Next Article in Journal
Mapping Surface Flow Velocity of Glaciers at Regional Scale Using a Multiple Sensors Approach
Next Article in Special Issue
A Comparison of OCO-2 SIF, MODIS GPP, and GOSIF Data from Gross Primary Production (GPP) Estimation and Seasonal Cycles in North America
Previous Article in Journal
Monitoring Harvesting by Time Series of Sentinel-1 SAR Data
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Remote Sensing
Volume 11
Issue 21
10.3390/rs11212497
remotesensing-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Mapping Periodic Patterns of Global Vegetation Based on Spectral Analysis of NDVI Time Series

by
Laura Recuero
1,2,
Javier Litago
3,
Jorge E. Pinzón
4,
Margarita Huesca
5,
Maria C. Moyano
6 and
Alicia Palacios-Orueta
1,2,*
1
Departamento de Sistemas y Recursos Naturales, ETSIMFMN, Universidad Politécnica de Madrid, C/José Antonio Novais 10, 28040 Madrid, Spain
2
Centro de Estudios e Investigación para la Gestión de Riesgos Agrarios y Medioambientales (CEIGRAM), Universidad Politécnica de Madrid, C/Senda del Rey 13, 28040 Madrid, Spain
3
Departamento de Economía Agraria, Estadística y Gestión de Empresas, ETSIAAB, Universidad Politécnica de Madrid, Avenida Complutense 3, 28040 Madrid, Spain
4
Science Systems and Applications Inc., Biospheric Sciences Laboratory, NASA Goddard Space Flight Center, Code 618, Greenbelt, MD 20771, USA
5
Land, Air and Water Resources Department, Center for Spatial Technologies And Remote Sensing (CSTARS), University of California Davis, One Shields Avenue, Davis, CA 95616, USA
6
Departamento de Ingeniería Agroforestal, ETSIAAB, Universidad Politécnica de Madrid, Avenida Complutense 3, 28040 Madrid, Spain
*
Author to whom correspondence should be addressed.
Remote Sens. 2019, 11(21), 2497; https://0-doi-org.brum.beds.ac.uk/10.3390/rs11212497
Submission received: 24 September 2019 / Revised: 17 October 2019 / Accepted: 18 October 2019 / Published: 25 October 2019
(This article belongs to the Special Issue Global Biospheric Monitoring with Remote Sensing)
Browse Figures
Versions Notes

Abstract

:
Vegetation seasonality assessment through remote sensing data is crucial to understand ecosystem responses to climatic variations and human activities at large-scales. Whereas the study of the timing of phenological events showed significant advances, their recurrence patterns at different periodicities has not been widely study, especially at global scale. In this work, we describe vegetation oscillations by a novel quantitative approach based on the spectral analysis of Normalized Difference Vegetation Index (NDVI) time series. A new set of global periodicity indicators permitted to identify different seasonal patterns regarding the intra-annual cycles (the number, amplitude, and stability) and to evaluate the existence of pluri-annual cycles, even in those regions with noisy or low NDVI. Most of vegetated land surface (93.18%) showed one intra-annual cycle whereas double and triple cycles were found in 5.58% of the land surface, mainly in tropical and arid regions along with agricultural areas. In only 1.24% of the pixels, the seasonality was not statistically significant. The highest values of amplitude and stability were found at high latitudes in the northern hemisphere whereas lowest values corresponded to tropical and arid regions, with the latter showing more pluri-annual cycles. The indicator maps compiled in this work provide highly relevant and practical information to advance in assessing global vegetation dynamics in the context of global change.

