Você está na página 1de 15

Landscape Ecol (2006) 21:13111325 DOI 10.

1007/s10980-006-0023-x

RESEARCH ARTICLE

Agricultural land-use patterns and soil erosion vulnerability of watershed units in Vietnams northern highlands
Karine Vezina Ferdinand Bonn Cu Pham Van

Received: 3 May 2005 / Accepted: 21 February 2006

Springer Science+Business Media B.V. 2006 Abstract Since the mid eighties, agricultural development and increased population growth in Vietnams northern highlands have modied land use patterns and thus, increased the runoff process and soil degradation induced by water erosion. In the last decade, Vietnamese literature has focused on the computation of soil losses over large areas. Most of these spatial and quantitative soil erosion studies do not consider the impact of agricultural land use diversity (spatial heterogeneity), particularly at the watershed scale, and the annual variability of seasonal landscape factors on soil erosion vulnerability and hence, landscape dynamics. We present an integrated approach combining eld measurements and observations, GIS and modeling to determine the spatial and temporal dynamics of soil erosion vulnerability according to watershed units and hence, the impact of physical environment components and agricultural land use patterns on landscape evolution. Tables and graphics showing the cropping systems, the periods within a year, and the watershed units that are most vulnerable are presented. The double cultivation cycles for paddy rice elds not only imply two periods of land preparation and establishment that expose the soil surface to raindrop impacts, but also increased soil management practices that decrease the soils resistance to detachment. Despite the low levels of soil management practices for the shifting cultivation system, the near absence of soil conservation practices clearly increases their vulnerability. Hence, rainfed cropping systems, mainly soya and cassava, cultivated on sloping lands (hills and mountains) where soil erosion vulnerability is the highest represent the watershed units which are the most prone to soil loss. Keywords Tropical environment Watershed Erosion dynamics Scaling in space and time GIS

K. Vezina (&) F. Bonn Remote Sensing Applications and Research Center (CARTEL), Sherbrooke University, 2500 University blvd., J1K 2R1, Quebec, Canada e-mail: karine.r.vezina@usherbrooke.ca Cu Pham Van Center For Remote Sensing and Geomatics (VTGEO) Institute of Geology, Vietnamese Academy of Science and Technology, 340 Bach Dang Chuong Duong, Hoan Kiem, Hanoi, Vietnam

Introduction Driven primarily by socioeconomic forces, land use and land cover predominantly determine the structure, functioning, and dynamics of most landscapes throughout the world and represent one of the most important and challenging research areas in landscape ecology, and indeed in global ecology (Wu and Hobbs 2002). Since the mid eighties, agricultural development

123

1312

Landscape Ecol (2006) 21:13111325

and increased population growth in Vietnams northern highlands have modied land use patterns and thus, increased the runoff process and soil degradation induced by water erosion. Steep slopes combined with intense rainfall concentrated 80% during the rainy season and high soil erodibility make upland elds prone to erosion (Nguyen and Le 1990). In 1999, soil losses by water erosion (runoff erosion) on upland areas were estimated at about 3.022 billion t yr1 for the whole 22 million ha area that represents approximately 70% of the total country area (Tran and Le Thai 1999). At watershed scale, soil water erosion generates important environmental and socio-economic consequences such as: decrease in soil fertility and productivity, transformation of land into fallow land not suitable for reforestation, irreversible reduction in arable soil, increase in ooding events, diffuse pollution of river networks, etc. The advent, since the 1980s, of Earth observation image processing tools, digital topographic data and geographical information systems (GIS) offering modeling capabilities have fostered the extrapolation of soil loss models over larger areas such as watersheds. The GIS approach is an effective tool to put site-specic models into a spatial context to examine important processes, identify important locations, and accommodate scenario analyses. Hence, in the last decade, much of the Vietnamese literature has focused on the integration of the physical and human landscape factors leading to accelerated soil losses into soil erosion models in order to determine soil erosion vulnerability and to create land use scenarios in the uplands (Nguyen 1992). In most of these soil erosion studies, GIS are combined with empirical models, and in particular, the Universal Soil Loss Equation (USLE), to quantify and map annual soil losses and erosion potential over large areas (Lai Vinh 2000; Thai et al. 2001). Parameters required for the erosion model are generated using data provided from different sources such as remote sensing data, a digital elevation model, cartographic data, meteorological data and ground surveys. The USLE and its revised version RUSLE (Renard et al. 1997) have been used in many other tropical countries to predict long term soil losses for different land uses (cultivation, dense forest, shrub grassland, etc.) due to torrential erosion, runoff and underground erosion (Jain and Kothyari

2000; Mati et al. 2000; Millward and Mersey 2000; Boggs et al. 2001; Jain et al. 2001; Cohen et al. 2005). Since the USLE and RUSLE represent the major factors affecting water erosion (Hudson 1971), transferring them to locations throughout the world requires only the determination of appropriate values for the different factors (Renard and Freimund 1994). However, most of the spatial scale studies found in the literature do not represent agricultural land use diversity (spatial heterogeneity), particularly at the watershed scale, and its impact on soil water erosion landscape vulnerability. Since the USLE and the RUSLE soil erosion models compute soil losses on an annual basis (time scale), the results do not consider the annual variability of seasonal landscape factors. We present here an integrated approach combining eld measurements, GIS and modeling to determine the spatial and temporal dynamics of soil water erosion vulnerability according to watershed units. The small size of the watershed under study allowed a very detailed assessment of the agricultural land use heterogeneity and hence, soil erosion parameters that lead to increased landscape unit soil water erosion vulnerability. In order to apply the USLE to tropical bioclimatic conditions and farming practices, the equations factors were reassessed. The monthly variability of the vegetation cover and rainfall erosivity was taken into account in the modeling to determine the temporal dynamics of the soil erosion vulnerability for each cropping system. Although USLE factors have been used by several authors to quantify soil losses and erosion potential, our approach focuses on establishing a relative scale of soil erosion vulnerability since we do not control all the factors for quantifying the phenomena. Quantitative results were considered as ordinal values rather than absolute in order to eliminate the false accuracy impression provided by specic values of soil erosion potential (Bonn 1998).

