A method for evaluating and optimizing the carrying capacity of alpine grasslands based on hyperspectral remote sensing

Abstract

A method for evaluating and optimizing the carrying capacity of alpine grasslands based on hyperspectral remote sensing, comprising: evaluating the nutritional value corresponding to the quality indicators of alpine forage (approximate nutrient indicators), calculating the ratio of the nutritional value of alpine forage to the nutritional value of standard forage, and using this ratio to optimize the carrying capacity of alpine grasslands, so that the carrying capacity of alpine grasslands optimized based on quality indicators has a higher confidence level. First, select the optimal forage yield and quality parameter inversion model; then determine the formula parameters according to industry standards and actual conditions, and calculate the theoretical carrying capacity of alpine grasslands in the study area based on the optimal forage yield modeling results and formula parameters; finally, correct the carrying capacity of alpine grasslands based on the optimal forage quality modeling results and approximate nutrient parameters, with the spatial resolution of hyperspectral images as the calculation unit, and determine it as the optimized carrying capacity of alpine grasslands. The present invention provides an effective solution for the evaluation and optimization of the carrying capacity of alpine grasslands, and provides reference for the livestock quantity and pastoral area configuration.

Description

Technical Field

[0001] The present invention belongs to the field of hyperspectral remote sensing, and particularly relates to a method for evaluating and optimizing the carrying capacity of alpine grasslands based on hyperspectral remote sensing. Background Art

[0002] Alpine grasslands are ecologically fragile and sensitive. Overgrazing causes grassland ecological degradation and threatens ecological security. Conducting an evaluation of the carrying capacity of alpine grasslands is the basis for regulating livestock numbers and selecting pastoral areas, and is of great significance for ensuring the sustainable development of animal husbandry and ecological security. Currently, most grassland carrying capacity estimation methods are based on yield indicators such as forage yield (Wu Dan, 2015), net primary productivity (NPP) (Wang Qi et al., 2019), and normalized difference vegetation index (NDVI) (Ren et al., 2021), but ignore the importance of quality indicators for grassland carrying capacity and cannot effectively and objectively evaluate the health status of the actual grassland ecosystem. Compared with ordinary forage, high-quality forage has a higher crude protein (CP) content, lower acid detergent fiber (ADF) and neutral detergent fiber (NDF), soft texture, good grazing tolerance, and higher feeding value (Chen Gu et al., 2010).

[0003] With the development of hyperspectral remote sensing technology, due to its advantages of high spectral resolution, wide wavelength coverage range, integration of spectrum and image, and continuous imaging, the ability to distinguish and identify ground objects has been greatly improved, and the number of imaging channels has increased significantly. This makes it possible to accurately invert grassland yield indicators and quality indicators (approximate nutrient parameters) (Gao Jinlong, 2020), and also brings the possibility of optimizing yield based on forage quality. Therefore, compared with the grassland carrying capacity estimated directly using yield indicators, the grassland carrying capacity optimized by quality indicators is more reasonable. Since different approximate nutrients in forage affect the digestible energy to different degrees, and the number of approximate nutrient parameters is very large, how to determine the optimal number and values of approximate nutrient parameters through method design to accurately and reliably estimate the carrying capacity of alpine grasslands is a problem worthy of study.

[0004] The references in the background art section are as follows:

[0005] [1] Chen Gu, Tai Jianhui. Forage Quality and Quality Standards in Commercial Applications in the United States [J]. China Animal Husbandry Bulletin, 2010, 48 - 49.

[0006] [2] Gao Jinlong, Liu Yin, Ge Hou, Feng Liang. Research Progress on Hyperspectral Remote Sensing of Forage Nutritional Quality in Natural Grasslands [J]. Acta Prataculturae Sinica, 2020, 29: 172 - 185.

[0007] [3] Wang Qi, Wu Chengyong, Chen Kelong, Zhang Xiao, Zhang Lele, Ding Junxia. Estimation of Grass Yield and Carrying Capacity in the Qinghai Lake Basin Based on MODIS NPP Data [J]. Ecological Science, 2019, 38: 178-185.