Graphical Abstract

1. Introduction

Seasonal environmental transitions are one of the most predictable variability that ecosystems experience. For example, temperate plants tend to adjust seasonally in response to the changing length of day (photoperiod), exhibiting a relatively predictable phenological timing of production and loss of leaves, flowers and fruits [1]. The recurrence of phenological cycles results in periodic vegetation patterns becoming a distinctive trait of ecosystems [2,3] with different levels of regularity at intra- and inter-annual time scales. To understand vegetation periodicities, in the present research we adopt the definition of seasonality introduced by Hylleberg [4] (p.4) ‘Seasonality is the systematic, although not necessarily regular, intra-year movement caused by the changes of the weather, …’, and we adapt it to an ecological context by adding other factors with influence in vegetation dynamics such as phenology, physical conditions or anthropogenic activities. In addition, when long-term data is available it is common to observe lower frequency periodic patterns whose length exceeds one or several years, commonly known as pluri-annual cycles [5] which are highly dependent on weather, particularly on rainfall–drought cycles. These two factors have been found significantly related to the 11-year solar cycle at some individual meteorological stations [6] and nowadays it is well known that these events are often spatially link across very distant regions through different teleconnections patterns [7].
Vegetation seasonality has a significant effect on land surface functioning. At local scales, the behavior of many birds and insects species are synchronized with vegetation seasonal variations [8,9] with significant disequilibrium when seasonality is altered [10,11,12]; at larger scales, vegetation seasonality plays a key role in biogeochemical cycling and energy fluxes with a strong impact on atmospheric composition and variability [13,14,15,16]. Accurate parameterization of vegetation dynamics by developing indicators of periodic patterns (e.g., related to the amplitude and stability of cycles) can help to improve models making it possible to strengthen our understanding of vegetation-climate interactions [17,18,19]. This is critically important in the current scenario of global change that may alter ecosystem dynamics by changing expected climate conditions and increasing the frequency of climatic extreme events (floods, heat waves, droughts) [20,21]. The extent and ubiquitous impact of global change processes makes necessary the development of tools to understand and operatively quantify the periodicity of phenological patterns and its relationship to climatic and other environmental drivers at large scales.
Several factors can affect the periodic patterns of vegetation. Firstly, climate limits plant growth with different precipitation, temperature and radiation regimes [22,23,24] strongly influencing vegetation seasonality and inter-annual cycles. Secondly, human activities and natural disturbances such as agricultural irrigation, land use changes, pests or fires, can alter seasonal vegetation patterns [25] with slight effects such as changes in seasonality stability or stronger effects such as the number of intra-annual cycles or the appearance of abrupt changes.
Spectral analysis as a part of time series analysis [26] is probably the most extensively used methodology to detect, identify, assess and model periodicities in time series. It has been widely applied in disciplines such as meteorology [27], astronomy [28], economy and agriculture [5], engineering or mathematics [29,30,31]. The Fourier spectral analysis [32] decomposes time series into sinusoidal functions at different frequencies providing a quantitative assessment of the relevance (i.e., power) of each periodic component which is known as spectral estimate or spectrum. The periodogram [27] is a tool to handle Fourier outputs, facilitating the detection and assessment of obvious and hidden periodicities in time series operatively.
Time series of field observations gathered through history have been a significant source of information, mainly in agriculture, providing empirical qualitative knowledge on past climates, trends and societal impacts [33,34]. More recently, several authors have shown the potential of spectral analysis for assessing periodic patterns and relationship with climatic drivers in reproductive phenology, from large data sets of flowering and fruiting dates in tropical regions at species level [35,36,37,38]. Specifically, Bush et al. [37] simulated phenological data by varying the level of uncertainty in the time series, finding that levels of confidence were higher when working with longer time series even in noisy data sets.
Long time series of remote sensing (RS) vegetation indices (VI) at high temporal frequency provide detailed information on vegetation dynamics at large scales facilitating the assessment of vegetation seasonality in a confident manner. The Normalized Difference Vegetation Index (NDVI) [39], based on the visible and near-infrared spectral bands, is the most commonly vegetation index used as a proxy of vegetation biomass or production [40]. Currently, NDVI3g dataset provides the longest NDVI time series with more than 30 years of data derived from the Advanced Very High Resolution Radiometer (AVHRR) sensor on board of the National Oceanic and Atmospheric Administration’s (NOAA’s) Polar Orbiting Environmental Satellites (POES) [41]. In spite of the limitations of NDVI that saturates in well-vegetated areas [42] and the 1/12° coarse resolution, the length of these time series makes it very valuable for vegetation studies [43,44,45,46].
Common approaches to assess vegetation dynamics and seasonality from RS time series rely on estimating the timing of the pheno-phases (i.e., start, end and length of growing cycles) through different methodologies [18,44,47,48]. The number of seasonal cycles has been estimated by identifying the number of maxima over smoothed time series based on threshold values and their dates of occurrence [49,50,51,52,53,54]. Other methodologies rely on time series similarities requiring the previous establishment of reference time series with typical seasonal cycles [55,56]. The majority of vegetation studies are mainly focus on the phenological events whereas certain characteristics of seasonality such as amplitude or stability of the cycles are not widely studied.
Little literature exists on the application of Fourier analysis to assess vegetation seasonality from RS VI indexes. Olsson and Eklundh [57] and Verhoef, Menenti, and Azzali [58] evaluated vegetation dynamics in Africa and South America respectively. Menenti et al. [59] were able to separate different agro-climatic zones in Africa and Andres, Salas, and Skole [60] and Moody and Johnson [61] derived land cover classifications of Brazil and southern California respectively. Azzali and Menenti [62] were able to map land units in southern Africa in terms of leaf bud burst phenology and could correlate amplitude with an aridity index. In agricultural areas, Jakubauskas, Legates, and Kastens [63] and Canisius, Turral, and Molden [64] could identified different crops and bimodal agriculture in Kansas (USA) and in Asia. More recently, Fourier analysis has been used to assess periodicities in natural processes such as fire seasonality [65] and the number of vegetation phenological cycles [66]. Palacios-Orueta et al. [67] used it to demonstrate the existence of a double cycle in the spectral shape angle AS1 index within the growing season of a cotton crop.
All of these studies were focused at local and regional scales being the application at global scale not fully exploited. In addition, most of them have been focused on identifying the number of intra-annual cycles (i.e., uni- or bimodal patterns). However, the information regarding to the significance of periodicities and pluri-annual cycles has not been taken into consideration.
We propose the periodogram as a powerful tool to identify periodic patterns as a type of “fingerprint” for vegetation cycles improving our ability to study phenology-related-cycle-indicators using the entire time series in a highly operative way. Given the easy implementation of spectral analysis and the present high computing capabilities, this methodology could be a valuable tool to complement other approaches when assessing vegetation dynamics and seasonality from RS time series.
The objective of this work is to generate maps containing quantitative information of periodic vegetation patterns (i.e., seasonality and pluri-annual cycles) at global scale from 1982 to 2016 using NDVI3g data set. The specific objectives are the following:
(1) To map the number of intra-annual cycles;
(2) To map the amplitude and the stability of the seasonal cycles;
(3) To assess the presence of pluri-annual cycles taking advantage of the NDVI3g data set length;
(4) To assess the periodicity indicators developed according to the different bioclimates.

