An intelligent optimization method for grazing strategies with different soil chemical properties

Abstract

An intelligent optimization method for grazing strategies for different soil chemical properties, based on the soil carbon and nitrogen monitoring data of different grazing intensities collected in recent years, detects outliers and missing values ​​of the data, uses the generated chemical property and grazing intensity correlation diagram to predict the soil chemical property data of the next cycle, combines the relationship coefficient, obtains the impact value of different grazing intensities on soil chemical properties, and performs correlation analysis; adopts a time series prediction method, uses five chemical properties of soil organic carbon, soil inorganic carbon, soil total carbon, soil total nitrogen, and soil C / N ratio as input, and grazing intensity as output, and finally returns the prediction result and selects the most suitable grazing strategy. The present invention provides the results of intelligent optimization of grazing strategies for different soil chemical properties by establishing a prediction model of the influence of grazing strategy and soil chemical properties, and provides a better strategy for reasonable grazing.

Description

Technical Field

[0001] The invention belongs to the field of computer technology, relates to grassland animal husbandry, and particularly relates to an intelligent optimization method for grazing strategies with different soil chemical properties. Background Art

[0002] Soil chemical properties include: soil organic carbon SOC, soil inorganic carbon SIC, soil total carbon STC, total N, soil C / N ratio, etc. The content of these organic and inorganic substances can make various mineral nutrients persistent and affect the growth of grassland plants. The higher the nutrient level of the soil, that is, the fertile soil, the higher the vegetation coverage of the grassland. Under good vegetation coverage, the grazing intensity can be appropriately increased. In a certain area, the higher the grazing intensity, the more grazing. Without considering grazing subsidies and price fluctuations, more grazing means higher grazing benefits.

[0003] Therefore, an intelligent optimization method for grazing strategies with different soil chemical properties is now needed. The prediction of this model is conducive to better development of grassland animal husbandry. The advantage of this method is that it reduces the impact of overgrazing on grasslands and protects the diversity of ecosystems. However, there are few studies on grazing strategies involving soil chemical properties. How to use various data of different soil chemical properties in actual scenarios to establish excellent grazing strategies is an urgent problem to be solved. Summary of the invention

[0004] In order to overcome the shortcomings of the above-mentioned prior art, the purpose of the present invention is to provide an intelligent optimization method for grazing strategies with different soil chemical properties, collect soil carbon and nitrogen monitoring data of different grazing intensities within a grassland T time, and use time series analysis to establish a time series model of five values ​​of different grazing intensities and soil organic carbon, soil inorganic carbon, soil total carbon, total nitrogen, and soil C / N ratio. Based on the prediction of soil chemical properties under different grazing strategies, a reasonable prediction of soil property data affected by future grazing can be made on the basis of the original data, and effective grazing operations can be achieved through the predicted data, avoiding the impact of overgrazing and unreasonable grassland operations on the ecological environment.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] An intelligent optimization method for grazing strategies with different soil chemical properties, including:

[0007] Step 1, collecting soil carbon and nitrogen monitoring data of different grazing intensities within a grassland T time, including five soil chemical property data of soil organic carbon content, soil inorganic carbon content, soil total carbon content, soil total nitrogen content and soil C / N ratio at different grazing intensities; the grazing intensity is divided into four categories: control group, mild, moderate and severe;

[0008] Step 2, making a correlation diagram between five soil chemical properties and grazing intensity based on the soil chemical property data, and obtaining the correlation coefficients between the five soil chemical properties and grazing intensity;

[0009] Step 3, predicting the soil chemical property data of the next cycle, and combining the relationship coefficient to obtain the impact value Q of different grazing intensities on soil chemical properties;

[0010] Step 4, constructing a time series model, taking the soil chemical property data of the next cycle and the impact value Q as input, and outputting the grazing intensity of the next cycle, and adjusting the grazing strategy according to the output result, wherein the grazing strategy is characterized by the carrying capacity per unit area.

[0011] Compared with the prior art, the beneficial effect of the present invention is that the grazing strategy on the grassland in the future can be reasonably optimized based on the data of soil chemical properties collected under different grazing intensities over the years, so as to maximize the economic benefits while ensuring the benign and healthy development of the ecological environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a flow chart of the present invention.