Methods Study area ` Ba Be National Park is situated 256 Km North of Ha noi in Northern Vietnam (Bac Kan Province, Ba Be district and Dong Phuc commune). Dened as a

123

Landscape Ecol (2006) 21:13111325

1313

natural reserve since 1977, Ba Be National Park became in 1992 the 8th national park in the country. With a total area of 7,610 ha, Ba Be National Park consists of limestone mountains, rivers and lakes that are endowed with a rich tropical fauna (222 species) and ora (354 species) (Le Trong et al. 2001). The Park is surrounded by a chain of calcareous mountains with three majors peaks: Pia Hoac at 1520 m, Hoa Son at 1517 m, Pu Sam Sao at 1175 m and Pu Moc at 1048 m. Ba Be Lake situated in the middle of the Park in a karstic depression surrounded by cliffs is considered as the most beautiful natural lake in Vietnam. With a total area of 500 ha, Ba Be Lake plays an important role for the local population since it insures the regulation of the regions water regime and the annual ooding of the agricultural elds, improves microclimatic conditions and contributes to the conservation of the aquatic ecosystem. The three main ` ` ` ` afuents of the lake, the Cho Leng, Bo Lu and Ta Han rivers, drain an important portion of the runoff waters from the regions watersheds. According to the Park director (personal communication), the measurements and observations that have been made over the last 20 years show an increase in the sediment rate and hence of erosion within the watersheds. The study was conducted within the 10 km2 watershed of the Khuoi Lung Phet river situated in the surrounding area (buffer zone) of the Ba Be National Park where intensive farming practices and shifting ` ` cultivation from ethnic groups (Hmong, Dao, Tay,

and Kinh) have lead to a conicting situation between production and protection that has proved to be very alarming since conservation policies are still unformulated (Zingerli et al. 2002) (Fig. 1). The topography of the watershed is characterized by steep mountain limestone ridges Pia Hoac, interspersed by non-limestone areas of more undulating topography. Elevation in the watershed ranges from 190 m to 1450 m and declivity from 0 to 65. Alluvial plains and high terraces represent low land areas. Alluvial areas are characterized by an altitude below 220 m and a declivity ranging from 0 to 10. High terraces occupy altitudes between 220 m and 300 m and slopes between 10 and 20. Hills and mountains, which represent sloping land areas, are characterized by slope steepness ranging from 20 to 65 and altitudes respectively between 300 m and 600 m, and 600 m and 1450 m. The extensive land exploitation allocated to the peasants in increasingly large numbers has completely transformed the landscape of the region. The area supports a mosaic of annual agricultural land-use types following the watershed topography and ethnic group settlements: shifting cultivation (cassava, corn and soya) on sloping land where forest has been cleared and paddy rice elds in the valleys (low land areas). The area experiences a mountainous humid tropical climate with a warm and humid season from April to September and a cold and dry season from October to March. Mean annual rainfall is about 1500 mm.

Fig. 1 Study area in Northern Vietnam

123

1314

Landscape Ecol (2006) 21:13111325

Soil erosion model An environmental characterization of the physical (climate, pedology, topography) and human factors (agricultural and conservation practices) governing water erosion at the watershed unit scale was made according to the Universal Soil Loss Equation (USLE). Created in the United States, the USLE is an erosion model designed to compute long-term average annual soil loss at the eld scale (A) as the product of six major factors: rainfall erosivity (R), soil erodibility (K), slope length (L) and steepness (S), cover and management practices (C), and conservation practices (P) (Wischmeier and Smith 1981). A R K LS C P 1

Soil erodibility (K) Values for the exact soil erodibility at a specic site can be computed by using the soil erodibility nomograph based on the percentage of silt and very ne sand, percentage of OM, soil-structure code, and prole-permeability class. However, the equations soil erodibility nomograph commonly gives unrealistic values for tropical soils (Mulengera and Payton 1999). Moreover, its use requires an important quantity of data concerning soil physical properties which, in our case, are not available and/or couldnt be measured on the eld because of nancial and technical issues. We used an equation developed in the framework of a research program initiated to identify a suitable soil loss equation for use under tropical conditions, in general, and particularly Tanzanian conditions (Mulengara and Payton 1999). The following equation developed by these authors obtained the best correlation coefcient (r=0.911) between calculated values and measured values in Tanzania and hence, it was retained for our study. This equation is based on texture-derived parameters and soil permeability that were found to be quite adequate for estimating the erodibility factor (K) in the tropics. K 1:82247 105 M 0:0045Pe 0:0097 3

