1. Overview of the study area
The study area is located in Kagadao Township, Cooperative City, Gannan Autonomous Prefecture, Gansu Province (Fig. 1), which lies in the transition zone between the Tibetan Plateau, the Loess Plateau and the Longnan Mountains, with the plateau being high in the north and south and low in the center, and the urban area is in the center of the Cooperative Basin. The average altitude is 2788 meters. It belongs to the humid alpine climate with a long cold season and a short warm season, with an average annual temperature of minus 0.5℃-3.5℃, an extreme maximum temperature of 28℃ and an extreme minimum temperature of -23℃; an average annual precipitation is 545 mm and is concentrated in July to September; the average frost-free period is 48 days and there is abundant sunshine all year round with a high proportion of solar energy utilization, and the terrain has three types of terrain, namely, alpine, mountainous and hilly areas. The terrain has three types of terrain, namely, high mountainous area, mountainous lowland area and mountainous hilly area.
2. Field investigation
2.1.Spectral Data Acquisition
Plant community spectral data was collected with a FieldSpec® 4 Hi-ResASD portable soil spectrometer from ASD (Analytica Spectra Devices. Inc.) in 2022 during the flowering period. Its wavelength is 350-2500 nm, including all visible light (380-780 nm) and part of the near-infrared light (780-2500 nm). The wavelengths of the spectral resolution were 3 nm for 350-1000 nm and 10 nm for 1000-2500 nm, with sampling intervals of 1. 4 nm and 2. 5 nm, respectively, and a resampling interval of 1 nm for data output (Table 1).
Table 1 Instrument parameters and requirements
Instrument name
|
Spectral band
|
Test time
|
Weather requirements
|
Assay requirements
|
FieldSpec® 4 Hi-ResASD
|
300~2500 nm
|
10: 00~14: 00
|
Dry, windless, clear and cloudless
|
Whiteboard correction is carried out in time after measurement
|
2.2.Sample plot selection and ground survey
Three sample plots were established in the study area, in each of which the age of the molehills was determined about soil properties, vegetation cover and species composition and categorized into four successional stages (Table 2): one-year-old molehills (ZM0), two-year-old molehills (ZM1), three-year-old molehills (ZM2) and perennial molehills (ZMM). Four molehills were selected at each successional stage. The molehills could be roughly considered egg-shaped, with the long axis × short axis of the molehills in each year ranging from 45 cm × 45 cm-65 cm × 68 cm and the sample quadrat of the molehills was 50 cm × 50 cm32. A total of 16 molehills were selected and the normal grassland vegetation adjacent to the molehills at each successional stage was used as a control group (CK). Their aboveground biomass, height, cover and abundance were analyzed. At the same time, the spectral data of the grassland vegetation of molehills and the CK group were determined using a portable ground spectrometer.
Table 2 Age division of rat mounds
|
Age of the mound
|
Character of mound
|
State of vegetation
|
1
|
Mounds through a winter
|
Less compaction soil
|
Total cover is more than 20%, and annual plants account for more than 90% of the total
|
2
|
Mounds through the second winter
|
More compaction soil
|
Total cover is more than 60%, and perennial Gramineae plants account for less than 15% of the total
|
3
|
Mounds through the third winter
|
Compaction soil
|
Total cover is more than 75%, and annual plants account for less than 10% of the total
|
4
|
Mounds through the fourth winter or more
|
Compaction soil
|
The total cover is more than 85%
|
5
|
Natural vegetations
|
Compaction soil
|
The total cover is more than 90%
|
3.Data processing
Calculate the important value (IV)33, Patrick34(hereafter simplified as PK), Shannon Diversity Index35(hereafter simplified as H), and Pielou34(hereafter simplified as P).
IV = (relative cover + relative height + relative biomass)/3
Richness index (Patrick), i.e., the total number of species occurring within the sample plot.
![](data:image/png;base64,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)
where S is the number of all species in the sample; Pi is the relative importance value of species.
The plant spectral data were preprocessed using View Spec Pro software of the feature spectrometer to remove data with large differences in the spectra of the same plant feature. GABS was obtained using 3 because the spectral curves were greatly affected by the external environment and noise36.
![](data:image/png;base64,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)
where R is the original reflectivity and λ is the wavelength (nm).
The acquired REF, ABS, GREF, and GABS data were comparatively analyzed to find the characteristic wavelength bands.
