated land area are loamy sand, loess sandy loam and black
sandy loam. These soils are loosely structured and coarse tex-
tured and thereby are highly susceptible to erosion by water and
wind (16). Three principal land-use management systems domi-
nate this region. In some drier parts of the region where aver-
age annual rainfall is below 300 mm and no water resources are
available for agricultural irrigation, rangeland pastoralism is
commonly practiced. In other drier parts of the region where,
although average annual rainfall is less than 300 mm, surface
and/or ground-water resources are available, irrigated agricul-
ture is usually carried out. In a large area of the region receiv-
ing a mean annual rainfall of over 300 mm, rainfed farming is
the most widespread land-use practice. Overall, rainfed cropland
occupies nearly 80% of the total cultivated land in the region
(17).
Although the level of agricultural productivity in the prevail-
ing rainfed farming system is restrained by many biophysical fac-
tors, the most serious limit is low availability of water. This re-
Rainwater harvesting agriculture (RHA), which was first
developed by scientists in Gansu province over a decade
ago, is an integrated system for water management on
rainfed land in semiarid areas. This system consists of
three main components including rainwater harvesting
system, water-saving irrigation system, and highly effective
crop production system. Its main function is to provide
farmers in water-limiting environments with access to the
water needed to meet domestic and agricultural water
needs. The preliminary implementation of RHA in Gansu
and other provinces in northwest China suggests that RHA
has the potential to improve performance in rainfed farming
systems and to address environmental problems such as
soil erosion. The small-scale and low cost of RHA systems
make application by household farmers simple. However,
to be successful RHA needs to be integrated in a com-
prehensive agricultural-management system; i.e. manage-
ment of RHA must be combined with other agricultural
technologies and management practices. In addition, the
spread of RHA over large areas entails consideration of a
range of technological, agrohydrological, ecological, social,
cultural, economic, and political factors. In particular, there
is a need to provide training and extension services to
farmers, to develop and disseminate more effective and
affordable types of RHA technologies as alternatives and
to design and develop alternative policy instruments and
social institutions that facilitate adoption of RHA practices.
Figure 1. Map of rainfall distribution in China.
Gansu province
478 © Royal Swedish Academy of Sciences 2000 Ambio Vol. 29 No. 8, Dec. 2000
http://www.ambio.kva.se
gion is strongly governed by the prevailing monsoon climate,
whereas the monsoon rains are highly variable in terms of their
amount and reliability. Hence, not only is the total amount of
rainfall often inadequate, but also it is subject to the high inter-
annual and interseasonal variations (18, 19). On average, ap-
proximately 60% of the annual rainfall usually falls during the
three months between July and September, often in the form of
heavy thunderstorms. This causes not only tremendous amounts
of erosion, but also problems of frequent and serious spring and
early-summer drought for both winter and summer crops. Some
studies in a semiarid region of Gansu province show that the
main water stress period for winter crops, such as winter wheat,
often occurs from May until the start of the rainy season in July,
whereas the main water-stress period for summer crops such as
corn often occurs during the period from mid-June to early-Au-
gust (Table 1). As the periods of water stress for both winter
and summer crops coincide with their heading and grain-filling
growth stages, it is extremely unfavorable for grain production
(20). For this reason, providing supplemental irrigation to crops
during their water-stress periods is essential to achieve a high,
sustainable yield in rainfed conditions.
For a long period of time, particularly since the founding of
the People’s Republic of China in 1949, the main strategy for
agricultural resource management on rainfed land in China’s
semiarid areas has been known as water-soil conserving farm-
ing (WSCF) (21, 22). A significant feature of WSCF is the con-
struction of terraced fields on hillsides to retain runoff and pre-
vent soil erosion. The principal technological components of this
system include intensive cultivation (e.g., ploughing, raking and
levelling of cropland), integrated application of inorganic and
organic (e.g., farm and green manure) nutrients, stubble, gravel
and plastic film mulches, cropping system practices (e.g., crop
rotation, ley farming, intercropping and multiple cropping) and
the choice of improved crop varieties. Many studies indicate that
adoption of WSCF practices has led to a significant increase in
crop productivity on rainfed land as a direct result of improved
availability of water and nutrients for crops. For instance, the
average yields of wheat crops on terraced fields have increased
from 1500–1800 kg ha–1 in the 1950s to 2200–2500 kg ha–1 in
the late-1980s, in areas with annual rainfalls of about 450 mm
(22). At the same time, WSCF has also played a beneficial role
in conserving soil and water resources. In spite of being invalu-
able for alleviating erosion and increasing productivity, conven-
tional WSCF can not solve the fundamental problem of poor
water availability (23). At present crop production levels, the wa-
ter availability for crops is below minimum requirements for full
yields in the semiarid areas. Thus, inadequate water supplies are
a major constraint for sustainable crop production. An increase
in crop yield will depend on whether water availability can be
increased.
According to our study in a semiarid region of Gansu prov-
ince (19), the relative water satisfaction of winter wheat, defined
as the percentage of growing-season precipitation relative to crop
water requirements for full yield is 62% in terms of the whole
phenology and only 35, 41, and 40% for the 3 critical growth
stages of jointing, heading and grain-filling. The relative water
satisfaction of corn is 87% for the whole phenology, but only
55 and 79% for the 2 critical growth stages of jointing and ear-
ring (Table 2). Some other studies in the same region also indi-
cate that the relative water satisfaction of grain crops ranges from
60 to 90% in areas with annual rainfall between 450 and 600
mm and only from 40 to 60% in areas with annual rainfall be-
tween 300 and 450 mm (24). Such a large water deficit is im-
possible to compensate for simply by implementing conventional
WSCF practices (23).