Since the parameters K, C, and P have values ranging between 0 and 1, if all equal 1 then soil loss (A) is equal to the product of rainfall erosivity (R) and slope length and steepness (LS). The soil erosion potential of the landscape can be computed as the product of R, K and LS. Environmental characterization Rainfall erosivity (R) The rainfall erosivity (R) represents the energy available for detachment and transport by rainsplash and is a function of kinetic energy (KE) and rainfall intensity (I) (mm h1). Since rainfall intensity represents the principal factor of kinetic energy, the following equation (Carson and Kirkby 1972) was used to calculate the monthly rainfall erosivity as the average rainfall intensity per day of rain (ro) for a given month (i) (mm day1), according to monthly rainfall amount (p) and number of days of rain per month (N). roi pi=N i 2

Where: K=soil erodibility in t ha h/(ha MJ mm); M=(si+vfs) (si+vfs+sa); si: silt % (0.050.002 mm), sa: sand % (0.20.10 mm), vfs: very ne sand % (0.100.05 mm) Pe permeability classes 1 Rapid [127 mm h1 2 Moderate to rapid 63:5 127 mm h1 3 Moderate 20 to 63:5 mm h1 4 Slow to moderate 5 20 mm h1 5 Slow 1 5 mm h1 6 Very slow \1 mm h1 Spatial and qualitative data for the different soil associations (FAO classication) found in the watershed were generated using the database

Since the vegetation cover protects the soil from raindrop impacts, the inuence of rainfall erosivity on soil erosion vulnerability will depend on the annual variation of the canopy cover for each cropping system.

123

Landscape Ecol (2006) 21:13111325

1315

(at 1: 50,000 scale) provided by the SAM (Mountainous Agrarian System) project members. Since we did not have any soil samples, the percentages of sa, vfs and si for the different soil types were estimated using typical tropical soil proles presented by the FAO (1998) mostly, and in Kalpage (1974). The soil textures were determined according to the soil particle size distribution within the triangle of texture. The permeability classes were estimated from the texture classes that determine the magnitude and the distribution of pore space (Wall et al. 2002). Even if the equation developed by Mulengara and Payton (1999) does not give actual values that can only be obtained from ground-measured samples, it is useful for estimating soil erodibility variability for the different soil types found in the watershed (Table 1). Topography (LS) The LS factor assesses the effect of slope steepness and slope length on soil erosion at the eld scale. The original equation for the calculation of the LS factor was created from data obtained on agricultural elds with slope steepnesses ranging from 3% to 18% and slope lengths from 10 m to 30 m. This ranging does not correspond to the topography of the study area. To avoid mistakes, a review of literature was made to nd a substitute equation better adapted to the studied environment. The approach used replaces the slope length factor (L) by the upslope contributing area (A) in order to incorporate the impact of ow convergence on increased erosion, an important factor in mountainous environments with steep slopes (Mitasova et al. 1996, 1998). LSr m 1Ar=a0 m sin br=b0 n 4

Where: A: upslope drainage contributing area per unit contour width, b: slope steepness (), m and n : empirical parameters of 0.6 and 1.3 for affected regions susceptible to rill and gully erosion a0=22.1 m slope length and b0=5.16 slope, a standard USLE plot, r =hillslope segment (x, y) The Digital Elevation Model (DEM) was created in the Arc View 3.2 GIS software by the interpolation (Kriging method) of digital contour lines with a contour interval of 20 m. The contour lines were extracted from the database provided by the PARC project members. Its precision accuracy has been evaluated with ground truthing samples realized with a GPS during eld campaigns. The slope map (S factor) was created in the GIS from the DEM. The main topographic units (alluvial plains, high terraces, hills and mountains) of the watershed were dened by the superposition of the DEM and the slope map. The ow concentration grid (ow accumulation) was created by applying the following script in Map Calculator. ([elevation].FlowDirection(FALSE)): FlowAccumulation(NIL) 5

The LS factor was computed for each grid cell of the DEM which represents a hillslope segment (r) from the ow concentration grid (owacc) and the slope map (slope) (Figure 2). In the script used, the value for resolution corresponds to the grid cell size of the DEM (20 m). (([flowacc]* resolution /22.1). Pow(0.6))* (((([slope] *0.01745).Sin)/ 0.09).Pow(1.3))* 1.6 6

Crop and soil management (C)


Table 1 Soil erodibility (K) values and ranking Soil type Erodibility (t ha h ha1 MJ1 mm1) 0.055 0.025 0.028 0.050 0.045 0.065

Fluvisols Regosols Leptosols Cambisols Alisols Phaozems

The C factor denes the relative effectiveness of a cropping system in terms of soil loss prevention according to a particular combination of cover, crop sequence and management practices and also on the growth stage and development of the canopy cover at the time of erosive rainfall. Since value tables developed in the USA for estimating the C factor are not applicable to agricultural and climatic conditions in the tropics (Mulengara and Payton 1999),

123

1316

Landscape Ecol (2006) 21:13111325

Fig. 2 LS factor values

new values were calculated by the combination of the methods presented by Wischmeier and Smith (1981), Morgan (1995), and Stone and Hilborn (2000). Six cropping systems were identied through eld observations and interviews on the watershed. On alluvial plains, because irrigation water is taken directly from the main river and hence available all year around, farmers can do two cultivation cycles of rice despite the dry season. On the other hand, irrigation water for paddy rice elds on terraces comes from rainfall and runoff water coming from the mountains and is transported into irrigation channels running along the foot of upslope terrace risers. Thus, rice can only be cultivated during the rainy season as rainfall is insufcient for a second cultivation cycle during the dry season. Nevertheless, a rotating crop such as corn can be cultivated during the dry season since it does not need as much water to grow. Since there is no irrigation setting for shifting cultivation systems (called rainfed cropping systems), elds are irrigated by rainfall. Thus, only one cultivation cycle per year is possible (Figure 3). The cropping system annual calendars are divided into a series of crop stages (land preparation, sowing, canopy growth and development, harvest, and fallow) subdivided into periods (months of year) dened so that cover and management effects may be considered approximately uniform within each period. Since multitemporal satellite imagery (without clouds) were unavailable and multitemporal canopy