[0013] Figure 2 Time series diagram of five chemical properties.

[0014] Figure 3 Correlation diagram of chemical properties and grazing intensity. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the prediction method of soil chemical properties based on different grazing strategies of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0016] The embodiments of the present invention are described in detail below with reference to the accompanying drawings and examples.

[0017] like Figure 1 As shown, the present invention is an intelligent optimization method for grazing strategies with different soil chemical properties, comprising the following main steps:

[0018] Step 1: Collect soil carbon and nitrogen monitoring data of different grazing intensities in the grassland within T time, including five soil chemical property data of soil organic carbon content, soil inorganic carbon content, soil total carbon content, soil total nitrogen content and soil C / N ratio at different grazing intensities. 2 ), light grazing (2 sheep / day / hm 2), moderate grazing (4 sheep / day / hm 2 ), heavy grazing (8 sheep / day / hm 2 ) Four grazing intensities were selected. The experimental design adopted a randomized block method, with three replicates for each grazing intensity.

[0019] Generally, T is measured in years, and the collection period is T1 (eg, 2 years). The soil chemical property data obtained in the present invention is time series data.

[0020] Step 2: In order to better analyze the relevant data and the strength of the correlation, a test was conducted to avoid missing or abnormal data. A correlation diagram was drawn based on the collected data on soil organic carbon, soil inorganic carbon, soil total carbon, soil total nitrogen, and soil C / N ratio. A correlation analysis was performed to obtain the relationship coefficients between the five soil chemical properties and grazing intensity.

[0021] Through the correlation diagram, the trend and cycle of changes can be displayed, and jump points and inflection points can be found. Jump points refer to observations that are inconsistent with other data. If the jump point is a correct observation, it should be taken into account in the time series analysis; if it is an anomaly, that is, it is out of the concentrated area, the jump point should be adjusted to the expected value. The inflection point refers to the point where the time series suddenly changes from an upward trend to a downward trend. If there is an inflection point, different models must be used in step 4 to segmentally fit the time series separated by it, such as using a threshold regression model.

[0022] Step 3: Predict and calculate the five soil chemical property data for the next cycle, and combine the relationship coefficients to obtain the impact value Q of different grazing intensities on soil chemical properties.

[0023] By analyzing the material cycle of the ecosystem, the calculation formula of the impact value Q is as follows:

[0024] Q=C1*k1+C2*k2+C3*k3+C4*k4+C5*k5

[0025] Among them, C1~C5 predict the soil chemical properties data of the next cycle, k1~k5 are the relationship coefficients obtained through the correlation diagram in step 1, C1 is the soil organic carbon content, C2 is the soil inorganic carbon content, C3 is the soil total carbon content, C4 is the soil total nitrogen content, C5 is the soil C / N ratio, k1 is the relationship coefficient between soil organic carbon content and grazing intensity, k2 is the relationship coefficient between soil inorganic carbon content and grazing intensity, k3 is the relationship coefficient between soil total carbon content and grazing intensity, k4 is the relationship coefficient between soil total nitrogen content and grazing intensity, k5 is the relationship coefficient between soil C / N ratio and grazing intensity. Through the above contents and relationship coefficients, Q is obtained to be the influence of grazing intensity on the chemical properties of grassland soil.

[0026] Soil chemistry is affected by factors such as precipitation, grazing intensity, and vegetation coverage. Precipitation carries a small amount of chemical substances, which the soil absorbs as it absorbs water, regulating the soil's pH to a certain extent; soil microorganisms decompose dead plants and livestock excrement to supplement the organic and inorganic substances in the soil. Appropriate grazing intensity and vegetation coverage can improve soil fertility, but excessive grazing can lead to problems such as an imbalance in soil pH.

[0027] For example, to calculate the soil organic carbon content in the latest year, the soil organic carbon content at the next moment = the soil organic carbon content at this moment + the soil organic carbon content supplemented by external factors - the soil organic carbon content consumed. The consumption value of soil organic carbon content mainly considers the vegetation coverage rate, and the supplementary value of soil organic carbon content mainly considers precipitation, plant wilting rate, and biological excretion, and then further establishes the calculation formula of soil organic carbon content.

