Introduction

As in numerous other countries, natural rangelands in México are characterized by accelerated soil erosion and vegetation degradation as a result of overgrazing or poor grazing management (Thomas and Middlenton 1994; INE–SEMARNAT 2006).

In the central region of northern Mexico south of the Zacatecas state, the poor grazing management levels are estimated to exceed 140% at a state average stocking rate of 12 ha by animal unit per year (Echavarría et al. 2015).

These land conditions result from a need for low cost livestock production in a limited ecosystem (low rainfall depths; eroded, stony and low fertility soils and low levels of vegetation productivity).

However, it has been shown that through the use of selected rangeland management practices, land degradation can be controlled or minimized. Reduced stocking rates, rangeland reseedings, scrubland plantations, livestock exclusion areas, soil conservation practices, etc., help increase vegetation cover while limiting soil erosion (Sánchez 1984; Ruíz-Fernández et al. 2007).

From a hydrological perspective, a watershed’s response to rainstorms is shaped by the natural environment and by management practices. Outflow hydrographs for rainstorms integrate incremental runoff levels and elemental contributing hydrographs from all upstream sub-areas of a basin. Such hydrographs can be modified based on the effects of transit time through a basin, storage levels in stream channels of natural and artificial structures (small dams), and runoff contributions from upslope areas (i.e., upstream land conditions).

Insofar as morphological features remain constant in a watershed, runoff volume and hydrograph changes can also be caused by dramatic vegetation manipulations [e.g., forest clear-cutting (Swindel et al. 1982), overgrazing (Sánchez 1984) or sediment yields for urbanization in small watersheds (Sutjiningsih et al. 2015)].

This presents a complex problem to owners, practitioners and managers of natural resources when making management decisions. Therefore, mathematical computer simulation models can be used to support decision-making processes (Pierson et al. 1996). While numerous hydrological models are available, one must use the model that complements data availability levels that is consistent with the degrees of precision that are required, reasonably mimics reality, and considers computer and data reduction and processing costs. Hydrological decision model parameterizations are limited by the lack of reliable information available for less-instrumented watersheds. Nevertheless, in Mexico, there are some instrumented watersheds from which basic information can be obtained and used for modeling and decision-making purposes. On the other hand, low parameter empirical hydrological models can be used as an initial tool for modeling cause and effect relationships when the availability of reliable data are a limiting factor.

Curve number (CN) method

The CN method was developed by the USDA-SCS during the mid-1950s from experimental plots and small watersheds for numerous soils and vegetative cover conditions in the USA (SCS-USDA, 1972). The equation is used:

$$Q=\frac{{{{(P - {I_a})}^2}}}{{(P - {I_a})+S}}\quad{\text{for}}\,\,P\,{\text{>}}\,{\text{Ia,}}$$
(1)

otherwise:

$$Q=0$$
(2)

where:

Q = volume of accumulated runoff (mm).

P = accumulated rainfall (mm).

Ia = initial abstraction, including surface storage, interception, evaporation, and initial wetting infiltration (mm).

S = potential maximum retention of rainfall in the watershed following the fulfillment of Ia (mm).

The value of Ia is assumed to be:

$${I_a}=0.2S$$
(3)
$${{\varvec{\uplambda}}}={\text{Ia}}/{\text{S}}$$
(4)

The value of λ originally had a value of 0.2; however, numerous studies cited by Hawkins et al. (2010) suggest that the value of λ has a range from 0.02 to 0.07. Woodward et al. (2010) recommends a value of λ equal to 0.05. For this study the value of λ was 0.2 in accordance with the model TR-55

And after substitution into Eq. (5), the runoff equation becomes

$$Q=\frac{{{{(P - 0.2S)}^2}}}{{(P+0.8S)}}\,\,{\text{for}}\,{\text{P}}\,>0.2{\text{S}},{\text{otherwise Q}}\,=\,0$$
(5)