Fig. 3 Cropping systems according to watershed toposequence in the study area

cover eld measurements impossible to realize because of nancial and time restrictions, the canopy cover (%) for each period was estimated using substitute methods found in the literature as follows: Land preparation During land preparation, canopy cover was estimated at 0% since the soil surface is completely bare either by residual crop burning (shifting cultivation crops) or burying (paddy rice elds with or without corn as rotational crop) during ploughing operations. Sowing During rice seedling sowing under water, an estimate of ground cover was derived from a vertical downward photography taken with a digital camera at a height of 2 m above the sampling plot so that its coverage was about 1 m2. ENVI software was used to classify the digital photographs into vegetation

123

Landscape Ecol (2006) 21:13111325

1317

present and vegetation absent classes using a maximum likelihood classication. The ratio of vegetation present pixels to the total number of pixels provided an estimate of fractional ground cover. Ground cover was considered nil (0%) during grain crop (soya and corn) and tuber (cassava) sowing. Canopy growth and development Between the sowing and harvest stages, the ground cover variation was assessed according to the average speed of growth and development of each canopy crop estimated from tropical studies (Elwell and Wendelaar 1977; Elwell 1978; Mati 1995; Dijk and Bruijnzeel 2001; Vaesen et al. 2001). The Leaf Area Index (LAI), dened as the one-sided proportion of leaf area per unit ground area, was used as ground cover indicator (Vaesen et al. 2001). During the period of plant maturity, the nal crop canopy covers were estimated from the average of maximum ground cover obtained in the above studies. These averages were adjusted to the crop plant spacing in the study area. Harvest and fallow Ground cover was estimated according to the different harvest techniques (picking (corn), cutting (rice) or uprooting (soya and cassava)) and crop residue management (left on the soil surface, buried or removed), and the speed of growth of adventices. The classication in ENVI of the downward eld (cassava and rice) photography in fallow was used to estimate the canopy cover following the harvest and growth speed of adventices, which govern the ground cover rate. Rainfall erosivity distribution The erosion control effectiveness of a cropping system depends, in part, on how the years erosive rainfall is distributed among the

crop stage periods of each crop included in the system (Wischmeier and Smith 1981). Assuming that erosivity is directly related to the precipitation amount, the percentages of rainfall total amount (% R value) within each period were used to compute the R factor distribution throughout the year (Morgan 1995).

Adjusted C factor values From canopy cover estimates, a rst C factor was determined for each period of the year according to the inversely proportional linear relationship between the C factor and the ground cover (1% of ground cover). The summation of the product of the C factor and adjustment factor (% R) for each period allows the computation of the C factor adjusted to annual distribution of ground cover and rainfall (Morgan 1995). Since the vegetation cover protects the soil from raindrop impacts, a high value of the adjusted C factor indicates a long period within annual cultivation cycles of low ground cover during heavy rain periods. In the assessment of the C adjusted factor values, the most important thing wasnt to determine precisely the percentage of ground cover but to know exactly the state of the canopy cover (crop stage) at the time of erosive rainfalls and to keep a relative ranking between the different canopy covers according to vegetation type and plant spacing. The methods used to estimate the canopy cover were then appropriate to obtain relative results (Table 2).

Soil management factor A soil management factor must be added to the adjusted C values obtained (Stone and Hilborn 2000). An inventory of the nature and frequency of soil management practices (ploughing, sowing, weeding, etc.) that affect the rate of soil ero-

Table 2 Calculation of the C factor adjusted to rainfall and canopy cover annual distribution: example of the soya cropping system Crop stage Fallow Land preparation Sowing Canopy growth and development Periods (months) FebMarchAprilMay June (2nd weeks) June (last 2 weeks) July August Sept Oct Nov (2nd weeks) Nov (last 2 weeks)-DecJan Canopy cover (%) 0.45 0.00 0.00 0.05 0.40 0.60 0.70 0.00 0.15 C factor 0.55 1.00 1.00 0.95 0.60 0.40 0.30 1.00 0.85 Total Rainfall (% R value) 0.25 0.07 0.07 0.21 0.21 0.10 0.05 0.01 0.03 1.00 Adjusted C factor 0.14 0.07 0.07 0.20 0.13 0.04 0.02 0.01 0.03 0.69

Harvest Fallow

123

1318

Landscape Ecol (2006) 21:13111325 Table 4 C factor values Cropping system Adjusted C factor 0.69 0.67 0.69 0.54 0.60 0.69 Management factor 0.80 0.60 0.80 0.40 0.20 0.40 Final C factor 0.55 0.40 0.55 0.22 0.12 0.28

sion (Wischmeier and Smith 1981; Turkelboom et al.1996) was carried out. A level of soil management ranging from 1 to 4 was determined according to the relative topsoil disturbance associated with the management practices. From these results, a management factor (values ranging between 0.2 and 0.8) was estimated for each cropping system (Table 3).