RAINWATER HARVESTING AGRICULTURE: AN
INTEGRATED SYSTEM FOR WATER MANAGEMENT
ON RAINFED LAND
Rainwater Harvesting Agriculture and Its Technological
Systems
It is evident that rapid, effective and low-cost solutions to the
problems relating to inadequate water availability are necessary
to increase yields and sustainable development of rainfed farm-
ing in China’s semiarid areas.
The development of large-scale irrigation development
projects which depend largely on massive and permanent water
sources such as stream water and river water, may offer a po-
tential solution to the problem, but this is not always the case
for several reasons. (i) Both surface and groundwater resources
are scarce in most of China’s semiarid areas. In some parts of
the region although groundwater is available, it is often highly
Table 1. Relationship between the supply and demand of rainfall
during different growth stages of winter wheat and corn in a
semiarid region of Xifeng (long-term mean annual rainfall
561 mm) in Gansu province.
Rainfall Potential Water Relative
(mm) ET deficit crop water
Phenology mean ± SD (mm) (mm) satisfaction (%)
Winter wheat
Sowing-regrowth 122±52 105 – 100
Regrowth-elongation 34±19 98 –64 35
Elongation-heading 48±31 116 –68 41
Heading-grainfilling 42±29 103 –61 41
Grainfilling-mature 41±32 36 – 100
Whole phenology 286±75 458 –172 62
Corn
Sowing-elongation 109±44 199 –90 55
Elongation-earing 97±51 122 –25 79
Earing-grainfilling 106±58 116 –10 91
Grainfilling-mature 88±54 34 – 100
Whole phenology 399±96 461 –62 87
% = as a percentage of the rainfall during different growth stages relative to
potential evapotranspiration (i.e. crop water requirements for full yield).
Table 2. Onset, duration and end of the water stress periods for winter wheat and corn
with different soil layers in a representative dry and wet year in a semiarid region of
Xifeng in Gansu province.
Dry year Wet year Duration (days)
Layer
(m) Onset End Onset End Dry year Wet year
Winter wheat
0.0–0.5 17 May 13 July 8 May 26 June 58 50
0.5–1.0 24 May 16 July 24 May 6 June 54 14
1.0–2.0 8 June 14 July None None 37 None
Corn
0.0–0.5 10 June 14 Aug. None None 66 None
0.5–1.0 5 June 22 Aug. None None 18 None
1.0–2.0 None None None None None None
479Ambio Vol. 29 No. 8, Dec. 2000 © Royal Swedish Academy of Sciences 2000
http://www.ambio.kva.se
saline or brackish and thus cannot be used for either irrigation
or human consumption. (ii) Almost 80% of the cultivated land
in the region consists of highly fragmented hillside slopes re-
sulting from the extensive mountainous and hilly topography.
In the case of Gansu province, some 62% of the cultivated land
is slopes with a steepness of between 3 and 15°, 23% consists
of slopes with between 15 and 25° and 6% has slopes steeper
than 25°. From an ecological perspective, less than 10% of
Gansu’s cultivated land is suitable for agricultural activities. The
topography makes large-scale irrigation projects difficult to de-
velop. (iii) The soils in the region are loosely structured and
highly susceptible to wind and water erosion and thus the con-
struction of large-scale irrigation works may cause environmental
problems such as subsidence and soil erosion. (iv) The high costs
of large-scale irrigation development projects are prohibitive in
poor areas with very limited financial resources. (v) High con-
struction, operation and management costs are likely to trans-
late into relatively high prices for the use of irrigation water,
which farmers in poor areas are virtually unable to afford.
If large-scale irrigation development projects are not the an-
swer to the problem of water shortages in semiarid mountain-
ous and hilly areas in northwestern China, then what is? In or-
der to be feasible, any workable solution must be affordable and
compatible with the biophysical and socioeconomic conditions
of the region.
Rainwater harvesting agriculture (RHA) based on the inclu-
sion of rainwater harvesting techniques may offer a solution.
RHA is an integrated system for water management on rainfed
land in semiarid areas. This system can help farmers to allevi-
ate water constraints. The RHA system consists of a synergetic
combination of different technological components including a
rainwater harvesting system, a water-saving irrigation system,
and a highly effective crop production system. Among these 3
components, the rainwater harvesting system is the core, its func-
tion is to provide water for domestic and agricultural needs. An
effective water-saving irrigation system is crucial for the eco-
nomical and effective use of water.
Scientists in China have developed a wide variety of simple,
low-cost mobile and semi-fixed micro-drip irrigation systems
that are suitable for small-scale farmers (25). However, only
through integration of the components can the expected objec-
tive of agricultural water management be achieved.
The rainwater harvesting system consists mainly of collection
surface, runoff channel, sediment tank and storage container (Fig.
2). Collection surfaces can be waterproofed by compacted dirt,
melted wax, asphalt, concrete, and plastic films and the like.
