Mapping and Evaluating Plant Phenology in the Qinghai-Tibet Plateau: A Digital Approach using the Plant Phenological Index (PI)

A new plant Phenological Index (PI) has been developed based on the visibility of plants, communities, and ecosystems. An evaluation and assessment of Zhang's Phenological Index (PI) were further examined based on the �eld observations of Alpine Shrub and Alpine Meadow vegetation at Haibei Alpine Ecosystem Research Station in the Northwest Qinghai-Tibetan Plateau. Zhang's PI is a measure of the plant phenological status on the time intervals, calculated the accumulation of PI area in two dimensions. Two phenological variables were described as the Phenological Average Ratio (PAR) and the Lasted Days (LD). The LD may overlap when a certain percentage of the plant population has a nutritional stage that lasts longer than one additional phenological stage or occurs between two phenological stages. Correspondingly, the standard deviations of the observed multiple plant species represent plant population variations. After reviewing the PI applications for phenology studies on Alpine Shrub and Alpine Meadow vegetation, we found that they provide a means of measuring and comparing plant phenology at various levels - population, community, and ecosystem. During the start and end of the season in the Alpine Plateau region, plant phenological changes were signi�cantly constrained by the environmental factors. However, during the summer season, Sunlight Hours (X4), Accumulated Ground Temperature (X3) and Accumulated Air Temperate (X1) above 0 o C had a more uniform impact on plant phenology across the region. Zhang's PI can be used for climate change research by altering warming temperatures, water conditions, and nutrient levels. We also discussed the concern of applying Zhang's PI to global warming research. Moreover, the Seasonal Phenological Index (SPI) can be described on a regional scale and used with ASOS, SEOS, SPAR, and SLD characters to evaluate the changes in the timing of seasonal events in Eq.


Introduction
Phenology studies were essential contents in biology and ecology, from digital mapping of insects, birds, and vegetation, to phenological analysis of plant populations and communities and global changes [1,2,3,4].Based on the distance and visibility of a subject [5], the phenology change and character are identi ed and described as a simple code [6], Phenological Index (PI) [7], NDVI, and EVI [8].Plant phenology, describing the annually recurring sequence of plant developmental stages, is a primary plant biological function and ecosystem feature and re ects biophysical and biogeochemical feedback to the climate system.Plant cyclical biological events are signi cant indicators of the seasonality of the environment, revealing the implications of rising temperatures on vegetation functioning [6,7,8].
Plant phenology studies were based on changes in the timing of seasonal events, such as budburst, owering, fructi cation, and senescence [8, 9,10].Plant phenology has received increasing public and scienti c attention due to the growing evidence that the timing of developmental stages is mainly dependent on environmental conditions and agricultural activity [9,11].
Plant phenology is directly related to climatic conditions and is essential in ecosystem processes, such as carbon and nutrient cycling, adaptive mechanisms, and survival strategies [12,13,14].Furthermore, at the individual plant level, phenology has been shown to in uence tness and reproductive success [12] and has played a direct role in species distribution [15].Eventually, changes in phenology not only have their consequences on affecting species distribution and disrupting species interactions but also altering the carbon cycle and in uencing global climate impacts [16,17,18].Early Studies of the phenological characteristics were restricted to the simple occurrence data [1,2,9,10,19], the trigger of leaf onset/offset, and variations of leaf area throughout the phenological periods [20].Therefore, modeling, assessing, and monitoring phenological dynamics are fundamental to understanding how plants and communities respond to the changing climate and how these changes in uence the ecosystems and their functions [7,12,13,14].An initial phenology study at Alpine Meadow was conducted in 1983 and 1984 [21].Twentyone plant species were observed to study the relationship between plant biomass and phenological stages.However, the phenological two-dimension diagram raised a question [21,22] about obtaining the Digital mapping result, Phenological Index [7].
Land Surface Phenology (LSP) on a large scale has been established from remote sensing-based vegetation indices (VIs) data.They re ect the relationship between the LSP data and vegetation growth status for the start and end of growing seasons as different vegetation indices with the algorithms [23,24].The leaf area index (LAI) was used in DGVMs as Phenology schemes, and land surface models can be established using satellite observational data.The statistical models were based on the relationship between phenology and climate and dynamic carbon models [20].
In this paper, we will present Plant Phenological Index Digital Mapping and discuss the methods of monitoring phenology.(1) Time series-based human visual observations on the individual scale.(2) Digital Plant Phenology Index Mapping from subtracting multiple-variable model.(3) Remote sensors-based observations for a local and regional phenology study.The visibility of the plant community presents the principle of plant phenological retrieval and allows for remotely capturing phenological variations at both small and large scales [8].The most widely applied Vegetation Index (VI) will be discussed using deep learning algorism.In the phenological study and plant phenological index mapping, Zhang's PI (1994) had been examined as a numerical tool for estimating plant growth and will enable improved vegetation monitoring, particularly of plant and community phenology related to digital mapping, clustering phenological types, and phenological change rate, and relationship with environmental climate factors.