Final factor C values The product of the adjusted C factor and management factor allowed the calculation of the nal C factor values for each cropping system (Stone and Hilborn 2000). Factor C is given in terms of its average annual value for a particular combination of cropping system, management, and rainfall pattern. In the worst case, the C factor value is equal to 1.0. Since the vegetation cover protects the soil from raindrop impacts, a high value of the adjusted C factor indicates a long period within annual cultivation cycles of low ground cover during heavy rain periods (Table 4). Conservation practices (P) The P factor represents the peoples participation of in soil protection for each cropping system. It takes
Table 3 Calculation of the soil management factor Cropping system Soil management (activities and tools) 4 tilling and 4 harrowing (plough and buffalos) 2 sowing (by hand) 2 weeding (rake) 2 tilling and 2 harrowing (plough and buffalos) 1 sowing (by hand) 1 weeding (rake) 4 tilling and 4 harrowing (plough and buffalos) 2 sowing (by hand) 2 weeding (hoe) 1 sowing (shovel or hoe) 1 weeding (hoe) 1 uprooting (machete) 1 slight tilling (rake) 1 sowing (branch) 1 weeding (hoe) 1 slight tilling (rake) 1 sowing (by hand) 1 weeding (rake) 1 uprooting (machete) Level

Paddy rice (2 cycles) Paddy rice (1 cycle) Paddy rice (1 cycle) with corn Cassava Corn Soya

Paddy rice (2 cycles)

Paddy rice (1 cycle)

Paddy rice (1 cycle) with corn

Cassava

Corn

Soya

into account practices that reduce water runoff speed and thus, the quantity of soil particles transported. However, these practices found in the literature (ex: contouring, contour strip-cropping, terracing) are not representative of those found in the study area since most of them present technical, social, and economic constraints. To avoid leaving the P factor to its basic values as found in many studies (1 for a cropping system using none of the above erosion-control practices and 0 for a complete erosion-control practice), we adapted the factor to the watershed conditions. We considered not only practices that reduce surface runoff water, but also those that enhance soil fertility (Roose 1999). For soil conservation fertility measures, we considered organic matter (OM) and fertilizer inputs. Green manure, which represents the turned over and the burying of crop residues (rice and corn), and animal manure help to maintain soil fertility and structure and thus, enhance soil resistance to rainfall impact. Natural (burning ash) and chemical fertilizers add nutritive elements to the soil (nitrates and phosphates) and hence, contribute in reducing the speed of soil degradation. For runoff reducing measures, we considered mechanical and agronomic methods. Mechanical methods such as dykes and terraces depend upon manipulating the surface topography to control the ow of water and the transport of soil particles to the bottom of the slope. On the alluvium plains, dykes are wide and in straight line while on high terraces they are narrower and run along contour lines. Terraces are constructed across the slopes and parallel to the contour lines. Agronomic measures utilize the role of vegetation (hedges) to stabilize and to drain excess water (Morgan 1995) along the terrace levels. According to the number of soil conservation practices, P factor values ranging from 0.1 to 0.9 were

123

Landscape Ecol (2006) 21:13111325

1319

attributed to each cropping system; in the worst case, the P factor value is equal to 1.0 (Table 5). Watershed units The integration of the spatial data (K and LS digital maps and C and P eld observations) into the GIS allowed the association of the cropping systems (C and P factors) and their geographic units (K and LS factors). Since only the factors K and LS could be mapped, the watershed units are presented by a gure representing the possible spatial combination of the different factors. Each watershed unit is characterized by physical and human factors having a similar hydrologic response and hence, a unique soil erosion vulnerability (Figure 4). According to the watershed units (Figure 4), Fluvisols found on alluvial plains are favorable for the cultivation of paddy rice (2 cycles) because of their high water retention capacity. On high terraces, when rainfall exceeds 750 mm per year, the entire prole of the Regosols is raised to its low water holding capacity early in the wet season (Williams and Joseph 1970) and hence the cultivation of paddy rice (1 cycle) is possible. Many Regosols form a hard surface crust early in the dry season. Since the crust hinders the emergence of seedlings and inltration of rain and irrigation water, most of the terraces are left in fallow during the dry season or are, in a few cases, cultivated with a dry crop such as corn. Cambisols are good agricultural lands that are intensively used for a wide range of crops from irrigated crops (ex: paddy rice 1 cycle on high terraces) to rainfed crops (ex: corn on hills and mountains). Since Leptosols have excessive internal drainage that makes them sensitive to
Table 5 P factor values for each cropping system Cropping system Soil fertility conservation Organic matter Fertilizers

drought, rainfed crops (ex: soya, cassava and corn) that do not need any irrigation system other than rainfall are cultivated on them. Alisols are characterized by high acidity, surface instability, and poor natural fertility, which all represent limitations to their cultivation. Thus, only rainfed crops are cultivated on Alisols. Phaozems, which are generally porous, relatively rich in nutrients, and structurally stable are cultivated with corn, cassava, and soya.