They are usually built in courtyards for household use and sur-
rounding or above agricultural fields for irrigation purposes. In
many areas, roofs, courtyards, rocky hillsides, and asphalt roads
are used as collection surfaces. Containers for the storage of rain-
water include wells, tanks, mini-reservoirs etc. In the loess ar-
eas the most common storage containers are wells. There are two
basic types of the wells, namely traditional earthen wells and
Figure 2. A schematic
diagram showing the
small-scale rainwater
harvesting system
being widely used by
household farmers in
semiarid areas of
Gansu province.
Rainwater storage wells with concrete collection surfaces for irrigation,
Yuzhong County, Gansu Province, July 1997. Photo: S. Cook.
Manual water pump
Ground
Inlet
Sediment tank
Runoff channel
Collection surface Water intake
Concrete well wall
Main pipe
Lateral line
Valve
Outlet
Well cover
Branch
pipe
Washing
valve
Filter
Well bottom
20 cm
450 cm
470 cm
600 cm
30 cm 80 cm
480 © Royal Swedish Academy of Sciences 2000 Ambio Vol. 29 No. 8, Dec. 2000
http://www.ambio.kva.se
tensive mountainous and hilly topography of the semiarid areas.
(ii) The small scale of this system makes it easy to construct,
operate and manage at the small farm level, which is the basic
unit of agricultural production in China today. (iii) The owner-
ship of rainwater harvesting facilities such as collection surfaces
and storage wells belong to individual households and the farm-
ers have a greater incentive to maintain them. (iv) Unlike large-
scale water development projects, this small-scale system does
not require long construction delays, and can provide benefits
the same year they are begun. (v) The construction of small-
scale rainwater harvesting systems requires relatively low invest-
ments and is more affordable to farmers in poor areas. For in-
stance, the average cost (including labor and material inputs) for
building a 30 m3 concrete well is RMB 1250 yuan (about USD
125), because of the low cost of labor in China. When compari-
son is made on a per unit irrigated area basis, the average cost
of small-scale rainwater harvesting systems amounts to only 50–
65% of that of conventional large-scale water development
projects (26).
The scale of the rainwater harvesting system to irrigate a small
farm can be determined according to (i) the annual amount of
runoff production per unit of catchment area, and (ii) the actual
area to be irrigated and the amount of irrigation water required
per unit of irrigated area under limited irrigation conditions. The
annual amount of runoff production per unit of catchment area
in a given area can be estimated by the following formula:
Yw = Cr x Hp/1000 Eq. 1
where Yw is the annual amount of runoff production per unit of
catchment area (m3 m–2); Cr is the runoff coefficient from a par-
ticular collection surface; and Hp is the annual rainfall depth
(mm) under different rainfall probabilities.
Because of the variability in rainfall, runoff coefficients of
the collection surfaces vary with the characteristics of rainfall.
Table 3 presents the annual average values of runoff coefficients
for different collection surfaces. The data in the Table are from
a 4-year runoff plot study conducted by a research team from
the Gansu Water Conservancy Science Institute (27). In this
study, researchers observed the runoff coefficients of 9 differ-
ent collection surfaces over 600 different rainfall events (vary-
ing intensities, depths, and antecedent moisture content) in sev-
eral representative sites of the semiarid areas of Gansu province.
Given crops to be irrigated, their annual amount of irrigation
water requirements per unit of irrigated area can be determined
using the following formula:
Iq= (Rw – 10Pe – Ws)/η Eq. 2
where Iq is the annual amount of irrigation water requirements
(m3 ha–1); Rw is crop water requirements for full yield, namely
potential evapotranspiration (m3), which can be calculated by a
modified Penman equation (28); Pe is the rainfall during the crop
growing season (mm); Ws is the available water storage in the
soil before planting (mm), which can be determined according
to field experiments or can be roughly estimated based on
(0.15~0.25) Rw in the absence of field data; and η is the factor
of water use and its values can be set 0.8~0.9 when drip irriga-
tion system is used.
According to field studies, using a drip irrigation system it is
possible in semiarid environments, to irrigate crops such as
spring or winter wheat 2 to 3 times during the whole growing
season, with a water quantity of 150–225 m3 ha–1 for each pe-
riod. At this level of supplemental irrigation, irrigated crops can
produce an average of 20–40% more yield than unirrigated crops
(29–31).
Effects of RHA Practices on Improving Agricultural
Production
As part of a long-term research effort aimed at establishing a
productive, sustainable RHA system in the semiarid areas of
China, a series of field studies has been carried out in a joint
research project between Lanzhou University and the Institute
of Dryland Farming in Gansu Academy of Agricultural Sciences,
in several counties of the Gansu province (29, 30). The effects
of RHA in improving production can be illustrated using Dingxi
county as a case study. In 1993, field research was conducted
in the Dingxi Agricultural Experimental Station. Long-term
(1960–1990) average climatic data for the study area are 415 mm
rainfall, 6.2°C mean temperature, 1432 mm open-pan evapora-
tion and a frost-free period of 140 days. For the year of the study,
annual rainfall was 420 mm, close to the 30-year average, and
the rainfall during the crop growing season was 289 mm, i.e.
slightly higher than the 30-year average (279 mm). Hence, this
study can be considered representative for long-term averages.
Spring wheat (Triticum aestivum, cv. Longchun 8139-2) was
planted at a seeding rate of 225 kg ha–1 in late March. The plot
size was 2.6 m . 5 m with a 1.2 m space between plots. Plots
Rainwater storage wells with plastic collection surfaces for irrigation, Dingxi County,
Gansu Province, June 1997. Photo: S. Cook.
modern concrete wells. Wells have vari-
ous shapes such as vase-like, ball-like
and column-like. Earthen wells are usu-
ally built by digging into the ground,
combined with special treatment using
cement or red clay soil to prevent seep-
age loss. To prevent evaporation loss, a
finished well often has a concrete cover.