Originally developed based on 24-hour storm rainfall and runoff, the method has been widely used for other study periods as well. The CN is a transformation of S that situates the runoff parameter along a 0-100 scale. Using the metric system and millimeters (mm):

$$CN=\frac{{25400}}{{(254+S)}}$$
(6)

Following further substitutions from Eq. (5) and algebraic manipulations, this becomes (Hawkins 1993):

$$S=5[P+2Q - {(4{Q^2}+5PQ)^{0.5}}]$$
(7)

CN tables based on experience and recommended for use with different soil types, hydrological conditions and land uses are available from SCS-USDA (1972). An area-based weighted average CN can typically be obtained based on fractions of drainage areas in each land use and soils group. However, when a model is structured as such, it is preferable to subdivide the studied watershed into discrete areas of different CNs, to determine the weighted runoff for each component and to then sum the total (the average runoff level, not the CN value).

CN values for range conditions

For agricultural land use, CN values are associated with several soil and crop features such as tillage systems, crop types, seed dates, crop cover, etc. (i.e., Sartori et al. 2011; Bonta and Shipitalo 2013). However, for rangeland sites, such factors are very different given changes in grass and shrubs conditions that occur throughout the growing season.

Much research has been carried out to relate CN to rainfall, soil, geomorphological, cover, and land use and vegetation information. For example, for most watersheds (arid and otherwise) a strong relationship between CN and event rainfall has been consistently found, although additional responses to prior rainfall patterns and initial abstractions are more difficult to detect.

One distinctive aspect of arid watersheds is their ephemeral nature and absorptive channels. Simanton et al. (1996) reported that data-defined CN decrease with increases in drainage areas within a semiarid watershed in southern Arizona. This was attributed to the spatial variability of rainfall patterns and to flow abstractions in ephemeral channels. For runoff plots in southern New Mexico, Hawkins and Ward (1998) found CN to be highly correlated with plant cover, and in regards to the most pronounced responses to better (i.e., lower CN) sites, Rietz and Hawkins (2000) via a CN analysis found distinct impacts of grazing on rainfall-runoff. Of note, there were several instances of grazed pasture showing a difference of up to 15 CNs over the same watershed when applied to un-grazed meadows. An overview of the CN method is presented in Ponce and Hawkins (1996).

The objectives of this paper are:

  1. (a)

    To determine rainfall runoff relationships and the representative CN value of management conditions for a watershed located at in southern region of the state of Zacatecas in northern Mexico.

  2. (b)

    To parameterize the TR-55 model for simulating impacts of hydrological conditions due to rangeland management in terms of runoff.

  3. (c)

    To determine the relationship between stocking rates and hydrological conditions of this watershed.

Methods

Experimental “La Cruz” range watershed

It is assumed that the hydrological response measured at the outlet of a watershed is the integrated response to effects of management within the rainfall-runoff relationship. The experimental watershed (La Cruz) is located in the Tabasco county in the State of Zacatecas, México located between 102° 59ʹ and 103° 09ʹ W and between 21° 51ʹ and 21° 52ʹ N. It presents a drainage pattern and an average elevation of 1700 m above sea level with an area of 46.8 ha as shown in Fig. 1. Its climate classification is BS1 hw (García 1988) with a mean yearly temperature of 20.7 °C and with an average annual rainfall level of 582.1 mm (Medina and Ruíz, 2004). Approximately 90% of the rainfall falls from June–November. Rainfall constitutes the main source of the watershed’s water yields (Velásquez and Serna 1994).

Fig. 1
figure 1

Layout of measurement devices within the La Cruz range watershed in Tabasco, Zacatecas

Full size image

The watershed is divided into four paddocks; the largest paddock (24 ha) is positioned close to the flume (Fig. 1), and it is overgrazed all year. The other paddocks are grazed during the rainfall season (July–October). For this study the watershed was considered as homogeneus unit eventhough there is empirical evidence about the spatial distribution of rainfall as well as some physical properties and attributes of the vegetation.