[0008] [4] Wu Dan. Research on the Causes and Carrying Capacity of Guanshan Grassland in Long County [M]. Northwest A&F University. 2015. Summary of the Invention

[0009] In view of the deficiencies of the existing methods for estimating the carrying capacity of alpine grasslands, the present invention provides an evaluation and optimization method for the carrying capacity of alpine grasslands based on hyperspectral remote sensing, which solves the problem of how to optimize the yield using forage quality indicators, and determines the optimal number and values of approximate nutrient parameters through method design to accurately and reliably estimate the carrying capacity of alpine grasslands.

[0010] The technical solution adopted by the present invention is: an evaluation and optimization method for the carrying capacity of alpine grasslands based on hyperspectral remote sensing, comprising the following steps:

[0011] Step 1: Select the optimal forage yield inversion model Y(X), where X is the inversion parameter and Y is the grass yield;

[0012] Step 2: According to the agricultural industry standard "Calculation of the Rational Carrying Capacity of Natural Grasslands" (NY / T 635–2015) and in combination with the actual situation of the research area, determine the grassland-animal husbandry parameters, where the grassland-animal husbandry parameters include the reasonable utilization rate G of the grassland, the proportion E of edible forage in the grassland, the standard sheep unit I, and the grassland grazing time D;

[0013] Step 3: Based on the optimal modeling results of the inversion parameter X and the grassland-animal husbandry parameters determined in Step 2, according to the agricultural industry standard "Calculation of the Rational Carrying Capacity of Natural Grasslands" (NY / T 635–2015), calculate the theoretical carrying capacity C of the alpine grassland for each grid cell in the research area, C = {C 1 , C 2 , C 3 ,..., C i}, where i represents the i-th grid cell, and its calculation formula is as follows:

[0014]

[0015] In the formula: C is the theoretical carrying capacity of the grassland, (SU·hm -2 ); Y is the optimized grass yield per unit area of the grassland (g·m -2 ); G is the annual grassland utilization rate (%); E is the proportion of edible forage in the grassland (%); I is the standard sheep unit; D is the grassland grazing time (d).

[0016] Step 4: Select the optimal forage quality inversion model*(X), where * is the forage approximate nutrient index, and the forage approximate nutrient index includes crude protein content CP, neutral detergent fiber content NDF, and acid detergent fiber content ADF, and X is the inversion parameter;

[0017] Step 5: Consult the reference literature and determine the un-inverted approximate nutrient parameters in combination with the actual situation of the research area, including 48-hour in vitro neutral detergent fiber digestibility NDFD and crude fat content EE;

[0018] Step 6: Based on the optimal forage quality modeling results of the inversion parameter X and the un-inverted approximate nutrient parameters determined in Step 5, calculate the relative feeding quality RFQ of alpine grassland for each grid cell in the research area. RFQ = {RFQ 1 , RFQ 2 , RFQ 3 , …, RFQ i}, where i represents the i-th grid cell, and its calculation formula is as follows:

[0019] RFQ = DMI × TDN / 1.23 (2)

[0020]

[0021] In the formula: CP is the crude protein content (%); NDF is the neutral detergent fiber content (%); ADF is the acid detergent fiber content (%); RFQ is the relative feeding quality (%); TDN is the total digestible nutrients (%); DMI is the voluntary intake of dry matter of roughage (%); NDFn is the nitrogen-free neutral detergent fiber content (%), and NDFn = NDF × 0.93; NDFD is the 48-hour in vitro neutral detergent fiber digestibility (%); NFC is the non-fiber carbohydrate content (%), and NFC = 100 - (NDFn + CP + EE + ash), where ash is the crude ash content (%); EE is the crude fat content (%).