Geohydrological research suggests that
in the loess regions the optimal volume
of an earthen well is 15–20 m3, and the
optimal volume of the concrete wells is
30–50 m3 (15). However, in some rocky
mountainous areas like Weihui county of
Henan province in central China, some
larger wells, the so-called ‘water cave’,
with a volume of over 500 m3 are built
by digging into rocky cliff sides. These
large wells are suitable for flood waters
collected during intense rainstorms.
Compared with large-scale water devel-
opment projects the rainwater harvesting
system shows the following superior
qualities. (i) It is well suited to the ex-
481Ambio Vol. 29 No. 8, Dec. 2000 © Royal Swedish Academy of Sciences 2000
http://www.ambio.kva.se
were arranged as a randomized complete block with 3 replica-
tions. Because the soil of this area is naturally rich in K and de-
ficient in N and P, farm manure was extensively applied to the
experimental plots prior to planting. In addition, according to soil
test recommendations, adequate amounts of chemical fertilizer
(bi-ammonium phosphate and urea) were applied at seeding
depth to provide 90 kg N and 60 kg P2O5 ha–1. The experiment
included 3 irrigation treatments: namely irrigating once in the
heading stage, twice in the tillering and heading stages and three
times in the tillering, jointing, and heading stages. Nonirrigated
spring wheat served as a control. In each treatment, the amount
of water supplied to the crop was approximately 350 m3 ha–1.
This study shows that average yields of spring wheat irrigated
for one, two and three times were 4.70, 5.02 and 5.49 Mg ha–1,
respectively, i.e. 28, 37 and 50% higher than the control (3.66
Mg ha–1). The water supply efficiency, defined as yield incre-
ment per unit of water supplied, of the 3 treatments averaged
2.97, 1.94 and 1.74 kg m–3, respectively.
In another study at the same site, 4 improved cultivars of corn
(Zea mays) including Lindan 160, Lindan 141, Jiudan 3 and
Danyu 13 were planted in rows (with a population of 60␣ 000
ha–1) in late April under the plastic film mulch condition. Ad-
equate amounts of fertilizers N and P were applied to the ex-
perimental plots according to soil test recommendations. All the
cultivars received supplemental irrigation of about 750 m3 ha–1
during the earring stage (a most sensitive period to water stress),
with a nonirrigated control. The 297 mm rainfall occurred dur-
ing the corn growing season. This study indicates that the aver-
age yields of irrigated cultivars were 9.05 (Lindan 160), 8.90
(Lindan 141), 8.58 (Jiudan 3) and 7.00 Mg ha–1 (Danyu 13), i.e.
88, 31, 34 and 20% higher than the nonirrigated cultivars. The
water supply efficiencies of the 4 cultivars averaged 6.20 (Lindan
160), 2.83 (Lindan 141), 2.91 (Jiudan 3) and 1.53 kg m–3 (Danyu
13). Furthermore, a field study involved in supplemental irriga-
tion of rainfed fruit trees was also carried out at the same site.
In this study, 3 cultivars of pear (Pyrus spp.) including Zaoshu,
Jinfeng and Korea Yangli were used. For each cultivar, 3 same-
sized, 5-year-old pear trees were chosen as a cohort with 3 rep-
licates. All pear trees were irrigated 4 times using the sub-
surface drip irrigation during the period from flowering to fruit
expanding stages, with a nonirrigated control. Each time, the
amount of water supplied was 15 kg tree–1, about 60 kg tree–1 in
total. This study shows that the average fruit yields per tree for
the 3 irrigated cultivars were 26.2 (Zaoshu), 34.4 (Jinfeng) and
23.6 kg (Korea Yangli), with increases of 58, 83 and 102% as
compared with nonirrigated trees. The average water supply
efficiencies of the 3 cultivars were 0.16 (Zaoshu), 0.26 (Jinfeng)
and 0.20 kg kg–1 (Korea Yangli), respectively. At the same time,
irrigation treatments significantly improved the quality of the
fruit. All these data suggest that implementing RHA practices
can improve performance in rainfed farming systems in semiarid
areas.
Environmental Benefits of RHA Practices
Problems of environmental degradation, particularly soil erosion
in China’s semiarid areas has received a great deal of attention
both nationally and internationally (32–34). These problems have
a biophysical root cause related to the climatic, topographic and
geological features of the region, which combine to produce an
erosion-prone landscape (35). However, soil erosion is closely
tied to poor land-use management. Since the beginning of the
century, particularly over the last several decades, rapidly grow-
ing human population has placed severe pressure on productive
soil resources, thus leading to a rapid decline in the average per
capita cultivated land in the region. In the case of Gansu prov-
ince, its total population has increased from about 11.4 mill. in
the early 1950s to 25. 6 mill. in the late 1990s, with an average
growth of over 320␣ 000 persons every year. As a result, the av-
erage per capita cultivated land has decreased from approxi-
mately 0.4 ha in the early 1950s to less than 0.2 ha in the late
1990s. To meet the increasing food needs resulting from popu-
lation growth, a considerable increase in food and fiber produc-
tion will be needed. Under these conditions, farmers were forced
to convert more and more forestland and grassland into cropland
and at the same time to increase cultivation of steep erodible
slopes in order to obtain an adequate harvest. Consequently, this
has caused the increase in the scale and severity of soil erosion
and a reduction in soil fertility, which are major factors affect-
ing the sustainability of agricultural system. It is estimated that
about 75% of the Loess Plateau in China has been severely af-
Table 3. Average annual values of runoff coefficients (%) for the 9 different collection surfaces
under different annual rainfall and rainfall probabilities (%).