Natural vegetation in the watershed includes Mesquite (Prosopis spp.), Huizache (Acacia spp.) and ozote (Ipomea intrapilosa Rose.), which are also found within the landscape (Fig. 2). Shrubs found here include Varaduz (Tysenhartia sp.) and Huiscolote (Mimosa sp.) and some native grasses, such as navajita (Bouteloua gracilis H.B.K.) and banderita [B. curtipendula (Michx) Torr], are also found here. In a monitoring site near the watershed, Sanchez (1984) found significant differences (P > 0.05) between the foliar cover of grazed areas (15.6%) and excluded sites (53.7%). In addition, mean basal bare ground cover was not found to vary statistically (P > 0.05) between grazed (75.1%) and excluded (55.6%) areas. Biological degradation is represented by a lack of desirable grass species and by the presence of non-desirabl e bushes for animal consumption.

Fig. 2
figure 2

Vegetation conditions of rangeland in the La Cruz watershed of Zacatecas, México

Full size image

According to the FAO’s classification (1990), soils in the watershed are shallow and moderately textured with stony phase calcaric Regosols and haplic Chernozems (Velásquez and Serna 1994). Stony phase shallow soils are present with an outcrop of clay horizons through the watershed.

The watershed’s soils have been severely degraded as consequence of overgrazing, gullies and channel soil erosion. It has been reported (Sanchez 1984) that interril soil erosion (small runoff plots of one square meter) in this area reaches 0.17 and 0.12 ton ha−1 for grazed and excluded sites, respectively. With respect to bed load, Velásquez (1991) reported a range of surface runoff transport capacity levels of 2.5–19.5 ton ha−1.

The hydrological behaviors of soils in the La Cruz basin correspond to hydrologic group “D” This hydrologic group is characterized by high levels of runoff potential, as sub-layers of the watershed’s soils present high levels of clay (including expandable clays) and slow infiltration and water transmission rates close to the surfaces of soil parental materials (USDA-SCS 1972; Wood and Blackburn 1984).

For soil in the La Cruz basin, the high runoff generation and low hydraulic conductivity levels are mainly due to high levels of soil clay content and lower store water soil capacities. This was reported by Serna and Velasquez (1986), who found flow variables (volume and time to peak) to be positively correlated (P > 0.01) with depth and the intensity of rainfall events, causing the watershed to generate a strong hydrological response to the magnitude of rainfall events in the examined year. Infiltration tests were carried out next to the La Cruz rangeland watershed (Sánchez 1984) using a Blackburn mobile rainfall infiltrometer (Blackburn et al. 1974) in grazed and ungrazed (exclusion areas) range sites; according to soil surface conditions and land management patterns, infiltration rates were found to be significantly different between grazed and ungrazed areas.

Watershed instrumentation

A single tipping bucket rainfall gauge is located in the upper region of La Cruz watershed. The runoff gage located in the lower region of the watershed is a Santa Rita supercritical flume (Gwinn 1964; Smith et al. 1981) (see Fig. 3). Runoff events begin in July and end in October of each year. It is important to note here that flow depths observed and reported here are the total flows at the lower end of the watershed where the flume is located (see Fig. 1).

Fig. 3
figure 3

Santa Rita supercritical flume at the outlet of the Zacatecas watershed in La Cruz, Tabasco

Full size image

Using the above listed devices, a daily hydrological database was generated for 1985 to 1988 and for 1990, 1993 and 1994 (n = 72 rainfall-runoff pair values). The discontinuity in the years of observed values of rainfall and runoff (missing years) was due to a reduction in logistical support for the watershed data collection.

Watershed runoff simulation

In order to quantify the hydrological impacts of management practices in the rangeland area, a model was used. We simulate different management scenarios, i.e., different CN values. We simulate a regular condition of CN = 84 for soil cover levels of 50–75%, with good conditions presenting a value of CN = 80 and soil cover levels exceeding 75%.