Study site
The research sites are located near Haibei Alpine Meadow Ecosystem Research Station (37°N, 101°E) at 3200-3350 m.The vegetation types of the eld sites are typical of Kobresia Humilis Meadow and Potentilla Fruticosa Shrub [15].The eld experiments were initiated and completed in the summer of 1988 and 1989 [7,18] and 1999 [13].The plots at Potentilla Fruticosa Shrub and Kobresia Humilis Meadow were selected and located separately on the homogeneous vegetation sites of 50X50 m 2 .Twenty dominant plant species were selected, and twenty same species duplicates were labeled for statistical observation; the plant population phenological rate (PR) is from 10% at the Start Of Status (SOS) to great than 90% at the End Of Status (EOS).Similar studies [13,14] were conducted at Kobresia Humulis meadow.

Measurement and Data Collection
When the plant phenology status was changing quickly, the observation date was once three days; when the plant phenology status was changing slowly, the observation date was once from ve to seven days.The meteorological data were measured and recorded at Haibei Alpine Meadow Ecosystem Station daily for the same period.Plant population phenological status is reported and based on nutritional stage, bud stage, owering stage, fruit stage, and after-fruit-matured nutritional stage, avescent and dormancy period [7,13].The dormitory stage starts on Oct. 20 when the average air temperature reaches zero degrees in the study area or the last-observed plant populations reach over 90% avescent.

Zhang's PI description
In Plant Population Phenology mapping (Fig. 1), the top line presents the Start Of Status (SOS) at 10%, with increasing the percentage of the status over time (t 1 , t 2 , t 3 , t 4 ), the mapping line reaches the bottom at t 4 .The Maximum Of Status (MOS) is great than 90%, up to 100%.At a certain time of the MOS passed, the End Of Status (EOS) reduced from t5, t6, t7, and t8 at 0-10% ending (Fig. 1).Hence, according to Zhang's PI, we calculate and obtain PIs of six plant-growing phenological stages.An example of one plant PI calculation for one phenological stage is shown in Table 1.Two interpretations were matched when comparing Zhang's and Zhou's phenological reports [7,13].

Average values and variants of PI, PAR, LD, and Environmental factors
A rate of 10% terminated the SOS, and 90% ended the EOS of plant phenology status.Thus, each plant species' lasted days (LD) was calculated.PI, PAR, and LD were calculated for six plant phenology stages for 20 species.We calculated the Average Values and Standard Deviation of the plant Phenology Index (PI) of the Alpine Shrub vegetation (Table 3) [7].The Standard Deviation presents the differences among the 20 plant species, not the error variant from the same plant populations.And the same experiment and calculation were completed for Alpine Meadow vegetation [13].