4.Vegetation index selection
Spectral measurements are susceptible to interference from external influences, such as the soil background, atmospheric aerosols, vegetation canopy, etc., but RVI and VARI are able to effectively correct for the effects of atmospheric aerosols and also remove some radiometric errors. NDVI670 and NDVI705 are very sensitive to canopy structure, PSRI can be used to detect and monitor vegetation health, GNDVI is more sensitive to canopy greenness, PRI is more sensitive to plant carotenoids, NDWI is more sensitive to canopy moisture, OSAVI can eliminate some soil background effects, and VARI can correct most macroaerosol effects. By selecting these vegetation indices, the spectral accuracy can be improved and some of the radiometric errorscan be eliminated37 and 20 vegetation indices can be selected38(Table 3).
Table 3 Spectral Index
Vegetation indices
|
Name
|
Calculation formula
|
GI38
|
Greenness vegetation index
|
R554/R667
|
ARVI37
|
Atmospheric impedance vegetation index
|
(R810-(2R680-R480))/(R810+(2R680-R480)
|
VARI37
|
Visual barometric impedance index
|
(R555-R680)/(R555+R680-R480)
|
NDVI70537
|
Normalized vegetation index 705
|
(R750-R705)/(R750+R705)
|
MSR70537
|
Improved red edge ratio vegetation index
|
(R705-R445)/(R705+R445)
|
NDVI67037
|
Normalized vegetation index 670
|
(R800/R670)/(R800+R670)
|
CI38
|
Chlorophyll index
|
(R750/R705)-1
|
PSRI37
|
Vegetation attenuation index
|
(R680-R500)/R750
|
RGI38
|
Relative green index
|
R690/R550
|
EVI37
|
Enhanced vegetation index
|
(2. 5×(R782-R675)/(R782+6R675-7. 5R445+1)
|
RVI37
|
Vegetation Index
|
R800/R760
|
SAVI37
|
Soil-adjusted vegetation index
|
1. 5×((R800-R760)/(R800+R760+0. 5))
|
MCARI37
|
Chlorophyll absorption correction index
|
((R700-R670)-0. 2×(R700-R550)) ×(R700/R670)
|
TCARI37
|
Improved chlorophyll absorption vegetation index
|
3×((R700-R670)-0. 2×(R700-R550)) ×(R700/R670)
|
OSAVI38
|
Soil-regulated vegetation index
|
1. 16×((R800-R670)/(R800+R670+0. 16))
|
CIrededge37
|
Red edge chlorophyll index
|
(R700-R800)/(R690-R720)-1
|
PRI39
|
Photochemical vegetation index
|
(R570-R531)/(R570+R531)
|
CRI39
|
Carotenoid reflex index
|
(1/R510) - (1/R550)
|
ANT39
|
Anthocyanin reflection index
|
((1/R550) - (1/R700))
|
SR29
|
The NIR/R ratio is called the simple ratio
|
R900/R680
|
5.Data analysis
Data was analyzed using R language, Excel 2016, and SPSS. Random Forest (RF) is a combinatorial classification method obtained by integrating K decision trees based on K decision trees. When the samples to be classified are input, the classification result of the Random Forest output is determined by a simple vote on the classification result of each decision tree40.
6. Results and analysis
6.1.Vegetation characteristics of molehills at different successional stages
The results of the survey on the cover, height, and biomass of the sample plots of molehills at different successional stages are shown in Table 4.
Table 4 Plant and cover survey of sample plots
Vegetation composition
|
Plant Statistics
|
ZM0
|
ZM1
|
ZM2
|
ZMM
|
Coverage/%
|
21. 62
|
60. 54
|
72. 12
|
88. 73
|
1,2 year Old plants
|
whitefly
Lappula myosotis V. Wolf
|
Whitefly
Lappula myosotis V. Wolf
|
Whitefly
Lappula myosotis V. Wolf
|
Whitefly
Lappula myosotis V. Wolf
|
microporous grass (botany)
Microula sikkimensis Hems
|
microporous grass (botany)
Microula sikkimensis Hems
|
chard (Beta vulgaris), a foliage beet
Potentilla chinensis Ser
|
microporous grass (botany)
Microula sikkimensis Hems
|
chard (Beta vulgaris), a foliage beet
Potentilla chinensis Ser
|
chard (Beta vulgaris), a foliage beet
Potentilla chinensis Ser
|
Rice hip celery
Ammi majus L
|
chard (Beta vulgaris), a foliage beet
Potentilla chinensis Ser
|
|
Amaranth Artemisia
Portulaca oleracea L
|
|
Amaranth Artemisia
Portulaca oleracea L.