Results Spatial dynamics of watershed unit soil erosion vulnerability Alluvial plains represent around 31% of the watershed area. Although they are occupied by high erodibility soils such as Fluvisols characterized by a high clay content, their low topographic LS values contribute signicantly in reducing the soil surface erosion vulnerability, which is the lowest in the watershed. High terraces representing about 16% of the watershed area also have low soil erosion vulnerability since they are characterized by moderate topographic LS values and occupied by low erodibility soils such as Regosols and Cambisols. Hills and mountains, representing about 53% of the watershed area, have, in certain cases, high soil erosion vulnerability. Topographic LS values on hills and mountains are the highest found in the watershed (if we dont take into account the river network, which has values of 101 and more since the ow accumulation and hence, soil water erosion vulnerability are maximal) and can be associated with high erodibility

Runoff reducing Mechanical methods Agronomic methods Factors P

Green manure Animal manure Naturals Chemicals Terraces Dykes Hedge Paddy rice (2 cycles) Paddy rice (1 cycle) Corn with paddy rice (1 cycle) Cassava Corn Soya X X X X X X X X X X X X X X X X X X X X 0.1 0.2 0.1 0.9 0.8 0.9

123

1320

Landscape Ecol (2006) 21:13111325

Fig. 4 Watershed units

soils such as Phaozems. Hence, approximately half of the watershed area is physically prone to important soil erosion vulnerability. If these vulnerable areas are associated with inappropriate agricultural practices, then soil losses can signicantly increase. The product of the C and P values give the soil erosion vulnerability for each cropping system (Table 6) which, combined to the watershed intrinsic vulnerability (K*LS), can provide an idea of the spatial dynamics of the watershed unit soil erosion vulnerability (A). The two cultivation cycles for paddy rice (2 cycles) and paddy rice with or without corn as rotational crop not only imply two periods of total soil surface exposure to rainfall erosivity, but more soil management practices that decrease the soils resistance to detachment by raindrop impacts or owing water. The low levels of soil management practices for rainfed cropping systems gives them a lower nal C factor.
Table 6 Soil erosion vulnerability of cropping systems Cropping system Paddy rice (2 cycles) Paddy rice (1 cycle) Paddy rice with corn Cassava Corn Soya C 0.55 0.40 0.55 0.22 0.12 0.28 P 0.10 0.20 0.10 0.90 0.80 0.90 C*P 0.06 0.08 0.06 0.20 0.10 0.25

Nevertheless, paddy rice cropping systems (1 and 2 cycles) have the highest number of soil protection measures, hence low P values (0.1 and 0.2), since rice represents the rst food crop for the households in the study area. Shifting cultivation systems have high P values (0.9 and 0.8) since no other soil protection measures than the use of crop residues as green manure (corn) and/or natural fertilizer (soya and cassava) is implemented. Burning eld activities in shifting cultivation contribute to the recycling of mineral richness contained in the vegetation and hence, represent an essential step for the success of rainfed crops. Despite the low levels of soil management practices for shifting cultivation systems that give them a relatively low value of C, the near absence of soil conservation practices clearly increases their vulnerability to soil erosion, which are the highest of the cropping systems found in the watershed (according to the product of the C and P factors). Rainfed cropping systems, mainly soya and cassava, cultivated on sloping lands (hills and mountains) occupied by Phaozems, where the intrinsic soil erosion vulnerability is the highest, represent the watershed units the most prone to soil loss. In order to decrease soil losses in the watershed, soil conservation practices should be implemented, rstly, to the rainfed cropping systems. To optimize the effectiveness of conservation measures, the temporal

123

Landscape Ecol (2006) 21:13111325

1321

Fig. 5 Ground cover and rainfall erosivity annual distribution in the paddy rice (2 cycles) cropping system. *We used two labels for each month to represent the rst 2 weeks and the last 2 weeks of each month

Fig. 6 Ground cover and rainfall erosivity annual distribution in the corn in rotation with paddy rice (1 cycle) cropping system

dynamics of soil erosion vulnerability for each cropping system, mainly the most vulnerable, has to be assessed in order to determine the critical period. Temporal dynamics of soil erosion vulnerability for each cropping system The comparison of the ground cover and rainfall erosivity distributions throughout the annual calendar for each cropping system showed the temporal dynamics (periods of the year 2002) of the water erosion vulnerability. Since the cropping system annual calendars do not usually change from 1 year to

the other, the results can be extrapolated over the years. The months of the year with the highest rain erosivity are, in decreasing order, July, August, May, September and June in the rainy season. Cropping systems of paddy rice (2 cycles) and corn in rotation with paddy rice (1 cycle) imply two periods of land preparation and establishment that expose the soil surface to rainfall erosivity. The rst cultivation cycle starts during the dry season when the soil surface exposure induces a low vulnerability to water erosion because of the low rainfall intensity. Erosion vulnerability rises to its maximum during the second

123

1322

Landscape Ecol (2006) 21:13111325

Fig. 7 Ground cover and rainfall erosivity annual distribution in the paddy rice eld (1 cycle) cropping system

Fig. 8 Ground cover and rainfall erosivity annual distribution in the soya cropping system

cultivation cycle since the land preparation (ground cover 0%) coincides with the onset of intensive rainfall (July and August) (Figures 5 and 6). Agricultural activities on paddy rice elds (1 cycle) also start during the rainy season, in June. Nevertheless, at the time of the onset of heavy rainfall, the rice canopy cover (between 5% and 30%) already offers minimal soil surface protection (Figure 7). Agricultural activities on soya elds, contrary to the other rainfed crops, start in the rainy season, in June. After sowing, the ground cover during the rst

2 months is low and does not protect the soil surface against the intensive rainfall of June and July (Figure 8). Cassava and corn are sowed just before the beginning of the rainy season, in March. Consequently, these crops offer better soil surface protection during July and August. Nevertheless, corn is harvested during high erosive rainfall of August which increases soil exposure to water erosion. Corn thatches left on the eld following the harvest helps to reduce raindrop impacts and runoff (Figures 9 and 10).