[0022] Step 7: Based on the dominant grass species type in the grassland of the research area, take it as the standard forage type, assume that its approximate nutrient parameters are standard fixed values, and calculate the relative feeding quality RFQ of the standard forage according to Step 6 s ;

[0023] Step 8: Calculate the ratio R i of the relative feeding quality RFQ s of each grid cell to the relative feeding quality RFQ i of the standard forage;

[0024] R i = RFQ i / RFQ s (5)

[0025] Step 9: Use the ratio R i to correct the theoretical carrying capacity C of each grid cell of the alpine grassland i to obtain the optimized theoretical carrying capacity C of the alpine grassland, oi and its calculation formula is as follows:

[0026] C oi = C i × R i (6)

[0027] Step 10: Repeat Steps 8 - 9 until all grid cells in the hyperspectral remote sensing image are calculated;

[0028] Step 11: Output the optimized theoretical carrying capacity C of each grid cell of the alpine grassland oi The optimized theoretical carrying capacity of the alpine grassland is C o = {C o1 ,C o2 ,C o3 ,…,C oi}.

[0029] Furthermore, the selection range of the optimal forage yield inversion model Y(X) includes: stepwise regression, partial least squares method, principal component analysis, physical model, neural network, random forest.

[0030] Furthermore, the selection range of the optimal forage quality inversion model *(X) includes: stepwise regression, partial least squares method, principal component analysis, physical model, neural network, random forest.

[0031] Furthermore, when selecting the model Y(X) or *(X) in Step 1 or Step 4, based on the actual laboratory analysis data after sampling the alpine grassland, referring to relevant scientific research papers and other literature, the optimal inversion model is determined according to the correlation coefficient R 2 and the root mean square error RMSE.

[0032] Furthermore, the selection range of the inversion parameter X of the model in Step 1 or Step 4 includes: spectral feature variables and vegetation indices. The spectral feature variables include band reflectance, higher - order differential of reflectance, absorption or reflection position, absorption or reflection depth, absorption or reflection width, and absorption or reflection symmetry; the vegetation indices include normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), difference vegetation index (DVI), ratio vegetation index (RVI), transformed vegetation index (TVI), modified soil - adjusted vegetation index (MSAVI), optimized soil - adjusted vegetation index (OSAVI).

[0033] Further, in step 6, the relative forage quality (RFQ) is based on the concept of total digestible nutrients (TDN) relative to a standard forage. Forages with a higher RFQ can provide more energy relative to the standard forage. The default standard forage is alfalfa in full bloom, and its relative forage quality RFQ is 100.

[0034] Further, in step 7, according to the differences in dominant species in the study area, the type of standard forage should be referenced based on the dominant species, and its relative forage quality RFQ is no longer 100 and needs to be recalculated.

[0035] The beneficial effects of the present invention are as follows:

[0036] Compared with the prior art, the proposed method combines industry standards. Based on the concept that the relative forage quality (RFQ) of forage is relative to the total digestible nutrients (TDN) of the standard forage, the optimal number and values of approximate nutrient parameters are determined, optimizing the grassland carrying capacity calculated only based on forage yield, solving the difficulty of formulating and correcting the grassland carrying capacity using forage quality indicators, and achieving the goal of being closer to the actual situation of the grassland. Description of the Drawings

[0037] Figure 1 is a flowchart of a method for evaluating and optimizing the carrying capacity of alpine grasslands based on hyperspectral remote sensing according to an embodiment of the present invention;

[0038] Figure 2 is a schematic diagram of the carrying capacity of alpine grasslands before optimization in an embodiment of the present invention;

[0039] Figure 3 is a schematic diagram of the carrying capacity of alpine grasslands after optimization in an embodiment of the present invention. Detailed Embodiments

[0040] To facilitate understanding and implementation of the present invention by those of ordinary skill in the art, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the embodiments described herein are only used to illustrate and explain the present invention and are not used to limit the present invention.

[0041] As Figure 1 shown, an embodiment of the present invention proposes a method for evaluating and optimizing the carrying capacity of alpine grasslands based on hyperspectral remote sensing. The technical solution provided by the present invention will be described in detail below for one embodiment. The embodiment is an evaluation and optimization of the carrying capacity of alpine grasslands, but the protection scope of the present invention is not limited to the described embodiment.