2. Materials and Methods

2.1. Remote Sensing Data

The dataset used was the Normalized Difference Vegetation Index-3rd generation version 1.1 (NDVI3g v1.1), derived from the Advanced Very High Resolution Radiometer (AVHRR) instrument onboard the series of the National Oceanic and Atmospheric Administration (NOAA) satellites. It is provided by NASA and compiled by the Global Inventory Monitoring and Modeling System (GIMMS) project [41]. The scenes correspond to fortnightly composites with 1/12° spatial resolution. This dataset provides the longest NDVI time series covering the Earth terrestrial surface enabling the extension of the analysis to the early 1980s. It contains simple quality flags associated with NDVI values informing on the retrieval process in case of low quality NDVI data. Thus, flag 0 indicates high quality NDVI whereas flag 1 and 2 indicate low quality NDVI, retrieving values from spline interpolation or from seasonal profile respectively.
In this study, negative values were masked with zero and missing values were replaced with average values of the previous and next observations. The most of negative and missing values are located at very high latitudes in the northern hemisphere (higher than 60 N) in winter and in the Antarctica. The study period comprises from January 1982 to December 2016 constituting 840 global scenes.

2.2. Bioclimates Map

The bioclimates dataset was obtained from the climate regime classification used for mapping the global ecological land units developed by the U.S. Geological Survey and Esri, published by the Association of American Geographers (AAG) [68]. This classification consists of 37 bioclimates based on the temperature and moisture regimes. It was provided at 250 meters spatial resolution in raster format and resampled to 1/12° in order to reconcile with the NDVI dataset. Figure 1a shows the global spatial distribution of the different bioclimates and Figure 1b shows the percentage of the number of pixels within each bioclimate, indicating their representativeness at global scale and within the thermal regime. Bioclimates under “very cold” and “cold” conditions present more pixels in wetter regimes, particularly in the subclass “wet” whereas “hot” and “very hot” bioclimates show pixels homogeneously distributed among water classes.