123

Landscape Ecol (2006) 21:13111325

1323

Fig. 9 Ground cover and rainfall erosivity annual distribution in the cassava cropping system

Fig. 10 Ground cover and rainfall erosivity annual distribution in the corn cropping system

Discussion and conclusions Due to rainfall erosivity and unsustainable land-use practices, tropical countries face more soil erosion problems than temperate climate countries. Since most tropical countries do not have the necessary resources to develop their own soil loss prediction models, the spatial representation at the watershed scale of the Universal Soil Loss Equation (USLE), if modied to suit tropical conditions, can be a useful tool for sustainable soil conservation planning. This study provided a better understanding of the soil water erosion spatio-temporal dynamics in a

mountainous tropical watershed characterized by a wide variety of upland cultivation practices on which millions of people depend on for their livelihood. Since the cropping systems and physical environment are quiet similar in Ba Be National Park and its surroundings, the results will be provided to the PARC project members and may be used as a decision tool to create and implement management and conservation practices to promote sustainable agricultural development in the region. The local decision-makers who are aware of the Ba Be lake increasing sediment rate will be able to determine the scope of the erosion problem within the watersheds and to concentrate

123

1324

Landscape Ecol (2006) 21:13111325 Universities and Colleges of Canada (AUCC). The eld campaigns were funded by a grant from the Canadian International Development Agency (CIDA). The authors would like to extend their gratitude to the members of the PARC (Creating Protected Areas for Resource Conservation using Landscape Ecology) and SAM projects (Mountains Cropping systems) for their collaboration. We gratefully acknowledge Professor Nguyen Quang My, Ms. Laurie St-Onge and Vo May Dung for their assistance during the eldwork and also the hospitality of all the people living in the study area.

corrective activities where and when they will be more protable. Using a watershed approach to apply conservation practices should help reduce signicantly soil erosion consequences since it is recognized that there is an intimate relationship between soil and water resources and that soil management (land use) affects surface runoff and so the quality of the stream ow (non-point source pollution). Moreover, with targeted (and not general) and justied corrective actions, the local population pressure should less affect the viability of the decision support process suggested. The study limitations come mainly from the lack of spatial and ground truth data in the factor estimating process. It would be preferable in a future study to have the necessary technical and nancial resources to obtain maximum ground truth data from different data sources. On the one hand, quantitative results could then be used as absolute values and, on the other hand, the availability of multidate imagery could provide for the mapping of soil erosion vulnerability. However, this study has demonstrated that even in a poor data context, it is possible to assess the watershed unit soil erosion relative vulnerability spatio-temporal dynamics to promote local sustainable agricultural development. Nevertheless, the results obtained can only be extrapolated in regions, particularly in Vietnam uplands, where the physical environment and human agricultural activities are similar to those found in the study area. Otherwise, all the equation factors should be reevaluated. In the long term, future research should focus on developing a soil loss prediction model designed to compute soil loss at the watershed scale in tropical upland conditions. This would be essential to the understanding of the landscape dynamics and the impact of land use patterns on soil erosion specic to tropical countries. In this framework, GIS and remote sensing used in conjunction with other data sources can provide unique spatially distributed information used for the spatial representation of soil erosion models. These tools should be used in combination with actual and long term sediment load measurements in different sub-basins, the latter being chosen according to the landscape units dened in this paper.
Acknowledgments This study was carried out in the framework of the GEOMENSA project (Geomatics of environment and health in Vietnam, Hanoi) funded by the Association of

References
Boggs G, Devonport C, Evans K, Puig P (2001) GIS-Based rapid assessment of erosion risk in a small catchment in the wet/dry tropics of Australia. Land Degrad Dev 12:417434 Bonn F (1998) La spatialisation des modeles derosion des sols ` a laide de la teledetection et des SIG: possibilites, erreurs et limites. Secheresse 19(3):185192 Carson MA, Kirkby MJ (1972) Hillslope form and process. Cambridge University Press, London England Cohen MJ, Shepherd KD, Walsh MG (2005) Empirical reformulation of the universal soil loss equation for erosion risk assessment in a tropical watershed. Geoderma 124(3):235252 Dijk AIJM, Bruijnzeel LA (2001) Modelling rainfall interception by vegetation of variable density using an adapted analytical model. Part 2. Model validation for a tropical upland mixed cropping system. J Hydrol 247:239262 Elwell HA (1978) Soil loss estimation. Compiled works of the Rhodesian multidisciplinary team on soil loss estimation. Cyclo. Salisbury. In: Morgan RPC (ed) 1995. Soil erosion & conservation. 2nd edn., Longman group, John Wiley & Sons Inc, New York, NY, USA, p 71 Elwell HA (1978) Soil loss estimation. Compiled works of the Rhodesian multidisciplinary team on soil loss estimation. Cyclo. Salisbury. In: Morgan RPC 1995. Soil erosion & conservation, Second Edition, Longman group, John Wiley & Sons Inc, New York., NY, USA, p 71 FAO (1998) Topsoil characterization for sustainable land management (Draft). Land and Water Development Division Soil Resources, Management and Conservation Service, Rome Italy, p 71 Hudson N (1971) Soil conservation. Cornell University Press, Ithaca, NY, USA Jain MK, Kothyari UC (2000) Estimation of soil erosion and sediment yield using GIS. Hydrol Sci 45(5):771786 Jain SK, Kumar S, Varghese J (2001) Estimation of soil erosion for Himalayan watershed using GIS technique. Water Res Manage 15:4154 Kalpage FSCP (1974) Tropical soils: classication, fertility and management. Macmillan Company of India Limited, India Lai Vinh C (2000) Soil erosion study in northwest region of Vietnam by integrating watershed analysis and universal soil loss equation (USLE). Vietnam Natl Univ J Sci: Nat Sci 11:142150 Le Trong T, Eames JC, Kuznetsov AN, Nguyen VS, Bui Xuan P Monastyrskii AL (2001) A biodiversity survey and