Average Different collection surfaces
rainfall
(mm) Probability
(%) T1 T2 T3 T4 T5 T6 T7 T8 T9
50 80 75 50 40 53 25 46 68 8
(40) (38) (25) (20) (27) (13) (23) (34) (4)
400~500 75 79 74 48 38 25 23 45 67 7
(39) (37) (24) (19) (23) (12) (23) (34) (4)
90 76 69 39 31 41 19 36 65 6
(38) (35) (20) (16) (21) (10) (18) (33) (3)
50 80 75 49 40 52 26 46 68 8
(32) (30) (20) (16) (21) (10) (18) (27) (3)
300~400 75 78 72 42 34 46 21 41 66 7
(31) (29) (17) (14 (18) (8) (16) (26) (3)
90 75 67 37 29 40 17 34 64 5
(30) (27) (15) (12) (16) (7) (14) (26) (2)
50 78 71 41 34 47 20 41 66 6
(24) (21) (12) (34) (14) (6) (12) (20) (2)
200~300 75 75 66 34 28 40 17 34 64 5
(23) (20) (10) (28) (12) (5) (10) (19) (2)
90 73 62 30 24 33 13 28 62 4
(22) (19) (9) (24) (10) (4) (8) (19) (1)
T1–Concrete; T2–Cement tile; T3–Machine-made brick tile; T4–Traditional brick tile; T5–Compacted surface of mixed red clay and
loess soil; T6–Compacted surface of loess soil; T7–Gravel-covered plastic film; T8–Asphalt road; T9–Untreated (natural) loess soil
surface. The values in parentheses are the annual amount of runoff production per unit of catchment area (m3 100 m–2) with 9 different
collection surfaces.
482 © Royal Swedish Academy of Sciences 2000 Ambio Vol. 29 No. 8, Dec. 2000
http://www.ambio.kva.se
fected by soil erosion (36). Each year nearly 1600 mill. tonnes
(Mt) of topsoil are lost from the region through water and wind
erosion, with associated loss of about 8 Mt of organic matter,
2.4 Mt of nitrogen, 3.2 Mt of potassium and 0.02 Mt of avail-
able phosphorus (37). The severity of this situation becomes ap-
parent when it is recognized that soil is formed only at approxi-
mately 1 t ha–1 yr–1 (38). To combat erosion and agricultural soil
degradation, both the central and regional governments have for-
mulated a series of policies to prohibit cultivation of steep slopes,
particularly slopes steeper than 25° (39). However, this limit is
often breached, and the cultivation of steep slopes is still going
on in many areas. An important cause for this is the low and
unstable productivity in rainfed land. This leaves farmers with
no choice other than to put more sloping lands into production.
It is evident that the key to prevent soil erosion is to stop culti-
vation of steep slopes, whereas the key to stop this cultivation
is to increase agricultural productivity in existing rainfed flat and
terraced fields by developing and adopting more effective wa-
ter management practices (40).
RHA has the potential to prevent soil erosion in two ways.
(i) RHA is in itself a useful measure in soil and water conserva-
tion, by capturing and storing runoff during heavy rainstorms,
it can directly contribute to the reduction of soil and water ero-
sion. Studies show that over 70% of the soil erosion that occurs
in the semiarid mountainous and hilly areas is caused by a few
intense rainstorm events (41). (ii) RHA is also an effective yield-
increasing practice, by implementing supplemental irrigation to
crops, it can significantly increase productivity of rainfed land
because of improved availability of water. While increased pro-
ductivity in flat and terraced fields can help reduce the incen-
tives that farmers have to cultivate steep slopes as an insurance
strategy against crop failure. Taken out of grain production, slop-
ing lands can be used to develop protective economic forest or
grassland production. This not only provides a far better soil
cover than grain crops, but it is also beneficial to the adjustment
of the structure of agriculture, decreasing the production of sin-
gle cereal crops and increasing the production of livestock. A
recent survey study in several selected villages which have
adopted RHA practices, in Dingxi and Yuzhong counties of
Gansu province, shows that adoption of RHA practices can re-
ally help decrease the incentives that local farmers have to cul-
tivate fragile sloping lands (42). A significant change is that more
and more villagers in these villages have already started to give
up the cultivation of steep slopes because they have been able
to produce adequate grains to meet their practical needs simply
by cultivating more productive flat and terraced fields. Some vil-
lagers use these retired sloping lands to grow either perennial
legumes for animal feed or economic trees for market sales, thus,
increasing cash income and the standard of living for the vil-
lagers.