2.3. Precipitation Data

Monthly precipitation data from 1982 to 2015 and 0.5° spatial resolution were obtained freely from the Climate Research Unit (CRU), version TS4 [69]. This dataset was used to assess the influence of the precipitation annual distribution on the number of intra-annual cycles in water limited areas.

2.4. NDVI Spectral Analysis: Periodogram

Determination of the seasonal dynamics present in each NDVI time series was carried out by means of classical spectral analysis [32,70,71]. More precisely seasonality was determined by the estimation of the respective periodograms that allowed us to identify the most significant periodic components of each NDVI time series.
Periodogram is based on the classical mathematical theory of Fourier series. In a Fourier series, the signal is decomposed into a series of sine and cosine waves at different frequencies called Fourier frequencies (fi). Thus, the Fourier transform decomposition of the series xt is [70]:
X t = a 0 + ∑ i = 1 m [ a i · cos ( 2 π f i t ) + b i · sin ( 2 π f i t ) ] + e t
where
  • xt are the equally spaced time series data
  • t is the time subscript, t = 1, 2, …, n
  • n is the number of observations in the time series
  • ai and bi are the Fourier coefficients; i = 1, 2, …, m
  • m is the number of frequencies: m = n/2 if n is even; m = (n − 1)/2 if n is odd
  • fi is the i-th Fourier frequency: fi = i/n
  • a0 is the mean term: a0 = x ¯
  • et: error term
The importance of each frequency (fi) for explaining the variance of time series is measured by the periodogram amplitude at that specific frequency (Equation (2)).
I ( f i ) = n 2 · ( a i 2 + b i 2 )
Thus, the periodogram is plotted as I(fi) versus fi for i = 1, 2, …, n/2 or (n − 1)/2. In this case, NDVI periodograms have 420 frequencies which is the half of the total time series observations.
A relatively large (small) value of I(fi) indicates more (less) variance explained respectively for the frequency fi in the time series oscillation. Thus, the most important periodicities are associated with high amplitudes and vice versa. After an exploratory analysis at global scale, the highest amplitudes were identified mainly at periods 24, 12 and 8, being considered as reference periods in this paper. They correspond to the presence of a cycle every 24, 12 and 8 AVHRR periods representing annual, half year and four months-terms respectively.
The seasonality significance was evaluated by means of the Fisher’s Kappa test (FK) [71] for the periodogram. The statistic for this test is computed as the ratio between the maximum amplitude value and the mean of all amplitudes according to the following equation 3:
F K = max ( I ( f i ) ) · m ∑ i = 1 m I ( f i )
The null hypothesis herein is that time series is white noise. The critical value for rejecting the null hypothesis at the 1% significance level is 10.53. Pixels with time series considered as white noise were removed for further analysis.
The NDVI periodogram was computed at pixel level using the “Genecycle” package [72] within R environment software [73].

2.5. Periodicity Indicators

In order to assess the main patterns of the vegetation dynamics at global scale, four periodicity indicators based on the spectral analysis of NDVI time series were defined: The seasonality mode (SM), amplitude (SA) and stability (SS) regarding to the intra-annual cycles and an indicator related to the relevance of pluri-annual cycles (PC) (Table 1). Note that the amplitudes at most higher periods are not included in these indicators because their relationship with abrupt changes in time series either by strong alterations in vegetation cover due to disturbances (e.g., fires, deforestation or flooding) or even errors in the time series.
Figure S1 displays several examples of NDVI time series and their periodograms of different ecosystem types and Table S1 shows the values of the seasonality indicators. The locations are shown in Figure S2.

3. Results

3.1. Global Vegetation Seasonality Mode

Three main types of seasonality stood out for their high ordinate values at 24, 12 and 8 AVHRR periods indicating the presence of one, two and three intra-annual cycles respectively. The geographic distribution of the predominant seasonality mode is shown in a three-color composition map (Figure 2a). Pixels with a clear dominance of one (93.18%), two (4.43%) and three (1.15%) intra-annual seasonal cycles are displayed in green, blue and red for the global land surface, with brighter colors indicating high statistical significance. Only 1.24% of the total land surface, the FK test was not significant (less than 10.53), suggesting the presence of areas with lack of seasonality in South America, Australia, and Central Asia (Figure S3). Most of the areas with two and three intra-annual cycles are distributed among “warm” to “very hot” bioclimates for all water regimes (tropical and arid areas) as well as in agricultural areas (Figure 2b).