Clustering plant phenological types
The responses of plant phenology to environmental factors relatively exhibit similarities and dissimilarities.As shown in Fig. 4, the correlative-clustering methods [7,13,15] were employed to categorize plant species of Alpine Shrub into distinct phenological types.Also, Zhou et al. (2002) [13] classi ed the plant phenological types of alpine species studied in the Alpine Meadow.Ye et al. (2014) [14] classi ed the plant phenological types based on lasted days (LD) of the Alpine Meadow.

Results And Discussion
3.1 Phenology observations and primary plant phenological index mapping West and Wein (1971) found that genetic differences increased variability and made quanti cation more di cult when ecologists studied the plant competitions among the species.As a result, they developed Phenological Index Scores(PIS), numerical ratings of more closely de ned phenological stages that would permit the data to be used in statistical tests for different sites or treatments.The PIS is a set of numbers for plant phenological stages.However, the same number of phenological index scores were based on

Clustering alpine plant phenology types by the plant PIs and LDs
Zhang's PI, PAR, and LD are important phenological characters and variables for studying plant phenology.
The correlational analysis of PI verse LD and PI Verse PAR was carried out in Table 2.The PI verse LD presented signi cant relations in six plant phenological stages (p < 0.01).Likewise, the PI verse PAR was signi cantly related to plant phenological stages (p < 0.01).Therefore, Zhang's PI, PAR, and LD are advanced quantitative methods for studying the phenology of plant populations concerning phenological stages.

Phenology characteristics and relation with environmental factors in alpine vegetation
We can calculate the PIs, LDs and PARs for six phenological stages of Alpine shrub and Alpine Meadow (Table 3).  4 and 5).
Alpine Shrub was predominantly found on the northwest slope, where Sunlight Hours (X4) had the greatest in uence on its phenology, ranking rst in ve out of six stages.Similarly, Accumulated Ground Temperature (X3) and Accumulated Air Temperature (X1) above 0°C signi cantly impacted most Alpine Shrub plant phenological stages over the growing season.Precipitation or snowfall (X5) had the most impact on the after-fruit-maturated nutritional stage and was ranked rst only once, and second in the avescent stage of Alpine Shrub (Table 4).
On the other hand, Alpine Meadow was distributed on at landscape, where Sunlight Hours (X4) ranked rst in the bud, owering, and after-fruit-matured nutritional stages.Accumulated Ground Temperature (X3) above 0°C ranked rst in the nutritional and fruit stages.In contrast, Accumulated Air Temperature (X1) above 0°C ranked rst in the avescent stage, showing a more critical in uence over time.Precipitation or snowfall (X5) ranked second and had negligible impacts on the after-fruit-maturated nutritional status of Alpine Meadow (Table 5), indicating a slight difference in the environmental factors that affect Alpine Meadow phenology.
In the nutritional stage, the environmental factors produced the constraints for plant growth beginning, and in the avescent stage, the environmental factors in uenced the ending of plant growth.Therefore, the relationships between PIs and environmental factors were considerably associated with the higher coe cients.On the other hand, during the bud and owering stages, environmental factors had fewer constraints on plant growth and were favorite to plant phenological growths.Thus, their correlation coe cients were reduced to lower levels.The diagrams present these relationships as "V" shape from nutrition, bud, owering, fruit, after-fruit-maturated nutritional, and avescent stage according to Table 4 and Table 5.  Note: The number on the upright corner is the rank of the correlation (R) in Table 4, and The essential content in the plant population ecology is studying plant phenological changes in different stages [7].Thus, Zhang's PI, PAR, and LD in Eq. (1) (2) are practice methods for obtaining plant population phenological characteristics and comparing them in timing matter.However, when we study the seasonal growth at the plant community level, the growing season will be divided into Average Start OF Season (ASOS) at t 1 , Average Maximum OF Season (AMOS) at t 2 and t 3 , Average End OF Season (AEOS) at t 4 (Fig. 5).