SUCCESSFUL IMPLEMENTATION OF RHA IN
GANSU PROVINCE
RHA was first practiced in Gansu. Since 1995, RHA has been
promoted by the authority as a ‘solution’ to the problem of wa-
ter shortages, thus leading to a rapid development of RHA in
the province. In order to bring about a massive application of
this technology, the Gansu provincial government launched a
program, the so-called 1-2-1 system, in 1995. Initially, the pri-
mary focus of the program was the central parts of the province,
an area with the most serious water shortages. This program built
upon prior success in fulfilling household domestic and produc-
tive requirements for water on an experimental scale. Its pur-
pose is to help each household farmer build about 100 m2 con-
crete collection surfaces, two concrete storage wells and irrigate
one mu (1/15 ha) of cropland for production of high market value
cash crops (e.g. vegetables). In this program, the government
provided farmers with materials (e.g. steel and cement) for build-
ing the collection surfaces and storage wells and necessary tech-
nical services, whereas farmers contributed the labor. In addi-
tion, the farmers involved in the program can also obtain finan-
cial support in the form of microcredits from the ongoing UNDP-
assisted projects in the province. After implementation of this
program for 3-years there are already significant changes. The
consequences are that approximately 22 mill. m2 collection sur-
faces and 9.1 mill. storage wells have been constructed, supply-
ing 1.31 mill. rural residents and 1.2 mill. livestock adequate sup-
plies of drinking water. At the same time, some 2.1 mill. mu
(about 140␣ 000 ha) of rainfed lands have been irrigated using
rainwater harvesting (43). The positive experience gained from
this program so far encouraged the Gansu provincial government
to project expansion of the program in 1998. The new program
covered the whole semiarid area of the province, the goal being
to help each rural household in the project areas irrigate 3 mu
(0.2 ha) of cropland using the rainwater harvesting system over
a 5-year period. The program has already made some progress.
The success of RHA in Gansu province has had a profound
impact on other provinces in northwest China. Over the last few
years, RHA has been introduced in neighboring provinces such
as Ningxia, Shanxi, Shannxi, and Inner Mongolia, as well as in
the central parts of China such as Hebei and Henan provinces
(43). In these provinces, some programs similar to the 1-2-1 sys-
tem in Gansu province have been launched by local govern-
ments, leading to the rapid spread of RHA over a wider range
of arid, semiarid and drought-prone sub-humid regions. A grow-
ing number of farmers have benefitted from the use of this tech-
nology. Currently, 4 basic types of agricultural production based
on the inclusion of rainwater harvesting systems are being
practiced by household farmers in China’s semiarid areas: (i)
semi-intensively managed field food crops (e.g. wheat, corn and
potatoes); (ii) semi-intensively managed field cash crops (e.g.
oilseed, sugar beet, tobacco and water melon); (iii) semi-inten-
sively managed fruit trees (e.g. pear, apple, peach and grape);
and (iv) intensively managed high market value crops grown in
greenhouses (e.g. vegetables, Chinese medicinal plants and flow-
ers).
CONCLUSIONS
Conventional rainfed farming in China’s semiarid areas is fac-
ing serious resource (especially water resource) and environmen-
tal (especially soil erosion) problems. The solution to these prob-
lems depends largely on the development of more effective wa-
ter-management practices. Research so far on RHA systems has
suggested that RHA is a rapid, effective and inexpensive ap-
proach to water management on rainfed land in semiarid areas
with unreliable rainfall. RHA has the potential to improve per-
formance in rainfed farming systems and to address environmen-
tal problems. Hence, from the perspective of sustainable devel-
opment in agriculture, developing and adopting RHA is essen-
tial to achieving a productive, sustainable rainfed agriculture in
China’s semiarid areas.
However, to be successful, the RHA system needs to be inte-
grated with a comprehensive agricultural management system.
This means that the management of RHA should be combined
with other agricultural technologies and management practices
including water and conservation measures, crop management,
and soil fertility management, as well as the selection of suit-
able crop varieties such as drought-tolerant crops and salinity-
tolerant crops. In addition, with the exception of the use of natu-
ral precipitation, other nonconventional water resources includ-
ing saline/brackish water and treated wastewater can be used as
potential water sources for agricultural purposes in semiarid ar-
eas in the future. Moreover, the spread of RHA in different re-
gions also requires consideration of a range of technological, bio-
483Ambio Vol. 29 No. 8, Dec. 2000 © Royal Swedish Academy of Sciences 2000
http://www.ambio.kva.se
References and Notes
1. Frasier, G.W. 1985. Technical economic and social considerations of water harvesting
and runoff farming. In: Arid Lands: Today and Tomorrow, International Research and
Development Conference. Whitehead, E.E. et al. (eds). Tucson, Arizona, USA, Octo-
ber 20–25, 1985, XIX+1435 pp.
2. Lavee, H., Poesen, H. and Yair, A. 1997. Evidence of high efficiency water-harvest-
ing by ancient farmers in the Negev Desert, Israel. J. Arid Environ. 35, 341–348.
3. Richards, K.S. 1972. Rainwater harvesting for domestic purposes. Rhodesia Agric. J.
Technol. Bull. 15, 45–51.
4. Boers, M.T. and Ben Asher, J. 1982. A review of rainwater harvesting. Agric. Water
Mgmt 5, 145–170.
5. Sheikh, M.I., Shan, B.H. and Aleem, A. 1984. Effect of rainwater harvesting methods
on the establishment of tree species. Forest Products J. 8, 257–264.
6. Boers, T.M., Zondervan, K. and Asher, B. 1986. Micro-catchment-water-harvesting for
arid zone development. Agric. Water Mgmt 12, 21–39.