3.2. Periodic Vegetation Patterns and Bioclimates: Relationship with the Main Climatic Drivers

Global maps of the NDVI SA, SS and PC values including the three seasonality modes are shown in Figure 3a–c respectively. SA values greater than 20 are located primarily north to the 50 parallel with some fluctuations depending probably on climate and orography. South to the latitude 50N, SA values decrease, reaching the lowest values in arid regions followed by tropical areas. Vegetation in mid and high latitudes of the northern hemisphere shows SS values higher than 60. These regions together with the African savannas show the highest stability patterns. Although lowest values are located in some of the arid areas, some drylands such as the Sahara Desert show stable patterns. Tropical forests show the lowest SS values in the Amazonian region. Lowest occurrence of pluri-annual cycles is observed at latitudes higher than 50 N and in the African savannas whereas the highest values are located in some arid areas.
Figure 4, Figure 5 and Figure 6 show the boxplots of the SA, SS and PC indicators across the range of bioclimates for the three different SM (e.g., one, two and three intra-annual cycles). To easy inspection of the boxplots the bioclimates were ordered firstly by thermal regime and secondly by water classes, being removed from the analysis those bioclimates with none or few pixels. Similar patterns of the SA, SS and PC were observed independently of the number of intra-annual cycles. In general terms there is a significant difference in seasonal amplitude (Figure 4a, Figure 5a and Figure 6a) between “very cold” to “cool” (values around 40) and “warm” to “very hot” bioclimates (values around 15) with higher values and larger within-class variability of all water classes in the former case. Seasonality stability (Figure 4b, Figure 5b and Figure 6b) shows high values (around 80) in “very cold” and “cold” bioclimates with low variability within water classes and between thermal regimes. Higher amplitude and stability values are observed in the “wet”, “moist” and “semi-dry” conditions across bioclimates from “cool” to “very hot”. “Very wet” classes show low amplitude and stability and “dry” regimes show also low values, even lower than “very dry” regimes. Pluri-annual cycles (Figure 4c, Figure 5c and Figure 6c) are mainly observed under water-limited conditions (i.e., from “semi-dry” to “very dry” water classes) from “warm” to “very hot” thermal regimes and with higher values in “dry” than in “very dry” conditions.
Amplitude and stability values are lower in regions with two intra-annual cycles while pluri-annual cycles show similar values. Areas with three intra-annual cycles show lower values in the three indicators.