[0042] Embodiment:

[0043] Step 1: Select the multiple linear regression model Y(X) as the optimal forage yield inversion model, where X is the inversion parameter, and its functional expression is as follows:

[0044] Y = 379.747 EVI - 561.583 ρ1396 + 104.802 PC3 + 17.824

[0045] In the formula: EVI represents the enhanced vegetation index, ρ1396 represents the spectral reflectance of the band with a central wavelength of 1396 nm, and PC3 represents the third principal component in the principal component analysis of all bands of the hyperspectral image.

[0046] Step 2: According to the agricultural industry standard "Calculation of the reasonable stocking rate of natural grasslands" (NY / T 635–2015), combined with the actual situation of the research area, the reasonable utilization rate G of the grassland is taken as 50%, the proportion E of edible forage in the grassland is taken as 90%, the standard sheep unit I is the daily food intake of each sheep unit livestock, and 1.8 kg of standard hay with a water content of 14% is converted into completely water-free hay, which is 1.548 kg. The grassland grazing time D is taken as 365 days;

[0047] Step 3: Based on the optimal modeling result (forage yield Y) of the inversion parameter X and the grassland - livestock parameters determined in Step 2, according to the agricultural industry standard "Calculation of the reasonable stocking rate of natural grasslands" (NY / T 635–2015), calculate the theoretical carrying capacity C of the alpine grassland for each grid cell in the research area. C = {C 1 ,C 2 ,C 3 ,...,C i}, where i represents the i - th grid cell, and its calculation formula is as follows:

[0048]

[0049] In the formula: C is the theoretical carrying capacity of the grassland (SU·hm -2 ); Y is the optimized forage yield per unit area of the grassland (g·m -2 ); G is the annual grassland utilization rate (%); E is the proportion of edible forage in the grassland (%); I is the standard sheep unit; D is the grassland grazing time (d).

[0050] Step 4: Select the multiple linear regression models CP(X), NDF(X), and ADF(X) as the optimal forage quality inversion models, where X is the inversion parameter, and their function expressions are as follows:

[0051] CP = 46.735 ρ1497 - 77.126 ρ733 - 99.366 ρ2516 + 326.639 ρ1918 + 21.464

[0052] NDF = 1414.260 ρ1884 - 112.185 ρ2347 + 114.522 ρ404 + 43.558

[0053] ADF = -494.601ρ1918 + 68.076ρ2415 + 34.942

[0054] Where: ρ* represents the spectral reflectance of the band with a central wavelength of * nm.

[0055] Step 5: Consult the references and, in combination with the actual situation of the study area, take 55% for the 48-hour in vitro neutral detergent fiber digestibility NDFD, 2% for the crude fat content EE, and 9% for the crude ash content ash;

[0056] Step 6: Based on the optimal forage quality modeling results (crude protein content CP, neutral detergent fiber content NDF, and acid detergent fiber content ADF) and the un-inverted approximate nutrient parameters determined in Step 5, calculate the relative feeding quality RFQ of alpine grasslands for each grid cell in the study area. RFQ = {RFQ 1 , RFQ 2 , RFQ 3 ,..., RFQ i}, where i represents the i-th grid cell, and its calculation formula is as follows:

[0057] RFQ = DMI × TDN / 1.23 (2)

[0058]

[0059] Where: CP is the crude protein content (%); NDF is the neutral detergent fiber content (%); ADF is the acid detergent fiber content (%); RFQ is the relative feeding quality (%); TDN is the total digestible nutrients (%); DMI is the voluntary intake of dry matter of roughage (%); NDFn is the nitrogen-free neutral detergent fiber content (%); NDFn = NDF × 0.93; NDFD is the 48-hour in vitro neutral detergent fiber digestibility (%); NFC is the non-fiber carbohydrate content (%); NFC = 100 - (NDFn + CP + EE + ash), where ash is the crude ash content (%); EE is the crude fat content (%);

[0060] Step 7: Based on the dominant grass species type in the grasslands of the study area, take it as the standard forage type, assume its approximate nutrient parameters as standard fixed values, and calculate the relative feeding quality RFQ of the standard forage according to Step 6 s = 125.30;