|
|
Rice hip celery
Ammi majus L
|
|
Rice hip celery
Ammi majus L
|
|
|
|
Plantain
Plantago asiatica L
|
Perennial herbaceous plant
|
Gentian
Gentiana scabra Bunge
|
Gentian
Gentiana scabra Bunge
|
Gentian
Gentiana scabra Bunge
|
Gentian
Gentiana scabra Bunge
|
Buttercup
Ranunculus japonicus Thunb
|
Buttercup
Ranunculus japonicus Thunb
|
Buttercup
Ranunculus japonicus Thunb
|
Buttercup
Ranunculus japonicus Thunb
|
Lris
Iris tectorum Maxim
|
Lris
Iris tectorum Maxim
|
Lris
Iris tectorum Maxim
|
Lris
Iris tectorum Maxim
|
Flat beansMelissitusruthenicus(L.)Peschkoua
|
Echinococcus beans
Oxytropis DC
|
Echinococcus beans
Oxytropis DC
|
Echinococcus beans
Oxytropis DC
|
Beaded tooth
Polygonum viviparum Linn
|
Flat beansMelissitusruthenicusPeschkoua
|
Artemisia cold
Artemisia frigida Willd. Sp. Pl
|
Artemisia cold
Artemisia frigida Willd. Sp. Pl.
|
Gentiana
G. macrophylla
|
Beaded toothPolygonum viviparum Linn
|
Flat beansMelissitusruthenicus Peschkoua
|
Flat beansMelissitusruthenicus(L.)Peschkoua
|
Siberian butterfly
Polygonum sibiricum Laxm
|
Gentiana
G. macrophylla
|
Beaded tooth
Polygonum viviparum Linn.
|
Beaded tooth
Polygonum viviparum Linn
|
|
Siberian butterfly
Polygonum sibiricum Laxm
|
Gentiana
G. macrophylla
|
Gentiana
G. macrophylla
|
|
Dandelion
Taraxacum mongolicum
|
Siberian butterfly
Polygonum sibiricum Laxm.
|
Dandelion
Taraxacum mongolicum
|
|
Dwarf grass
Kobresia humilis Sergievskaya
|
Dwarf grass
Kobresia humilis Sergievskaya
|
Dwarf grass
Kobresia humilisSergievskaya
|
|
Arrowroots
Carumcarvi
|
Arrowroots
Carumcarvi
|
Arrowroots
Carumcarvi
|
|
Anemone
Anemone catharsis Kitag
|
Anemone
Anemone catharsis Kitag
|
Anemone
Anemone catharsis Kitag
|
|
|
Oriental strawberry
Fragaria orientalis Losinsk
|
|
Gramineae
|
|
Rough shadow grass
Cleistogenes squarrosa Keng
|
Draped grass
Elymus dahuricus Turcz
|
Precocious grass
Poa Annual
|
|
|
|
Precocious grass
Poa Annual
|
|
perennial weeds
|
Ask Jing
Equisetum arvenseL
|
Ask Jing
Equisetum arvenseL
|
Ask Jing
Equisetum arvenseL
|
Ask Jing
Equisetum arvenseL
|
Figure 2-A shows that the Pk of aboveground organisms in molehills increased continuously with the succession of years and was lowest at ZM0, and the differences between CK and ZMM, ZMM and ZM1, and ZM1 and ZM0 were significant (P<0.05); Figure 2-B shows that the H of aboveground organisms in molehills increased continuously and was lowest at ZM0, and the differences between CK and ZMM, ZMM and ZM1, and ZM1 and ZM0 were significant (P<0.05); Figure 2-C shows that the P of aboveground organisms of molehills first increased, then decreased and finally increased and was lowest at ZM0, and the difference between CK and ZM2 was significant (P<0.05); Figure 2-D shows that T of molehills increased steadily and was lowest at ZM0, and the differences between CK and ZMM, ZMM and ZM2, ZM2 and ZM1, ZMM and ZM1, and ZM1 and ZM0 were significant (P<0.05); Figure 2-E shows that B of molehills first increased, then decreased and finally increased and was lowest at ZM0, and the difference between CK and ZMM was significant (P<0.05).