[0028] Specifically, the soil chemical property data prediction formula for the next cycle is as follows:

[0029] X t+1 =X t +X 补 -X 耗

[0030] Among them, X is C1~C5, X t+1 Represents the five soil chemical properties data for the next cycle, X t Represents five soil chemical properties data of the current cycle, X 补 represents the supplementary value of a soil chemical property data at time t~t+1, X 耗 Represents the consumption value of a certain soil chemical property data from t to t+1.

[0031] Among them, for X 耗 , only considering the changes in soil chemical properties within time t, the consumption of soil organic carbon content is mainly related to vegetation coverage. Therefore, the relationship between soil organic carbon content and vegetation coverage is established, and the formula is as follows:

[0032] X 耗 =h1*C pt

[0033] For X 补 , considering that soil chemicals are mainly replenished through precipitation, microbial decomposition of livestock excrement and withered plants. Insects on grasslands also feed on plants, but the amount consumed is negligible compared to livestock. Therefore, the organic and inorganic matter obtained by microbial decomposition of insect excrement is not considered here. The established relationship formula between soil organic carbon content replenishment and precipitation, plant withering rate, and biological excretion is as follows:

[0034] X 补 =h2*P t +h3*D t +h4*Po t

[0035] Among them, h1 is the coefficient of vegetation coverage, C pt is the vegetation coverage at time t, h2 is the coefficient of precipitation, P t is the precipitation at time t, h3 is the coefficient of plant wilting rate, D t is the plant withering rate at time t, h4 is the coefficient of biological excretion, Po t is the biological excretion at time t.

[0036] The calculation formula for plant wilting rate is:

[0037]

[0038] Among them, β t is the soil moisture content at time t, w * is the characteristic value of the grassland (it is only an approximate scale, introduced to make the corresponding dimensionless quantities such as w / w* equal to 1 for ease of calculation).

[0039] When the daily feed intake of a standard sheep unit is 1.8 kg, the conversion factor for large livestock is 6.0, and the conversion factor for cattle, horses, camels, and large livestock cubs is 3.0, the calculation formula for biological excretion is:

[0040] Po t =1.8μ*M

[0041] μ is the livestock metabolic coefficient, and M is the livestock carrying capacity per unit area.

[0042] In summary, the calculation formula for a certain soil chemical property data in the next cycle can be expressed as:

[0043]

[0044] The stocking capacity per unit area is used to represent the grazing strategy, and t = T1 is taken. The above parameters are obtained when collecting data and are all known. When calculating different data, x t Different meanings have different values.

[0045] Step 4: The collected data is collected in years, and the soil chemical property data obtained is time series data, so time series analysis is used. A time series model of different grazing intensities and five values ​​of soil organic carbon, soil inorganic carbon, soil total carbon, soil total nitrogen, and soil C / N ratio is established, and the soil chemical property data of the next cycle and the impact value Q are used as inputs. The output is the grazing intensity of the next cycle. According to the output results, the grazing strategy is adjusted, where the grazing strategy is characterized by the carrying capacity per unit area.

[0046] Time series analysis usually uses the Auto-Regression and Moving Average Model (ARMA model), which includes two different parts, namely autoregression and moving average. The number of autoregression terms p represents the upper limit of the order value of the first part, and the number of moving average terms q represents the upper limit of the order value of the second part. The autoregression and moving average process can be expressed as ARMA (p, q). The specific expression is as follows:

[0047]

[0048] X t represents observation data; {Z t}~WN(0,σ 2 ) represents a random variable sequence with a mean of zero and no correlation between the two. The mean square error of this random variable sequence actually represents the error in the measurement of chemical properties. is the autoregressive coefficient, p is the autoregressive order; (θ1, θ2, …, θ q ) is the sliding average coefficient, and q is the sliding average order.

[0049] First, a stationarity test is performed. The time series of soil chemical property data for different grazing strategies in different years meet the following requirements: 1) For any time T1, its mean is always constant; 2) For any time T1 and T2, the correlation coefficient of this time series is determined by the time period between the two time points, and the starting points of the two time points will not cause any influence.

[0050] Therefore, such a time series can be considered as a stationary time series. Using the data graph test method, five chemical property time series graphs are made. It can be seen from the graph that the collected data over the years conform to the stationary time series. The latest soil chemical property data prediction results are combined with the relationship coefficient between different grazing intensities and soil data to obtain the impact of grazing strategies on grassland soil chemical properties.