[0061] Step 8: Calculate the ratio R i of the relative feeding quality RFQ s of the forage (within the spatial resolution range of the hyperspectral remote sensing image) for each grid cell to the relative feeding quality RFQ i of the standard forage;

[0062] Ri = RFQ i / RFQ s (5)

[0063] Step 9: Use the ratio R i to optimize the theoretical carrying capacity C of alpine grassland within the spatial range of hyperspectral remote sensing image resolution for each grid cell i , and obtain the optimized theoretical carrying capacity C of alpine grassland oi , and its calculation formula is as follows:

[0064] C oi = C i ×R i (6)

[0065] Step 10: Repeat Steps 8 - 9 until all grid cells in the hyperspectral remote sensing image are calculated

[0066] Step 11: Output the optimized theoretical carrying capacity C of alpine grassland for each grid cell oi , and the optimized theoretical carrying capacity of alpine grassland is C o ={C o1 , C o2 , C o3 , …, C oi}.

[0067] To verify the effectiveness of the proposed method, hyperspectral remote sensing satellite images of the Dangqu River Basin are used, combined with field sampling and laboratory analysis data to verify the effectiveness of the proposed method. The results show that the average value of the theoretical carrying capacity C of alpine grassland calculated only based on forage yield in the study area is 0.51 SU·hm -2 (as Figure 2 shown), while the average value of the optimized theoretical carrying capacity of alpine grassland after optimizing with yield indicators is C o with an average value of 0.63 SU·hm -2 (as Figure 3 shown). The high values of the optimized theoretical carrying capacity C of alpine grassland are mainly concentrated in the southeastern side of the basin. The optimization method can improve the reasonable livestock carrying capacity of high-quality forage grass, which has important significance for realizing the rational utilization of alpine grassland, the sustainable development of animal husbandry and ecological environment protection. This shows that the alpine grassland carrying capacity obtained by the method proposed in this invention contains higher data value and is superior to previous estimation methods o .

[0068] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims

Claims

1. A method for evaluating and optimizing the carrying capacity of alpine grassland based on hyperspectral remote sensing, characterized in that: The following steps are involved: Step 1: Select the optimal forage yield inversion model Y(X), where X is the inversion parameter and Y is the forage yield; Step 2: According to the agricultural industry standard "Calculation of Reasonable Carrying Capacity of Natural Grassland" (NY / T635-2015), combined with the actual situation of the study area, determine the pasture parameters, which include reasonable grassland utilization rate G, proportion of edible grass in grassland E, standard sheep unit I and grassland grazing time D; Step 3: Based on the optimal modeling results of the inversion parameter X and the pasture parameters determined in step 2, the theoretical carrying capacity C of the alpine grassland of each grid unit in the study area is calculated according to the agricultural industry standard "Calculation of Reasonable Carrying Capacity of Natural Grassland" (NY / T635-2015), where C = {C1, C2, C3, ..., C i }, i represents the i-th grid unit, and its calculation formula is as follows: In the formula: C is the theoretical carrying capacity of grassland; Y is the optimized grass yield per unit area of ​​grassland; G is the annual grassland utilization rate; E is the proportion of edible grass in grassland; I is the standard sheep unit; D is the grassland grazing time; Step 4: Select the optimal forage quality inversion model *(X), where * is the approximate nutrient index of the forage, the approximate nutrient index of the forage includes crude protein content CP, neutral detergent fiber content NDF and acid detergent fiber content ADF, and X is the inversion parameter; Step 5: Consult references and determine the approximate nutrient parameters that have not been inverted based on the actual situation in the study area, including 48-hour in vitro neutral detergent fiber digestibility NDFD and crude fat content EE; Step 6: Based on the optimal forage quality modeling results of the inversion parameter X and the uninverted approximate nutrient parameters determined in step 5, calculate the relative forage quality RFQ of the alpine grassland in each grid cell in the study area, RFQ = {RFQ1, RFQ2, RFQ3, ..., RFQ i }, i represents the i-th grid unit, and its calculation formula is as follows: RFQ=DMI×TDN / 1.23 (2) Wherein: CP is crude protein content; NDF is neutral detergent fiber content; ADF is acid detergent fiber content; RFQ is relative feeding quality; TDN is total digestible nutrients; DMI is ad libitum intake of roughage dry matter; NDFn is nitrogen-free neutral detergent fiber content, NDFn = NDF × 0.93; NDFD is 48-hour in vitro neutral detergent fiber digestibility; NFC is non-fiber carbohydrate content, NFC = 100-(NDFn + CP + EE + ash), ash is crude ash content; EE is crude fat content; Step 7: Based on the dominant species of grassland in the study area, take it as the standard forage type, assume its approximate nutrient parameters as standard values, and calculate the relative feeding quality FRQ of the standard forage according to step 6 s ; Step 8: Calculate the relative forage quality (FRQ) of each grid cell i RFQ forage quality relative to standard forage s The ratio R i ; R i =RFQ i / RFQ s (5) Step 9: Using the Ratio R i Correct the theoretical carrying capacity of alpine grassland C in each grid unit i , and the optimized theoretical carrying capacity of alpine grassland C is obtained oi , and its calculation formula is as follows: C oi =C i ×R i (6) Step 10: Repeat steps 8 to 9 until all grid cells in the hyperspectral remote sensing image are calculated; Step 11: Output the theoretical carrying capacity C of the alpine grassland of each grid cell after optimization oi The theoretical carrying capacity of the alpine grassland after optimization is C o ={C o1 , C o2 , C o3 , …, C oi }.