SLD ---Seasonally Lasted Days
The SPI in Eq. (3)(4) has the bene t of eliminating the overlapping on the seasonally lasted days (SLD) compared with LD in Eq. ( 2).
Extensional study of Plant phenology has been applied for seasonal and interannual variations in climate.
In dynamic global vegetation models (DGVMs), the impacts of phenology on the ecosystem are considered through the changes in leaf area index (LAI) [20].With the concern of global climate change and its potential impacts, establishing international phenology had rapid progress in remote sensing technologies.
The expanded scope of phenology studies dramatically improved the understanding of vegetation phenology from local to the globe [23,24,25].
Accurate monitoring of vegetation phenology (e.g., the start and end of status) [25] helps understand the impacts of climate change on vegetation and the terrestrial carbon cycle.The enhanced vegetation index (EVI) and normalized difference greenness index (NDGI) are the primary data sources for phenology monitoring at regional and global scales [4].

Conclusion
The phenological studies on plant populations showed that different species exhibited variations in their PIs, PARs, and LDs.These differences demonstrated that certain individuals went through complete phenological stages while others only went through the nutritional stage (as shown in Fig. 3) in the alpine ecoregion.Accordingly, the mapping results of PIs, PARs, and LDs were associated with the timing of SOS, MOS, EOS of the various phenological stages.Zhang (1994) primarily developed a new plant Phenological Index (PI), Phenological Average Ratio (PAR), and Lasted Days (LD) based on the observations of the different plant phenological stages.For example, the total PI for Alpine Shrub in the growing season is 131.23, and the total LD is 231.3; the total PI for Alpine Meadow in the growing season is 137.41, and the total LD is 238.16 from Table 3.The calibrated annual growing days for Alpine Shrub in the study area is 153.5 ± 14.8 [7,21] and 164.16 ± 10.22 for Alpine Meadow [13,30].Moreover, to study phenology at a regional scale, it is highly recommended to use the Seasonal Phenological Index (SPI), which considers seasonal and interannual variations.SPI is calculated using Eq.(3), while SPAR and SLD are calculated using Eq.(4) (5).
In the individual population scale, the phenology studies focused on identifying phenological stages that take time to complete over a year [1,7,22,31].During the start and end of the season in the Alpine Plateau region, plant phenological changes were signi cantly constrained by the environmental factors.And during the summer season, Sunlight Hours (X4), Accumulated Ground Temperature (X3) and Accumulated Air Temperate (X1) above 0 o C had a more uniform impact on plant phenology and plant growth across the region [30,32].However, for remote sensing and mapping data, the focus was on identifying and localizing plant species based on species-speci c phenological characteristics [2,3,4,5].In recent studies, a wide range of Deep Learning methods has been applied, showing their great potential to take plant phenology research to the next level [28].
The vegetation phenology directly indicates that ecosystems respond to environmental changes, which has attracted increasing attention from the academic community.Long-term vegetation phenology data is critical for conducting phenological research.The phenological parameters obtained through experimental observations have become an independent indicator of terrestrial ecosystem change [7,13,14,27,31,33,34].Plant phenology mapping in Alpine Shrub for 20 plant species [7]

Declarations Figures
) between PIs and environmental factors in six phenological stages of Alpine shrub

Figure
Figure 3

Table 1
Examples of observation records, and a calculated PI, and LD for one phenological stage