[0051] Secondly, the model is identified and the order is determined. The second-order characteristics of the ARMA (p, q) process, AR (p) process and MA (q) process have the rules shown in the table below. The model is identified by calculating the tailing or truncation of the sample autocorrelation function and the sample partial autocorrelation function. After calculation, the autocorrelation function and partial autocorrelation function of soil chemical properties show that the dynamic model is the ARMA (p, q) process.

[0052] ARMA(p,q) sequence MA(q) sequence AR(p) sequence Autocorrelation function Trailing q-step truncation Trailing Partial autocorrelation function Trailing Trailing P-step truncation

[0053] When the truncation of the autocorrelation function and the partial autocorrelation function is used to determine whether the model is an ARMA model, the order of p and q cannot be determined. In order to determine the order of p and q more accurately, it must be combined with the commonly used order determination criteria. The order determination criterion used is AIC (A-Information Criterion), and the order is determined in combination with the input Q value. The final positioning order with the smallest value is selected using the AIC criterion. After calculation, when p=3, q=5, AIC and BIC obtain the minimum value, and the measured data should meet the ARMA (3, 5) model.

[0054] Finally, the model parameters are estimated and the model is tested. The measured data of soil chemical properties are now determined to be fitted with ARMA (3, 5). In order to prove the accuracy of the model, a model fit test is also required. The core idea of ​​the test is that if the dynamic model is appropriate, the random sequence represented on the right side of the model should be a reality that satisfies the Gaussian distribution random process. Using the chi-square test method, first construct the test statistic, giving a significance level of α = 0.05; then, find the upper quantile χ 2 α (m-r), when χ 2 ≤χ 2 α (m-r), it is considered that Z t is a Gaussian distributed random sequence, and the model passes the test.

[0055] Grazing strategy optimization means that after obtaining different grazing intensity results for different areas, the latest two grazing intensities and five soil content data for the same area are compared. If the soil chemical property data is higher than that of the previous year under light grazing intensity, it means that this area can accept light grazing or even adjust to a stronger grazing intensity, and the grazing strategy can be adjusted next year; on the contrary, if the data is lower than that of the previous year under light grazing intensity, it means that this area should not be grazed next year. In summary, reasonable adjustments to grazing strategies follow the principle of sustainable development.

[0056] Taking a representative grassland in Xilin Gol League, Inner Mongolia Autonomous Region as an example, the data set uses the soil carbon and nitrogen monitoring data set of different grazing intensities collected from the grassland from 2012 to 2020. Some of the data are shown in the following table:

[0057]

[0058] Step 1: Collect soil carbon and nitrogen monitoring data of grasslands with different grazing intensities from 2012 to 2020, as shown in the table above; detect outliers and missing values ​​in the collected soil carbon and nitrogen monitoring data sets with different grazing intensities from 2012 to 2020. Through detection, it is found that the data is complete, without outliers or missing values. And judge that the time series is a stationary time series, perform stationarity test and processing on the data, and use data graphs to make time series graphs of five chemical properties for the data in the data set (such as Figure 2 The results show that the given data conform to the stationary time series.

[0059] Step 2: Make a correlation diagram based on the soil organic carbon, soil inorganic carbon, soil total carbon, soil total nitrogen, and soil C / N ratio data in the data set to obtain a correlation diagram between chemical properties and grazing intensity (e.g. Figure 3 As shown in the figure, correlation analysis was performed to obtain the relationship coefficients between five soil chemical properties and grazing intensity. The autocorrelation function (ACF) reflects the correlation between the values ​​of the same sequence at different time series. Formula:

[0060]

[0061] The numerator is used to calculate the covariance matrix, and the denominator is used to calculate the sample variance. The ACF value is [-1,1]. According to the correlation diagram, the trend and cycle of the change can be analyzed, and the jump point and inflection point can be found. If the jump point is the correct observation value, that is, it exists in the concentrated area, it should be taken into account when modeling. If there is an inflection point, different models must be used to fit the time series in segments when modeling, such as using a threshold regression model.