The WIN TR-55 model

The computer model considered here is the WIN TR-55, a single-event rainfall-runoff small watershed hydrologic model originally developed by the USDA (USDA-NRCS-CED 1986). The model generates outflow hydrographs for urban, agricultural and range areas at selected points along stream systems. Hydrographs are routed downstream through channels and/or reservoirs. Multiple sub-areas can be modeled within a watershed. To calculate runoff depths, the WIN TR-55 Model uses the SCS CN method (USDA-SCS 1972), and the unit hydrograph method is used to route runoff to channels. The following criteria for model TR-55 selection were used:

  • A model for a watershed version

  • Readily available information

  • Easy to use and understands the findings of others

  • Suited to a small watershed scale

The TR-55 model procedure follows the following basic procedural steps:

  1. 1.

    Estimation CN from look-up tables based on soil and land use characteristics representing different land uses and drainage areas.

  2. 2.

    Calculations of the time of concentration. These are used in unit hydrograph routing procedures for the model.

  3. 3.

    Determination of the Type II rainfall distribution and of 24-h rainfall depths.

  4. 4.

    Using the unit hydrograph approach, a hydrograph of direct runoff for the watershed can be formed.

  5. 5.

    From the above, using the estimated rainfall depths, incremental runoffs are determined, and using the unit hydrograph approach, runoff levels for each unit or paddock are calculated, lagged and summed to the outflow from the total drainage area. Table 1 presents model parameter values used for the simulation.

Table 1 Features of parameters of the rangeland watershed under traditional management
Full size table

A possible contribution to the use of the tabular values presented in Table 1 is referred to the weighted value of roughness of Manning, instead of an absolute value. This weighting considers various factors that affect the roughness of the average physical (\(n\)):

$$n=({n_b}+{n_1}+{n_2}+{n_3}+{n_4})m$$
(8)

where n b is a coefficient which corresponds to an even channel with gradual slope with natural materials; n 1 is a correction factor due to surface irregularities; n 2 is a value that considers variations in the size and shape of the cross section of the channel; n 3 is a value that considers the obstructions; n 4 a value takes into account the vegetation and conditions of flow and \(m\) is a factor of correction for the meanders of the channel (Arcement and Schneider 2004; Limerinos 1970; Chow et al 1988; French 1993).

Curve number determination

The CN value input for the WIN TR-55 Model was calculated two different ways: the first involved using Eq. 5 (Hawkins 1993) in consideration of all rainfall-runoff events (n = 72). The second approach involved fitting. To determine a CN value for the La Cruz watershed from a P:Q relationship, an adjusted method involving Eq. (7) was used to fit the P:CN original database

$$CN=C{N_\infty }+(100 - C{N_\infty }){\exp ^{( - kP)}}$$
(9)

where CN∞ is the constant value as P → ∞, and k is a fitting constant that quantifies the rate at which CN approaches CN∞. Bonta and Shipitalo (2013) suggest that larger k values result produce CN∞ values applicable to smaller P values and vice versa. Equation 9 was fitted using a least-squares procedure for CN∞ and k (Sneller 1985; Zevenbergen 1985). A recent report to NRCS (Woodward et al. 2010) recommends this procedure as the preferred technique for CN parameterization.

The criteria proposed by Nash and Sutcliffe (1970) were used to determine the model’s efficiency and to compare predicted runoff values to the mean of the observed runoff data (Wang et al. 2006) using:

$$EI=1 - \frac{{\sum\nolimits_{{t=1}}^{r} {{{(Q_{0}^{t} - Q_{m}^{t})}^2}} }}{{\sum\nolimits_{{t=1}}^{r} {{{(Q_{0}^{t} - {{\bar {Q}}_0})}^2}} }}$$
(10)

where EI is the Nash–Sutcliffe Efficiency Index (EI), Q0 is the mean runoff depth (mm), and Q m is the modeled runoff depth (mm) at time t. The EI can range from −∞ to 1. EI = 1 corresponds to a perfect match between observed and modeled data. When EI = 0, the model predictions are as accurate as the mean of the observed data, and if the EI is less than zero, the observed mean is a better predictor than the model. The coefficient of determination (R2) was also used to determine how well data fit a statistical model or to determine levels of association between hydrological variables.