4. Discussion

Useful indicators of vegetation dynamics were derived from NDVI time series by means of periodogram, extracting information related to the number, amplitude and stability of intra-annual cycles as well as the relevance of pluri-annual cycles. Throughout this discussion, the main periodic vegetation patterns across bioclimates together with the possible driving factors were commented.
Due to the low spatial resolution of the data, this work is mainly focus on the relationship between periodic vegetation patterns and climate. Only agricultural areas with a distinct seasonal pattern were commented.
Most of the global vegetated land surface showed a single seasonal cycle with different values of amplitude, stability and pluri-annual cycles (Figure 2a and Figure 3). Vegetation with double and triple intra-annual cycles showed a consistent spatial distribution, most of them located in arid and tropical regions as well as in agricultural areas. These areas are coincident with those where several authors found difficulties to identify phenological events due to the existence of multiple growing seasons [74,75].
Areas with unimodal seasonality and high amplitude are mainly distributed in the circumpolar region corresponding to climates from “very cold” to “cool” (Figure 4a). This pattern result from the short growing season together with the large difference between vegetation and snow NDVI values [76]. In these areas, vegetation dynamics show highly stable intra- and inter-annual patterns specially in the “very cold” and “cold” climates (Figure 3b–c), resulting probably from the low inter-annual variability of temperature and illumination as main climatic constraints for vegetation growth in these regions [77,78]. Other studies based on phenometric approach have shown similar results with low variability in greenness onset and/or end of senescence [79] and length of growing season [80].
Higher amplitude values in boreal forest than in tundra ecosystem (Figure S1(1,2) and Table S1) indicate that spatial variability in these regions may be related to land cover type, this is shown in a consistent belt of boreal forest around 60–65 N that crosses latitudinally Eurasia and North America (blue in Figure 3a). Verger et al. [81] found similar results in Eurasia but not in North America where high values of Leaf Area Index (LAI) amplitude were located in the eastern part of the continent.
In climates from “warm” to “very hot”, amplitude is consistently low while the rest of the indicators show high variability. In general terms, across temperature regimes lower stability values are observed in the “very wet” and “dry” classes (Figure 4, Figure 5 and Figure 6).
Within the arid regions (“dry” and “very dry” classes), extremely low amplitude of the seasonal cycles must be due to NDVI low values throughout the year. Due to this reason, other authors found difficulties in characterizing the seasonal cycles using a phenometric method [25,82] and others exclude these areas from their studies [25,77,81,83] resulting in a lack of information about these areas. The variability found in the rest of the indicators suggest that they contain relevant information to assess vegetation dynamics in many arid regions of the world. This open new ways to evaluate more accurately local patterns of vegetation dynamics derived from water availability at different scales due to climatic patterns, atmosphere-land surface interactions and soil water retention capacity among other reasons [84,85]. To illustrate the potential of the indicators, we present some of our findings in several arid regions: (1) The Sahara Desert, (2) the Rub-al Khali Desert (Arabian Peninsula), (3) the Sahel and (4) the Australian deserts.
The Sahara Desert shows a clear dominant unimodal pattern with high spatial variability of the stability indicator (Figure 3b, Figure S1-5 and Table S1). Geomorphological features seem to have a significant influence with higher SS located in high elevation areas (e.g., the Tibesti and Ahaggar Mountains) whereas flat areas showed lower SS values. Rocky regions are able to maintain soil moisture for longer periods than adjacent areas favoring vegetation growth. The influence of geomorphological features on land surface phenology by controlling soil moisture has already been reported by other authors [82].
The Rub’Al Khali Desert (a sand desert in the Arabian Peninsula) shows a trimodal seasonal pattern (three intra-annual greening cycles) with consistent spatial distribution but low stability (Figure 2a and Figure 3b). While in this region the precipitation is extremely low and irregular, the periodogram analysis has revealed also three intra-annual cycles due to probably to the occurrence of precipitation events throughout the year (Figure S4). Our results indicate that water availability in sandy soils at specific moments may trigger a fast vegetation response resulting in several short greening periods [86].
In addition, three intra-annual cycles were found in the Sahel area in a narrow band less than 100 km wide around latitude 15 N (Figure 2a, Figure S1-4 and Table S1). The presence of this pattern is statistically significant and shows a high spatial coherence. Other authors mapped this area based on other criteria such as NDVI threshold values [87] and precipitation amount [88]. However, the trimodal pattern found in this work has not been identified in previous studies [54,81]. As in the Rub’Al Khali desert, the periodogram of the precipitation suggests also three intra-annual cycles (Figure S5). Other reasons for this pattern could be: (1) The presence of dry spells within a single rainy season between July and October [89,90], (2) the coexistence of species with different phenological patterns and mixed pixels [91] and (3) the presence data artifacts.
Near to random greening dynamics in large regions of Australian drylands accounted for 30% of the pixels with statistically non-significant seasonality (Figure S1-6, Table S1 and Figure S3). In the remaining area, seasonality stability is low and the occurrence of pluri-annual cycles high (Figure 6b,c). This is in agreement with the intra- and inter-annual variability found in phenological events due to irregularity of precipitation patterns [83,92].
A consistent bimodal seasonality pattern in the Horn of Africa, as well as Southern United States and Northwestern Mexico responds to the existence of two rainy seasons (Figure 2a, Figure S1-3 and Table S1). The displacement of the ITCZ in the one case [48] and the North American Monsoon in the other [93] seems to have a significant impact in the vegetation seasonality.
In the evergreen broadleaf forest in Equatorial regions, the small seasonal amplitude found (Figure 3a) is probably due to the permanent vegetation cover with subtle NDVI variations through the year. The main climatic drivers that condition the seasonal dynamics in these regions have been subject of controversy [78]. While some authors reported that greenness is higher during dry season [94], others proposed that the vegetation index seasonality could be determined by the dormancy of the forest in the dry periods [54].
In this work large areas with statistically significant and stable double seasonal pattern were found in Africa responding to the displacement of the ITCZ [17,95,96]. Guan et al. [17] also found a double cycle in Congo basin using Fourier spectrum of SEVERI LAI time series. In contrast, Vrieling et al. [54] and Adole et al. [48] could not identify this pattern; however they found a double pattern in the Gulf of Guinea using a phenometric approach.
Most of the Amazon basin showed one seasonal cycle which agrees with the results presented by other authors [94]. The seasonality is characterized by a lower stability and a less consistent spatial pattern than in Africa (Figure 3b, Figure 4b). “Very wet” bioclimates close to the equator showed the lowest stability values resulting even in non-significant seasonality (Figure S3, Figure S1-8, and Table S1). It has been shown previously that high cloudiness in the Amazon basin in wet seasons result in many satellite corrupted observations [97,98] and high noise level in the vegetation indices time series, making difficult to capture low variations in evergreen vegetation [25,74]. Our results indicate that the presence of cloudiness may be a significant factor functioning on top of the canopy seasonal dynamics and resulting in the low stability values observed. The presence of a double cycle together with cloudiness artifacts is in agreement with the difficulty found when assessing phenological events in these areas [74,75]. In addition, the high spatial heterogeneity in the seasonality stability indicator may indicate the occurrence of land use changes resulting in a high seasonal variability at local scale.
Disturbances and anthropogenic processes also modify NDVI temporal patterns [99]. Agricultural intensification with multiple cropping systems produces a strong modification on natural vegetation seasonality and NDVI dynamics. Two intra-annual cycles were clearly identified in large rice plantations of China, India, the Nile Delta and Southern Vietnam as previously reported by other authors [54,64,100,101,102]. These results show the capacity of the periodogram derived indicators to identify hot spots of high agriculture intensification.
In this work, we have assessed quantitatively a large variety of NDVI periodic patterns across bioclimates. The northern regions limited by temperature and radiation show stable seasonality, while in the water limited areas vegetation dynamics is highly variable within and between years, with areas showing even a lack of statistically significant seasonality. Geomorphological features and the variability of precipitation events are the main reasons. In particular, it has been possible to identify consistent areas with a trimodal pattern maybe due to the fast response of vegetation to precipitation events. In addition, it has been shown that cloudiness has a significant effect in the NDVI dynamics.
This information provides a better understanding of vegetation dynamics in terms of seasonality and pluri-annual cycles that could be used as complementary information in vegetation phenology studies. In this sense, the number of intra-annual cycles and stability could be used to parameterize the models and assess the accuracy of the results.