123

Landscape Ecol (2006) 21:13111325 assessment of the Dong Phuc, Ban Thi-Xuan Lac and Sinh Long areas, Tuyen Quang and Bac Kan Provinces, Vietnam. Creating Protected Areas for Resource Conservation using Landscape Ecology (PARC) Project, Na Hang/Ba Be Component, Ba Be National Park, Be Be District, Bac Kan Province, Vietnam, p 55 Mati BM (1995) The effect of rainfall characteristics on splash erosion under crop covers. Trop Agric (Trinidad) 72:1:1822 Mati BM, Morgan RPC, Gichuki FN, Quinton JN, Brewer TR, Liniger HP (2000) Assessment of erosion hazard with the USLE and GIS: a case study of the Upper Ewaso NGiro basin of Kenya. International Int J Appl Earth Observ Geoinformat 2(2):7885 Millward AA, Mersey JE (2000) Adapting the RUSLE to model soil erosion potential in a mountainous tropical watershed. Catena 38(2):109130 Mitasova H, Hoerka J, Zlocha M, Iverson LR (1996) Modeling topographic potential for erosion and deposition using GIS. Int J Geogr Informat Sci 10(5):629641 Mitasova H, Mitas L, Brown WM, Johnston D (1998) Multidimensional soil erosion/deposition modeling and visualization using GIS. Final report for USA CERL. University of Illinois, Urbana-Champaign Illinois Morgan RPC (1995) Soil erosion & conservation, 2nd edn. Longman group, John Wiley & Sons Inc, New York NY USA Mulengera MK, Payton RW (1999) Estimating the USLE-soil erodibility factor in developing tropical countries. Trop Agric (Trinidad) 76(1):1722 Nguyen QM, Le TC (1990) Observation of soil erosion in Vietnam. Vietnam National University, Hanoi Vietnam Nguyen QM (1992) Initial research of present geo-environment in Vietnam. Regional Seminar on Environmental Geology, November 1113, Hanoi University, Vietnam, pp 105-111 Renard KG, Freimund JR (1994) Using monthly precipitation data to estimate the R-factor in the revised USLE. J Hydrol 157:287306 Renard KG, Foster GR, Weesies GA, McCool DK and Yoder DC (1997) Predicting soil erosion by water: a guide to conservation planning with the Revised Universal Soil

1325 Loss Equation (RUSLE). U.S. Department of Agriculture, Agriculture Handbook no. 703, USA, p 404 ` Roose E (1999) Introduction a la gestion conservatoire de leau, de la biomasse et de la fertilite des sols. GCES, Bulletin pedologique de la FAO Stone RP, Hilborn D (2000) Universal Soil Loss Equation (USLE). Ontario Ministry of Agriculture and Food, Agriculture and Rural Division; Factsheet, order no.00001, ISSN: 1198-712X. http://www.gov.on.ca/OMAFRA/ english/engineer/facts/00-001.htm Thai P, Mai VT, Do CD (2001) Erosion du sol sur les regions de collines, District Ninh Son, province de Ninh Thuan. Vietnam Soil Sci 15:161169 Tran K, Le Thai B (1999) Land environment (report on actual land environment in Vietnam). Ministry of Science, Technology and Environment, Hanoi Vietnam, p 21 Turkelboom F, Poesen J, Ohler K, Van Keer K, Ongprasert S, Vlassak K (1996) Assessment of tillage erosion rates on steep slopes in northern Thailand. Catena 29:2944 Vaesen K, Gilliams S, Nackaerts K, Coppin P (2001) Groundmeasured spectral signatures as indicators of ground cover and leaf area index: the case of paddy rice. Fields Crops Res 69:1325 Wall GJ, Coote DR, Pringle EA, Shelton IJ (2002) RUSLEFAC-Revised Universal Soil Loss Equation for application in Canada: a handbook for estimating soil loss from water erosion in Canada. Research Branch, Agriculture and Agri-Food Canada, Ottawa, Canada, pp 5360 Williams CN, Joseph KT (1970) Climate, soil and crop production in the humid tropics. Oxford University Press, Ely House, London, Great Britain Wischmeier WH, Smith DD (1981) Predicting rainfall erosion losses -a guide to conservation planning. Supplement to Agriculture Handbook No. 537. USDA, Washington DC, USA, p 58 Wu J, Hobbs R (2002) Key issues and research priorities in landscape ecology: an idiosyncratic synthesis. Landscape Ecol 17:355365 Zingerli C, Castella J-C, Pham Hung M, Pham VC 2002 Contesting policies: rural development versus biodiversity conservation in the Ba Be National Park Area. Hanoi, Vietnam, p 14

123

Você também pode gostar