Results

Rainfall-runoff relationship

From a hydrological point of view, the rainfall runoff relationship allows one to infer the hydrological behaviors of a watershed. Figure 4 presents the rainfall-runoff relationship (72 daily events) for the La Cruz watershed based on our analysis of the hydrological database (1985–1988 and 1990, 1993 and 1994).

Fig. 4
figure 4

Rainfall-runoff relationship for the La Cruz rangeland watershed in Zacatecas

Full size image

The watershed hydrological behaviors shown are typical, as they present a threshold rainfall value of 13 mm for initiating runoff (see the intercept of the rainfall runoff relationship shown in Fig. 4). While a better fit with a nonlinear model could be obtained, the intercept would appear fuzzy and would be difficult to interpret.

Serna and Velasquez (1988) reported that the spatial variability of rainfall and physical properties of soils in the La Cruz basin cause antecedent soil moisture and infiltration conditions to shape hydrological responses within the watershed. For this reason, for rainfall events of the same magnitude, surface runoff events of varying magnitudes occur at the outlet of this watershed.

Nevertheless, for this watershed, some factors that were not considered in this study (e.g., the physical degradation of the watershed, soil gravel content, the impervious riverbed, etc.) may also influence the magnitude of the rainfall-runoff relationship.

CN determination from La Cruz watershed data

Considering the rainfall runoff relationship for the watershed, the CN value of 83.1 obtained whit the Eq. 5 is acceptable, meaning that the relative rapid response to rainfall events would not be apparent for CN values of less than 80. The ordering of runoff data for obtaining a fitted model of CN did not yield different results compared to the CN value derived from Eq. 5.

It has been reported elsewhere that the CN value decreases as rainfall depth increases (Hawkins 1993; Sánchez et al. 2003; Bonta and Shipitalo 2013). A similar trend was found for ordered rainfall events and their respective CN values (large dash discontinuous) and for ordered runoff events and their respective CN values (continuous line) as shown in Fig. 5 (considering P > 0.2S and CN0 = 100/(1 + P/50.8) denoted by (short dash discontinuous line)).

Fig. 5
figure 5

CN-rainfall plot for the La Cruz rangeland watershed. The empty circles denote original data and the filled circles denote rank-ordered (frequency-matched) points. The dashed and solid lines are fitted to the potential models for original and rank-ordered rainfall and runoff, respectively

Full size image

As shown in Fig. 5, the fitting line of the potential model (y = 87.28 x−0.01) for runoff ordered data tends to be asymptotic. Hawkins (1993) reported this watershed behavior as a common and standard hydrological response. Thus, for the La Cruz watershed, the rainfall depth:CN relationship is classified as a standard response. The runoff depth of this watershed may be shaped by a source process (e.g., overland flows due to shallow soils (0.33 m)) based on advanced erosion processes and slope steepness (19.4%) into the watershed. In the same figure, when mm is applied, CN0 = 100/(1 + P/50.8). This trend line (lower slope) equilibrium for ordered runoff roughly emerges when rainfall values ~13 mm and when CN is close to 82.4. This value of precipitation coincides with the threshold value of runoff initiation.

On the other hand, an additional result of the least-squares procedure for CN∞ (Eq. 7) is that the CN value for this rangeland watershed is calculated at 82.9. The procedure generated a value of 0.2 k. A statistical summary of the CN is presented in Table 2.

Table 2 Summary of the CN values found using different procedures and the rainfall-runoff La Cruz watershed database
Full size table

It is recommended that the appropriate CN value for a watershed be computed from measured data, as the tabulated values (SCS-USDA 1972) of vegetation conditions are somehow discrete and present a wide range of values, thus preventing us from making an adequate CN selection.