[0062] Step 3: First, based on the soil carbon and nitrogen monitoring data sets of different grazing intensities collected from 2012 to 2020, a mathematical model is established to obtain the contents of five soil chemical properties in 2022. When calculating the content of chemical properties, the influence of factors such as precipitation, grazing intensity, and vegetation coverage should be considered. Calculate the soil organic carbon content at the next time, the soil organic carbon content at the next moment = the soil organic carbon content at this moment + the soil organic carbon content supplemented under external factors - the consumed soil organic carbon content. The consumption value of the soil organic carbon content mainly considers the vegetation coverage rate, and the supplementary value of the soil organic carbon content is mainly considered to be related to precipitation, plant wilting rate, and biological excretion, and then further establishes the calculation formula for the soil organic carbon content. The contents of five soil chemical properties in 2022 are shown in the following table:

[0063]

[0064] And the established formula for the influence of grazing strategies on the chemical properties of grassland soil is used, Q=C1*k1+C2*k2+C3*k3+C4*k4+C5*k5.

[0065] Among them, C1 is the soil organic carbon content, C2 is the soil inorganic carbon content, C3 is the soil total carbon content, C4 is the soil total nitrogen content, C5 is the soil C / N ratio, k1 is the relationship coefficient between soil organic carbon content and grazing intensity, k2 is the relationship coefficient between soil inorganic carbon content and grazing intensity, k3 is the relationship coefficient between soil total carbon content and grazing intensity, k4 is the relationship coefficient between soil total nitrogen content and grazing intensity, k5 is the relationship coefficient between soil C / N ratio and grazing intensity. Through the above contents and relationship coefficients, it can be concluded that Q is the influence of grazing intensity on the chemical properties of grassland soil.

[0066] Step 4: Establish an intelligent optimization model for grazing strategy and perform curve fitting, that is, use a general random model to fit the observed data of the time series, such as soil carbon and nitrogen monitoring data at different grazing intensities. The core idea is to continuously correct the weights and thresholds of the neural network through the sample training set, and gradually approach the expected output value. At the beginning of training, propagate along the network forward, and then adjust the weights and thresholds according to the error between the network output value and the expected output value through back propagation. The network training is completed by repeatedly updating the network weights and thresholds to minimize the error. Five chemical properties of soil organic carbon, soil inorganic carbon, soil total carbon, soil total nitrogen, soil C / N ratio and Q value are used as input, and grazing intensity is used as output. In the process of network design, it is very important to determine the number of hidden layer neurons. Too many hidden layer neurons will increase the amount of network calculation and easily cause overfitting problems; too few neurons will affect the network performance and fail to achieve the expected effect. The number of hidden layer neurons in the network is directly related to the complexity of the actual problem, the number of neurons in the input and output layers, and the setting of the expected error. At present, there is no clear formula for determining the number of neurons in the hidden layer. In order to ensure accuracy, this example was debugged repeatedly and the number of hidden neurons was finally determined to be l = 10. Finally, the established prediction model was used to predict the soil organic carbon, inorganic carbon, total N, soil C / N ratio and other values ​​in the same period in 2022, and the soil chemical properties data of different grazing areas under four different grazing intensities were obtained (the results are shown in the table below), making grassland animal husbandry operations more rational.

[0067]