Table 1 .
speci c7]lant characteristics.For example, the base scores of 1, 2, 3, 4, 5, and 6 for Atriplx nuttallii are the Winter Dormancy, Leaves Regreening and Buds Swelling, Twigs Elongating, Floral Buds Developing, Flowers Opening, and Fruit Developing, respectively.Some base scores for Hilaria jamesii are Winter Dormancy, Growth Initiation, 2 Leaf Stage, 3 Leave Stage, 4 Leaf Stage, and 5 Leave Stage, respectively, in Fig.2 [6].The percentage of plant phenological stages based on twenty plant duplications were recorded as x.10 for 10%, x.20 for 20%, …… x.90 for 90%.West's PIS method described the plant phenological index technique linearly.It did give an indication of the growth process.However, there were limitations in comparing the phenology among the different plant species if we did not use the commonly de ned phenology stages[1,  2,7].West et al. developed a plant phenological index that was presented by the base scores and mapped in the two-dimension diagram with 95% con dence limits.It provided a clearer picture of plant species' phenological change with seasonal times.But West's phenological index is not used as a direct index to plant physiological processes.the further calculations of plant Phenological Index (PI) and PAR were based on Eq. (1)(2) and referred to an example in [7, 21]nt Phenology Diagram mapping in SOS, MOS, EOS, PR, and LD for 20 plant speciesWith precise and detailed records of plant species' phenological status, the two-dimension diagram of Alpine Shrub was completed in Fig.3[7, 18].Zhang and Shi reported the two-dimension diagrams of Alpine Meadow in 1989[18,21].Each plant species had records of SOS, MOS, EOS, , and LD, and

Table 2
[13]elational analysis (P < 0,01, R = 0.537) of PI vs LD, PI vs PAR in six plant phenological stages of Alpine Shrub The dormitory stage starts on Oct. 20 when the average air temperate reaches zero degrees in the study area.Then PI vs PAR correlational analysis is approved (p < 0.01, R = 0.598) in avescent stage based on 20 plant species[7].According to the plant PIs, Zhang et al. used the correlation coe cient[15]to cluster six plant phenological types based on 20 plants of Alpine Shrub.And Zhou et al. used the relative Euclidean distance[13]to cluster three to six plant phenological types based on 19 plants of Alpine Meadow.Finally, based on the studied 12 plant LDs [14], Ye et al. clustered three plant phenological types (Fig. 4A.B. C.).

Table 3
Zhou et al. (2002)PAR in six phenological stages of Alpine Shrub and Alpine Meadow that of Alpine Meadow, 137.41.The growth season LD total of Alpine shrub is 231.30,smallerthan that of Alpine Meadow, 238.30.As the de ned variable, the phenological lasted days (LD) may overlap if a certain percentage plant species experienced a stage that lasted longer than one more phenological stages (Table3), or during the periods between two phenological stages.For instance, in Fig.3, these are observed in plant species such as 13 Asler accidus, 16 Cremanthodium plantagineum,19Lancea tibetica, and 20 Saussurea superba.Their average PARs in the growing season are closer to 50.05 for Alpine Shrub and 49.16 for Alpine Meadow.Therefore, the calibrated annual growing days for Alpine Shrub in the study area is 153.5 ± 14.8[7,21]and 164.16 ± 10.22 for Alpine Meadow[13].We can calculate the environmental factors in six phenological stages for Alpine Shrub and Alpine Meadow.Zhang et al. (2014)andZhou et al. (2002)studied the relationship between PI and environmental factors in six phenological stages.In each phenological stage, we identi ed the environmental factor's correlation coe cient, ranking them in order of 1, 2, and 3, labeled on the upright corner of the numbers for comparison (Tables [7,13]pine Meadow.For example, PI's value in the owering stage is14.82based on the 20 observation species.And the Standard Deviation presents the differences of 20 observed plant species, not the error variant in this case.Dataset[7,13]Table3shows the nutritional stage had the largest PI, LD, and PAR within six phenological stages for Alpine Shrub and Alpine Meadow.The growth season PI total of Alpine Shrub is 132.24, and smaller than

Table 5 Table 5
Correlation(R) between PIs and environmental factors in six phenological stages of Alpine Meadow chan29]26,27].Scaling up this analysis may improve understanding of climate change effects, phenology, and plant productivity on a global scale.With standardized recording systems, ground-based phenological data can be recorded in the future.Such data are essential for better understanding and predicting future environmental processes.In addition, ground-based camera systems and automated image analysis can provide high temporal resolution for calibrating satellite-based monitoring initiatives[28,29].
Yang and He et al. (2017) comprised about 3,000 trees from forested areas across the Tibet Plateau.They found that the start of the growing season and the end of the growing season were indicators of climate