Runoff depth simulation based on the CN value

For the only purpose of parametrize the WIN TR-55 Model and quantify the impact of improvement watershed’s hydrological conditions on runoff production, seven rainfall events were selected from rainfall records, ranging from 16.4 mm to 65.2 mm with the objective to simulated runoff events. This model was parametrized to different CN values. The model’s capacity prediction was evaluated using the Regression Coefficient (R2) and the Index Efficiency of Nash and Sutcliffe (IE) as a decision criterion.

The results of TR-55 Model runoff simulation with CN = 80 shows good agreement with the observed runoff catchment data (R2 = 0.86); the model’s capacity to represent runoff processes at the watershed’s outlet is thus acceptable according to the Nash–Sutcliffe Efficiency Index (0.65). With CN = 83, a regression coefficient similar was found (R2 = 0.90); however, a low Nash–Sutcliffe Efficiency Index was obtained (0.43). In the case of runoff simulations with CN = 89, a regression coefficient was equal to the Nash–Sutcliffe Efficiency Index was −1.52, for this the average measured runoff is better predictor than the average simulated runoff value. Here, the average of the runoff simulated values was two times more than the observed runoff values.

The results of the runoff simulations with the WIN TR-55 model as a function of different hydrological conditions (CN) in the small watershed are presented in the Fig. 6.

Fig. 6
figure 6

Hydrological impact modeling on outlet watershed runoff for different rainfall depth and CN values: CN = 80 (empty circle), CN = 83 (empty square) and CN = 89 (empty triangle). The filled circles represent the observed rainfall-runoff relationship

Full size image

The impact of different hydrological conditions (CN) can see in this Figure; according with WIN TR-55 runoff simulations with CN = 80 (good hydrological condition), needs more than 25 mm to start surface runoff; while for CN = 83 (fair hydrological condition) was found that rainfall events of 22 mm or more can produce surface runoff. On the other hand, under fair hydrological conditions (CN = 83) the rainfall threshold for producing runoff was reached with 16 or more mm. Also, this figure shows that after 37 mm rainfall events all runoff predictions under fair and poor hydrological conditions over predict the watershed’s observed water yield. As a summary, we can say that it’s possible reduce surface runoff improved the hydrological conditions; mainly when the rainfall depth is lower than 50 mm.

The general results show a tendency toward increasing the difference between observed and simulated runoff values, mainly in the case of longer rainfall events (more than 50 mm). The results also showed that a when the CN value was high (CN = 89) the model capacity to over predict was high. May be, a specific initial abstraction parameter under longer rainfall events can be relation with this sensitivity (Jacobs and Srinivasan 2005). The diversity of watershed responses may also be attributed to an inadequate selection of model parameters available based on hydraulic conductivity and CN values (Hernandez et al. 2000). With respect to this, Ponce and Hawkins (1996) suggest that the retention parameter for initial abstractions should be interpreted as a regional parameter.

On the other hand, the soil and land watershed conditions (advanced physical degradation) suggest the existence of a special CN value that is not considered in the USDA-SCS Handbook. This is interesting to note because it is necessary to identify modeling hydrological responses to heavily degraded watersheds, wherein some physical processes (e.g., soil property improvements, water erosion decline, etc.) are impossible to improve over a short time period while increasing soil, water and vegetation quality.

Reducing runoff volume may be achieved by reducing CN value as is showed in the Fig. 6 via management practices such as reseeding; the same effect may be obtained by reducing animal stocking rates and with the establishment of exclusion areas; any of these management practices, also implies a reduction in soil erosion rates (Sánchez 1984).

Linking hydrology and rangeland management (stocking rates)

In regards to the deterioration of natural resources in rangelands as a result of por management practicies in the southern region of the State of Zacatecas, for those production units in which primary economic activities involve livestock production, and according to its low net primary production 53.8 animal units ha−1 year−1 may be reached and for some áreas during very dry periods the coefficient of rangeland can reach 64.3 animal units ha−1 year−1 (Echavarría et al. 2009). For the southern region of the state, where values of average annual rainfall exceed those for the rest of the state (more than 700 mm average annual), the coefficient of rangeland for this region is minor relative to that for the rest of the state.