7. Alcock, P.G. and Verster, E. 1987. Investigation into unconventional sources of water
for a periurban-rural district of Kwazulu South Africa. South Africa J. Sci. 83, 348–
352.
8. Cater, D.C. and Miller, S. 1991. Three years experience with an on-farm macro-
catchment water harvesting system in Botswana. Agric. Water Mgmt 19, 191–204.
9. Shah, B.H. 1992. Development of agroforestry model using water harvesting system
in semi-arid and arid zones. Pakistan J. Forestry 42, 190–199.
10. Kronen, M. 1994. Water harvesting and conservation techniques for smallholder crop
production systems. Soil Tillage Res. 32, 71–86.
11. Tabor, J.A. 1995. Improving crop in the Sahel by means of water-harvesting. J. Arid
Environ.30, 83–106.
12. Kaarakka, V. 1996. Management of bushland vegetation using rainwater harvesting in
eastern Kenya. Acta Forest. Fenn. 253, 2–93.
13. Rockström, J. and Valentic, C. 1997. Hillslope dynamics of on-farm generation of sur-
face water flows: The case of rainfed cultivation of pearl millet on sandy soil in the
Sahel. Agric. Water Mgmt 33, 183–210.
14. Agarwal, A. and Narain, S. 1997. Dying wisdom: The decline and revival of tradi-
tional water harvesting systems in India. The Ecologist 27, 112–116.
15. Zhao, S.L. 1996. An Introduction to Catchment Agriculture. Shannxi Sci. Technol.
Press, Xian. (In Chinese).
16. Zhu, X.M. 1989. Soils in the Loess Plateau and Agriculture. China Agric. Press, Beijing.
(In Chinese, summary in English).
17. Shan, L. and Cheng, G.L. 1993. Theory and Practice of Dryland Farming in the Loess
Plateau of China. China Sci. Press, Beijing. (In Chinese).
18. Wei, H., Zhao, S.L. and Wu, G.H. 1996. Meteorological aspects of rainwater harvest-
ing agriculture in the semi-arid loess plateau. J. Arid Land Resources Environ. 10, 64–
70. (In Chinese, summary in English).
19. Li, F.R. 1998. Studies on Arid Agricultural Ecosystems. Shannxi Sci. Technol. Press,
Xian. (In Chinese, summary in English).
20. Li, F.R., Zhao, S.L. and Geballe, G.T. 2000. Water use patterns and agronomic per-
formance for some cropping systems with and without fallow crops in a semi-arid en-
vironment of northwest China. Agric. Ecosyst. Environ. 79, 129–142.
21. Lu, Z.F. and Zhao, G.S. 1991. Soil-water conserving farming and its yield-increasing
technical systems. China. J. Soil Water Cons. 5, 166–171. (In Chinese, summary in
English).
22. Shan, L. 1998. Technological measures for sustainable dryland farming in the semi-
arid regions of China. China Science Daily (Overseas Edition), Zhongguo Kexuebao
Publishing House, January 25. (In Chinese).
23. Yang, Q.Y. 1990. The characteristics of water deficits for certain grain crops in the
loess plateau and the problems of rainfed agriculture. China J. Natural Resources 1,
24–31. (In Chinese, summary in English).
24. Zhao, S.L., Wang, J. and Li, F.M. 1995. Limitations of soil-water conserving farming
in the semi-arid loess plateau. Acta Bot. Boreal. Occident Sin. 15, 1–7. (In Chinese,
summary in English).
25. Wu, F.X. 1995. Engineering design of rainwater harvesting systems. Gansu Water
Conserv. Water-Electricity Technol. 4, 34–39. (In Chinese).
26. Zhang, Z.X., Gong, S.H. and Wang, X. L. (eds). 1999. Rainwater Harvesting Engi-
neering Technologies. China Water Conserv. Water-Electricity Press, Beijing, pp. 50–
66.
27. The research term by the Gansu Provincial Water Conservancy Science Institute 1990.
Field Studies on the use of rainwater in arid and semi-arid areas of Gansu province.
Gansu Water Conserv. Water-Electricity Technol. (special issue). (In Chinese).
28. Pei, B.X. 1989. Measurement and Calculation of Evaporation and Evapotranspiration.
Meterol. Publishing House, Beijing, pp. 598–638. (In Chinese).
29. Li, F.M., Zhao, S.L., Duan, S.S., Gao, S.M. and Feng, B. 1995. Preliminary study on
limited irrigation for spring wheat field in semi-arid loess region of China. China J.
Appl. Ecol. 6, 259–264. (In Chinese, summary in English).
30. Gao, S.M. and Zhu, R.S. 1996. Effects of supplemental irrigation and plastic film mulch-
ing on yields of rainfed crops. J. Arid Land Resources Environ. 2, 42–48. (In
Chinese,summary in English).
31. Ren, Z. X., Bi, J.T. and Gu, Y.T. 1998. Research and application of well-based rain-
water harvesting systems in arid mountain area of Ningxia province. J. Arid Land Re-
sources Environ. 12, 58–63. (In Chinese, summary in English).
32. Robinson, A.R. 1981. Erosion and sediment control in China’s Yellow River basin. J.
Soil Water Cons. 36, 125–127.
33. Lee, H. 1984. Soil conservation in China’s loess plateau. J. Soil Water Cons. 39, 306–
307.