[0068] The above specifically describes the specific implementation of the present invention, but it should not be considered that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention and the protection scope defined by the claims, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. An intelligent optimization method for grazing strategies with different soil chemical properties, characterized in that: include: Step 1, collecting soil carbon and nitrogen monitoring data of different grazing intensities within a grassland T time, including five soil chemical property data of soil organic carbon content, soil inorganic carbon content, soil total carbon content, soil total nitrogen content and soil C / N ratio at different grazing intensities; the grazing intensity is divided into four categories: control group, mild, moderate and severe; Step 2, making a correlation diagram between five soil chemical properties and grazing intensity based on the soil chemical property data, and obtaining the correlation coefficients between the five soil chemical properties and grazing intensity; Step 3, predicting the soil chemical property data of the next cycle, and combining the relationship coefficient to obtain the impact value Q of different grazing intensities on soil chemical properties; Step 4, constructing a time series model, taking the soil chemical property data of the next cycle and the impact value Q as input, outputting the grazing intensity of the next cycle, and adjusting the grazing strategy according to the output result, wherein the grazing strategy is characterized by the carrying capacity per unit area; Wherein, in step 3, by analyzing the material cycle of the ecosystem, the calculation formula of the impact value Q is as follows: Q=C1*k1+C2*k2+C3*k3+C4*k4+C5*k5 Among them, C1~C5 predict the soil chemical property data of the next cycle, k1~k5 are the relationship coefficients obtained through the correlation diagram, C1 is the soil organic carbon content, C2 is the soil inorganic carbon content, C3 is the soil total carbon content, C4 is the soil total nitrogen content, C5 is the soil C / N ratio, k1 is the relationship coefficient between soil organic carbon content and grazing intensity, k2 is the relationship coefficient between soil inorganic carbon content and grazing intensity, k3 is the relationship coefficient between soil total carbon content and grazing intensity, k4 is the relationship coefficient between soil total nitrogen content and grazing intensity, k5 is the relationship coefficient between soil C / N ratio and grazing intensity; The soil chemical property data prediction formula for the next cycle is as follows: X t+1 =X t +X 补 -X 耗 Among them, X is C1~C5, X t+1 Represents the five soil chemical properties data for the next cycle, X t Represents five soil chemical properties data of the current cycle, X 补 represents the supplementary value of a soil chemical property data at time t~t+1, X 耗 Represents the consumption value of a certain soil chemical property data at time t~t+1; in: X 耗 =h1*C pt X 补 =h2*P t +h3*D t +h4*Po t Among them, h1 is the coefficient of vegetation coverage, C pt is the vegetation coverage at time t, h2 is the coefficient of precipitation, P t is the precipitation at time t, h3 is the coefficient of plant wilting rate, D t is the plant withering rate at time t, h4 is the coefficient of biological excretion, Po t is the biological excretion at time t; Po t =1.8μ*M Among them, β t is the soil moisture content at time t, w * is the characteristic value of the grassland, μ is the livestock metabolic coefficient, and M is the livestock carrying capacity per unit area.

2. The intelligent optimization method for grazing strategies based on different soil chemical properties according to claim 1, characterized in that: In the step 1, T is measured in years, the collection period is T1, T1 is 2 years, and the obtained soil chemical property data is time series data.

3. The intelligent optimization method for grazing strategies based on different soil chemical properties according to claim 1, characterized in that: Through the correlation diagram of the five soil chemical properties and grazing intensity, jump points and inflection points are found. The jump point refers to an observation value that is inconsistent with other data. If a jump point is out of the concentrated area, the jump point is adjusted to the expected value; the inflection point refers to the point where the time series suddenly changes from an upward trend to a downward trend. If there is an inflection point, different models are used in step 4 to fit the time series in segments.

4. The intelligent optimization method for grazing strategies based on different soil chemical properties according to claim 1, characterized in that: In step 4, the time series model is an autoregressive moving average model, and the autoregressive moving average process is expressed as follows: X t represents observation data; {Z t }~WN(0,σ 2 ) represents a random variable sequence with a mean of zero and no correlation between the two. The mean square error of this random variable sequence actually represents the error in the measurement of chemical properties. is the autoregressive coefficient, p is the autoregressive order; (θ1, θ2, …, θ q ) is the sliding average coefficient, and q is the sliding average order.

5. The intelligent optimization method for grazing strategies based on different soil chemical properties according to claim 1, characterized in that: Grazing strategy optimization means that after obtaining different grazing intensity results in different areas, the two most recent grazing intensities and five soil content data of the same area are compared. If the soil chemical properties data are higher than those of the previous year under light grazing intensity, it means that the area can accept light grazing or even adjust to a stronger grazing intensity, and the adjustment of the grazing strategy can be considered next year; conversely, if the data is lower than that of the previous year under light grazing intensity, it means that it is necessary to consider not grazing the area next year.

6. The intelligent optimization method for grazing strategies based on different soil chemical properties according to claim 1, characterized in that: The control group had a grazing intensity of 0 livestock / day / hm 2 ; The light grazing intensity is 2 animals / day / hm 2 ; The moderate grazing intensity is 4 animals / day / hm 2 ; The grazing intensity is 8 animals / day / hm 2 .

Images