6.2.Raw spectra of molehills at different successional stages
The original spectral reflectance of molehills at different successional stages (Fig. 3) showed that all molehills, with the exception of ZM0, exhibited varying degrees of "green peaks" at around 550 nm, in descending order ZMM>ZM2>CK>ZM1, with the first nadir occurring at 675 nm and a plant-specific red-edge phenomenon forming at 760 nm. The reflectance of the other molehills except ZM0 increased rapidly, and various molehills formed the plant-specific "red edge" phenomenon. Although the overall trend of this red edge phenomenon was not very different, there were still differences in REF between different treatments, and in the near red wavelength range (750-970 nm), REF increased slowly. The 2nd nadir of the NIR band appeared at 1 150 and 1 250 nm for ZM1 and CK, and the 1st nadir of the NIR band appeared for ZM0, ZM2 and ZMM.
6.3.Hyperspectral classification of successional stages of molehills
6.3.1Relevance analysis
The correlation analysis between vegetation-characteristic indicators and vegetation indices using the person correlation coefficient method (Fig. 4) showed that the vegetation indices NDVI670, NDVI705, EVI, TCARI and SR were highly significantly (P<0.01) were correlated with the vegetation-characteristic indicators Pk, H, B and T; the vegetation indices EVI and SR were highly significantly (P<0.01) correlated with the vegetation-characteristic indicators P NDVI670, NDVI705 and TCARI were significantly (P<0.05) correlated with the vegetation-characteristic index P.
6.3.2.Hyperspectral classification model
Multiple stepwise regression model
The results of modeling using the screened vegetation indices with strong correlation are shown in Table 5.
Table 5 Multiple stepwise regression modeling
Vegetation indicators
|
Modelling
|
R2
|
RMSE
|
Patrick
|
Y=20.039X(NDVI705)-18.139X(NDVI670)+24.190X(TCARI)-0.187X(SR)+9.635X(EVI)-0.976X(Ant)+5.844
|
0.388
|
1.888
|
Shamon.wiener
|
Y=12.559X(NDVI705)-9.603X(NDVI670)+17.918X(TCARI)-0.021X(SR)-1.480X(EVI)+0.029X(Ant)+1.967
|
0.482
|
0.284
|
Pielou
|
Y=3.652X(NDVI705)-2.769X(NDVI670)+4.790X(TCARI)+0.001X(SR)-0.780X(EVI)+0.048X(Ant)+1.09
|
0.342
|
0.049
|
Total. coverage
|
Y=405.711X(NDVI705)-319.057X(NDVI670)+2668.599X(TCARI)+4.102X(SR)-424.667X(EVI)-17.812X(Ant)+71.715
|
0.704
|
18.663
|
Biomass
|
Y=682.292X(NDVI705)-518.034X(NDVI670)+1285.846X(TCARI)+5.645X(SR)-314.295X(EVI)-8.066X(Ant)+71.183
|
0.477
|
24.151
|
The results of the model accuracy test are shown in Fig. 5, which shows that Pk (R2= 0.388, RMSE=1.888), H (R2= 0.482, RMSE= 0.284), P (R2= 0.338, RMSE= 0.049) T (R2= 0.704, RMSE= 18.663), B (R2= 0.4772, RMSE= 24.151), indicating that T has the best predictive ability for vegetation characterization indicators.
Random forest model
Random forest inversion modelling was performed for all spectral variables, ntree was set to 600 and mytry to 3. The accuracy of the random forest model for each indicator is shown in Table 6, in which T, B, Pk, H random forest model variables with a high explanatory rate meet the standard of model accuracy, and the variable explanatory rate is divided into 64.72%, 51.33, %, 46.6%, 47.23%, and Pielou index random forest model variance explained rate is poor (-9.06%) and does not meet the model accuracy criteria.
Table 6 Random forest model parameters
Vegetation indicators
|
The mean of squared residuals
|
%Var explained
|
R2
|
RMSE
|
Patrick
|
2.9786
|
46.6
|
0.4617
|
1.7258
|
Shamon.wiener
|
0.0776
|
47.23
|
0.474
|
0.2787
|
Pielou
|
0.0035
|
-9.06
|
0.069
|
0.0598
|
Total. coverage
|
339.9304
|
64.72
|
0.6485
|
18.4372
|
Biomass
|
475.728
|
51.33
|
0.5192
|
21.8112
|
The model test of the random forest model created was carried out with the test set (TEST) and the results are shown in Figure 6. The results of the linear fit test of the model for the predicted and true values show that Pk (R2=0.537, RMSE=0.823), H (R2=0.459, RMSE=0.143), T (R2=0.635, RMSE=24.287) and B (R2=0.948, RMSE=15.173) have good predictive power.