34. Wen, D.Z. 1993. Soil erosion and conservation in China. In: World Soil Erosion and
Conservation. Pimentel, D. (ed.). Cambridge University Press, Cambridge, pp. 63–85.
35. Jing, K. 1988. A study on the relationship between soil erosion and the geographical
environment in the Middle Yellow River basin. J.Arid Land Resources Environ. 1, 289–
299. (In Chinese, summary in English).
36. Zhang, Z.H. 1992. Grassland agriculture and its technical systems. In: Proc. Int. Con.
on Agroecosystems in the Loess Plateau. Ren, J.Z. (ed.).Gansu Sci. Technol. Press,
Lanzhou, pp. 37–42. (In Chinese, summary in English).
37. Ren, J.Z. 1992. Ecological productivity of grassland farming system on the Loess Pla-
teau of China. In: Ren, J.Z. (ed.). Proc. Int. Conf. on Agroecosystems in the Loess Pla-
teau. Gansu Sci. Technol. Press, Lanzhou, pp. 37-42. (In Chinese, summary in Eng-
lish).
38. Pimentel, D. (ed.) 1992. World Soil Erosion and Conservation. Cambridge University
Press.
39. The Gansu Provincial Commission of Science and Technology 1994. A Guideline of
the Gansu Agrotechnological Policies. Gansu Sci. Technol. Press, Lanzhou, pp. 3–9.
(In Chinese).
40. Li, F.R. and Li, S.J. 1998. A choice of strategy for sustainable development of rainfed
farming in water-limiting areas of northwest China. In: The 1st Annual Symposium of
the National Soft Science Research in China. China Sci. Technol. Press, Beijing, pp.
876–886. (In Chinese).
41. Li, S.K. 1992. On Harnessing Sloping Lands in Eroded Areas in the Middle Reaches
of the Yellow River. Gansu Sci.Technol. Press, Lanzhou. (In Chinese).
42. Wei, H.L., Bai, J.M. and Yang, X.T. 1998. Analysis of benefits of water-harvesting
farming in semi-arid loess areas. J. Arid Land Resources Environ. 12, 41–47. (In Chi-
nese, summary in English).
43. Yang, A.M. 1998. Practice of Rainwater Harvesting Systems in Gansu province. In:
The Challenges and Strategies for Development of Rainfed Agriculture in Semi-Arid
Regions. Xu, J. (ed.). Lanzhou University Press, Lanzhou, pp. 308–312. (In Chinese).
44. Giampietro, M. 1997. Exploring the link between technology, natural resources and
socioeconomic structure of human society: A theoretical model. In: Adv. Human Ecol.
Vol. 6. Freese, L. (ed.). JAI Press, Greenwich CT, pp. 73–128.
45. This research was supported by a special President’s fund from the Chinese Academy
of Sciences and the visiting scholar foundation from the State Key Laboratory of Arid
Agroecology in Lanzhou University.
46. First submitted 20 Aug. 1999. Accepted for publication after revision 11 April 2000.
Li Fengrui is professor in agroecology at the Department of
Ecology and Agriculture in the Cold and Arid Regions
Environmental and Engineering Research Institute,
Academia Sinica. His address: Engineering Research
Institute Academia Sinica, 260 Donggang West Road,
Lanzhou 730000, PR China.
E-mail: frli@fm365.com
Seth Cook is a PhD student in environmental studies at the
School of Forestry and Environmental Studies in Yale
University.
E-mail: seth.cook@yale.edu
Gordon T. Geballe is professor and Associate Dean of the
School of Forestry and Environmental Studies in Yale
University.
E-mail: gordon.geballe@yale.edu
William R. Burch, Jr. is Hixon Professor in natural resource
management at the School of Forestry and Environmental
Studies in Yale University.
E-mail: william.burch@yale.edu
Their address: Yale School of Forestry and Environmental
Studies, 205 Prospect Street, New Haven, CT 06511, USA.
physical, geohydrological, ecological, social, cultural, economic,
and political factors, especially the presence of tremendous het-
erogeneity among different regions. On the whole, the success
of RHA is likely to hinge upon: (i) whether research scientists
and extension agents can provide long-term training and tech-
nical services to local farmers, thereby, increasing their ability
to apply the technology and to ensure adequate maintenance of
rainwater harvesting facilities; (ii) the need to develop a wide
variety of rainwater harvesting technologies that are ecologically
sound and economically viable as alternatives for farmers; (iii)
the need to design and develop alternative policy instruments and
implementing strategies that facilitate the spread of RHA. It has
long been recognized that there are interactive forward and back-
ward linkages between technology, infrastructure and institu-
tional structures (44). In many cases, technological innovations
in the agricultural sector often fail to be widely applied, largely
due to the organizational and/or institutional incompatibility
rather than technological complexity. When the use of RHA is
expanded from a household level to a village level and to a wa-
tershed scale, it not only requires technological development but
also entails parallel development of social institutions and man-
agement systems in order to ensure an equitable allocation and
efficient use of water, soil, and other resources.
As with any agricultural technology, the effects of RHA are
likely to be complex and ambiguous. Therefore, further research
is needed to assess the ecological, social and economic impacts
of RHA from a plurality of perspectives in a wide range of dif-
ferent localities. This study can contribute to a better understand-
ing of the complex RHA system, which may help in the discus-
sion of trade-offs related to sustainability of this agrosystem. The
study may also help in applying RHA more effectively and in
mitigating some of the potential problems.