5. Conclusions

The use of periodogram derived indicators has shown to be an operative and robust methodology to assess global NDVI time series enabling to describe the Terrestrial Biosphere in terms of periodic vegetation patterns, even in those regions with noisy or low NDVI values.
Although GIMMS NDVI3g dataset has been frequently used, this is the first time that it has been analyzed using spectral analysis at global scale. Furthermore, the length of the time series with more than 30 years of observations provided high power and robustness to the analysis.
The seasonality indicators (number, amplitude and stability) and pluri-annual cycles have shown distinct patterns across bioclimates. The most important findings were the following:
(1) In the circumpolar regions, vegetation showed highly stable seasonality with a clear unimodal pattern and high amplitude.
(2) Large areas with unimodal and bimodal patterns were identified in equatorial rainforest of Africa and America respectively with higher stability in Africa. Not significant seasonality was found in very wet areas in the Amazon Basin maybe due to cloudiness.
(3) Large semi-arid areas in Australia showed near to random seasonality patterns, linked probably to the high precipitation variability. Also, pluri-annual cycles were identified.
(4) Three intra-annual cycles were found in Sahel region and the Arabian Peninsula suggesting a vegetation synchrony with precipitation events.
(5) In the Sahara Desert high spatial variability in seasonality stability was found probably due to different geomorphological features.
The availability of the information provided open new research paths in the context of vegetation dynamics. This methodology is operative and could be implemented with other data sets at better spatial resolution. Within the context of climate change, global warming and the increase of human population, this would facilitate future management and conservation policies.

Supplementary Materials

The following are available online at https://0-www-mdpi-com.brum.beds.ac.uk/2072-4292/11/21/2497/s1, Figure S1: NDVI time series (green) and their periodograms (blue) of different ecosystems. Locations in Figure S2. Figure S2: The red dots are the locations of selected sites for detailed examination of NDVI time series and periodograms. Figure S3: Spatial distribution of locations with non-significant seasonality (in red). Figure S4: (a) Seasonality mode of NDVI and (b) precipitation time series in the Arabian Peninsula. Figure S5: (a) Seasonality mode of NDVI and (b) precipitation time series in the Sahel area. Table S1: Seasonality indicators derived from periodograms of NDVI time series from the locations of Figure S2.