The results of the importance of the variables in the random forest model are shown in Figure 7, with TCARI, EVI, SR, NDVI705, NDVI670 and Ant in A in descending order of importance, with the TCARI variable having the highest importance and an IncMSE value of 17.79; in B with TCARI, EVI, NDVI670, SR, Ant and NDVI705, with the TCARI variable having the highest importance and an IncMSE value of 16.64; C with TCARI, EVI, SR, NDVI670, NDVI705 and Ant in descending order of importance, with the TCARI variable having the highest importance and an IncMSE value of 18.93; and D with SR, NDVI670, EVI, EVI, SR, NDVI670 and NDVI705 in descending order of importance, NDVI705, EVI, TCARI and Ant, with the SR variable having the highest importance and an IncMSE value of 15.48. The analysis revealed that TCARI plays an important role in the inversion of Pk, H and T and SR plays an important role in the inversion of B. The analysis also revealed that TCARI plays an important role in the inversion of Pk, H, a.
6.3.3.Classification of molehills
Classification of indicators of vegetation characteristics
The ntree error of the RF classification model shows that the error of the grass species within the model is stable when ntree=600, i.e. ntree is assumed to be 600. Fig. 8 shows the importance diagram of the variables of the RF model. The out-of-bag error OOB is 36.96%, and the importance indicators of the RF model, from the largest to the smallest, are B, T, H and Pk, respectively. The Gini coefficients of the RF model, from largest to smallest, are B, T, H and Pk. B is the variable with the highest importance indicator and Gini coefficient.
The results of modeling using vegetation characteristic indicators are shown in Table 7, where ZM1 was misclassified as ZM0 (50%), ZM0 and ZMM were misclassified as ZM1 (18.75%), ZM0, ZM1, and ZMM were misclassified as ZM2 (100%), ZM1 and ZM2 were misclassified as ZMM (50%), and ZMM was misclassified as CK (11.10%).
Table 7 Classification and error rate of aboveground plants in Rat Mound in different years
|
ZM0
|
ZM1
|
ZM2
|
ZMM
|
CK
|
Error rate
|
ZM0
|
4
|
4
|
0
|
0
|
0
|
0.5
|
ZM1
|
2
|
13
|
0
|
1
|
0
|
0.1875
|
ZM2
|
0
|
2
|
0
|
3
|
0
|
1
|
ZMM
|
0
|
1
|
3
|
4
|
0
|
0.5
|
CK
|
0
|
0
|
0
|
1
|
8
|
0.110
|
Vegetation index classification
The ntree error of the RF classification model shows that the error of grass species in the model is stable when ntree=600, i.e., ntree is taken as 600. Fig. 9 shows the importance of the variables in the RF model, and the out-of-bag error OOB is 23.91%, and the importance indicators of the RF model, from largest to smallest, are NDVI670, SR, TCARI, EVI, and NDVI705. The Gini coefficients of the RF model, from largest to smallest, are TCARI, NDVI670, SR, EVI, and NDVI705, respectively. Gini coefficients from largest to smallest are TCARI, NDVI670, SR, EVI, and NDVI705.SR and TCARI are the variables with the highest importance indicators and Gini coefficients.
The results of vegetation index modeling are shown in (Table 8).ZM1 was misclassified as ZM0 (50.0%), ZM0 and ZM2 were misclassified as ZM1 (12.5%), ZM1 and ZMM were misclassified as ZM2 (40.0%), ZM2 and CK were misclassified as ZMM (37.5%), and there was no misclassification in CK.
Table 8 Classification and error rate of Rat Mound spectra in different years
|
ZM0
|
ZM1
|
ZM2
|
ZMM
|
CK
|
Error rate
|
ZM0
|
4
|
4
|
0
|
0
|
0
|
0.5
|
ZM1
|
1
|
14
|
1
|
0
|
0
|
0.125
|
ZM2
|
0
|
1
|
3
|
1
|
0
|
0.4
|
ZMM
|
0
|
0
|
2
|
5
|
1
|
0.375
|
CK
|
0
|
0
|
0
|
0
|
9
|
0
|