2. The method for evaluating and optimizing the carrying capacity of alpine grassland based on hyperspectral remote sensing according to claim 1, characterized in that: The selection range of the optimal forage yield inversion model Y(X) includes: stepwise regression, partial least squares method, principal component analysis, physical model, neural network, and random forest.

3. The method for evaluating and optimizing the carrying capacity of alpine grassland based on hyperspectral remote sensing according to claim 1, characterized in that: The selection range of the optimal forage quality inversion model *(X) includes: stepwise regression, partial least squares method, principal component analysis, physical model, neural network, and random forest.

4. The method for evaluating and optimizing the carrying capacity of alpine grassland based on hyperspectral remote sensing according to claim 1, characterized in that: When model Y(X) or *(X) is selected in step 1 or step 4, the correlation coefficient R is calculated based on the actual laboratory analysis data after sampling of alpine grassland and reference to relevant scientific research papers and other literature. 2 , root mean square error RMSE determines the optimal inversion model.

5. The method for evaluating and optimizing the carrying capacity of alpine grassland based on hyperspectral remote sensing according to claim 1, characterized in that: The selection range of the inversion parameter X of the model in step 1 or step 4 includes: spectral characteristic variables and vegetation indexes, wherein the spectral characteristic variables include band reflectance, high-order differentials of reflectance, absorption or reflection position, absorption or reflection depth, absorption or reflection width, and absorption or reflection symmetry; the vegetation index includes the normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), difference vegetation index (DVI), ratio vegetation index (RVI), conversion vegetation index (TVI), modified soil adjusted vegetation index (MSAVI), and optimized soil adjusted vegetation index (OSAVI).

6. The method for evaluating and optimizing the carrying capacity of alpine grassland based on hyperspectral remote sensing according to claim 1, characterized in that: The relative feeding quality RFQ in step 6 is based on the concept of total digestible nutrients TDN relative to standard forage. Forage with higher relative feeding quality RFQ can provide more energy relative to standard forage. The default standard forage is alfalfa in full bloom, and its relative feeding quality RFQ is 100.

7. The method for evaluating and optimizing the carrying capacity of alpine grassland based on hyperspectral remote sensing according to claim 6, characterized in that: In step 7, according to the dominant species in the study area, the type of standard forage should be based on the dominant species, and its relative feeding quality RFQ is no longer 100 and needs to be recalculated.

Images