In relating this management information to hydrologic behavior, Table 3 shows that for poor conditions (less than 50% of the soil by vegetation cover) and a CN of 89, 17 ha are needed to maintain an animal unit for one year; for sites with more than 75% of coverage and a CN of 80, only 3 ha are required annually per animal unit, in accordance with the approach of the condition hydrological documented in the method of curve numerical (SCS-USDA 1972; USDA-NRCS-CED 1986).

Table 3 Curve number values of different management scenarios for the rangeland catchment of La Cruz, Tabasco, Zac.
Full size table

Based on this, it is relevant to note that in terms of handling or stocking, owners of grassland areas that require rangeland deterioration reversal will need to reduce their animal unit numbers by more than 80% to improve the overall biotic parameers that defines runnof yield. Rangeland stocking rates adjustments reduce water losses in the form of runoff by approximately 70% relative to expected values for traditional or continuous grazing for this region of the state (Sánchez 1984).

The Fig. 7 shows the relationship between stocking rates and CNs for the watershed. The figure shows that the watershed is more sensitive (sensitivity is defined herein as the first derivative of the fitted model of CN with UA) to stocking rate changes for low CNs, suggesting that one should be careful when managing rangeland in this case, as a minor change in ha UA−1 will produce significant changes in CN. For this watershed, a stocking rate of more than 15 ha UA−1 will not disturb watershed conditions any further. The vertical line shows a preliminary rangeland threshold for sustaining actual watershed conditions; moreover, the projection of the intersection of the curve of sensitivity and the model fitted to the CN axis generate a value of CN = 83.

Fig. 7
figure 7

Relationship between rangeland management and hydrological (CN) factors for the La Cruz watershed

Full size image

The main contributions of this study are related to the use of a rainfall and hydrological database of small partially instrumented watershed which were used to generate hydrological information (CN) which was integrated to the TR-55 model. Thus, the application of the method for the generation of a local value of numerical curve contributes to the understanding of the relationship between the management of natural resources and the temporal dynamics of physical processes (surface runoff, among others). This procedure can be used and validated for other basins of the country.

Conclusions

Hydrological behaviors of the watershed described illustrate the generality of arid watersheds in the central region of northern Mexico. Limited ground cover and low infiltration rates caused by por management practices or overgrazing along with summer high rainfall intensities are the main cause of water yields and soil losses in rangeland watersheds. The hydrological database for the La Cruz rangeland watershed was used to determine the CN value.

By adjusting the P:CN of the La Cruz watershed database by applying a least-squares procedure for CN∞ and k, we found a CN value that describes the runoff amount for a rangeland watershed in central México. CN = 82.9; as an CN representative value for rangeland watersheds characterized by poor cover brush-brush-weed-grass mixtures, hydrologic soil group D and ground cover levels of less than 50% with a 19% slope steepness and continuous grazing system.

In this study, the functionality of the TR-55 WIN model and the predictive capacities of the CN method in simulating individual events of surface runoff in a small rangeland watershed of less than 50 ha poor grazing management (fair condition, (CN = 83) were found to be adequate according to the regression coefficient (0.90) and Nash and Sutcliffe efficiency index (0.87) for the observed and simulated runoff depths).

It is worth noting that the TR-55 WIN model is empirically based on the CN and that physically based models may generate better predictions of the rainfall runoff relationship, though they require access to more parameters. It is necessary to obtain CN values in the field and the use of tabular values should be avoided, as these tend to overestimate such values, and one may in turn make inappropriate decisions in regards to runoff use. A tabular value is the last resource needed to develop an understanding of how rainfall runoff may develop in ungagged watersheds.

It is important to obtain local values of stocking rates and correlations with CN values in order to determine the threshold value that defines the start of the deterioration of watershed’s natural resources.