Author Contributions

Conceptualization, L.R., J.L. and A.P.-O.; Data curation, J.E.P.; Formal analysis, L.R., and J.L.; Investigation, L.R., J.L., J.E.P., M.H., M.C.M. and A.P.-O.; Methodology, L.R., J.L. and A.P.-O.; Software, L.R. and J.L.; Visualization, L.R., J.E.P., M.H., M.C.M. and A.P.-O.; Writing—original draft, L.R. and A.P.-O.; Writing—review & editing, L.R., J.L., J.E.P., M.H., M.C.M. and A.P.-O.

Funding

This research was funded by the Spanish Ministry of Science, Innovation and Universities through the pre-doctoral grant of Laura Recuero [FPU014/05633].

Acknowledgments

We would like to thank NASA for providing the NDVI3g dataset and to the anonymous reviewers for their valuable comments.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Chapin, F.S. Effects of plant traits on ecosystem and regional processes: A conceptual framework for predicting the consequences of global change. Ann. Bot. 2003, 91, 455–463. [Google Scholar] [CrossRef] [PubMed]
  2. Jax, K. Ecosystem Functioning; Cambridge University Press: Cambridge, UK, 2010; ISBN 978-0-521-70523-3. [Google Scholar]
  3. Jax, K.; Jones, C.G.; Pickett, S.T.A. The Self-Identity of Ecological Units. Oikos 1998, 82, 253–264. [Google Scholar] [CrossRef]
  4. Hylleberg, S. General Introduction. In Modelling seasonality; Hylleberg, S., Ed.; Oxford University Press: Oxford, UK, 1992; pp. 3–14. ISBN 9780198773184. [Google Scholar]
  5. Beveridge, W.H. Wheat Prices and Rainfall in Western Europe. J. R. Stat. Soc. 1922, 85, 412–475. [Google Scholar] [CrossRef]
  6. Wood, C.A.; Lovett, R.R. Rainfall, drought and the solar cycle. Nature 1974, 251, 594–596. [Google Scholar] [CrossRef]
  7. Feldstein, S.B.; Franzke, C.L.E. Atmospheric Teleconnection Patterns. In Nonlinear and Stochastic Climate Dynamics; Franzke, C.L.E., O’Kane, T.J., Eds.; Cambridge University Press: Cambridge, UK, 2017; pp. 54–104. [Google Scholar]
  8. Bartomeus, I.; Ascher, J.S.; Wagner, D.; Danforth, B.N.; Colla, S.; Kornbluth, S.; Winfree, R. Climate-associated phenological advances in bee pollinators and bee-pollinated plants. Proc. Natl. Acad. Sci. USA 2011, 108, 20645–20649. [Google Scholar] [CrossRef] [Green Version]
  9. Visser, M.E.; Noordwijk, A.V.; Tinbergen, J.M.; Lessells, C.M. Warmer springs lead to mistimed reproduction in great tits (Parus major). Proc. R. Soc. B Biol. Sci. 1998, 265, 1867–1870. [Google Scholar] [CrossRef] [Green Version]
  10. Renner, S.S.; Zohner, C.M. Climate Change and Phenological Mismatch in Trophic Interactions Among Plants, Insects, and Vertebrates. Annu. Rev. Ecol. Evol. Syst. 2018, 49, 165–182. [Google Scholar] [CrossRef]
  11. Mayor, S.J.; Guralnick, R.P.; Tingley, M.W.; Otegui, J.; Withey, J.C.; Elmendorf, S.C.; Andrew, M.E.; Leyk, S.; Pearse, I.S.; Schneider, D.C. Increasing phenological asynchrony between spring green-up and arrival of migratory birds. Sci. Rep. 2017, 7, 1902. [Google Scholar] [CrossRef]
  12. Kharouba, H.M.; Ehrlén, J.; Gelman, A.; Bolmgren, K.; Allen, J.M.; Travers, S.E.; Wolkovich, E.M. Global shifts in the phenological synchrony of species interactions over recent decades. Proc. Natl. Acad. Sci. USA 2018, 115, 5211–5216. [Google Scholar] [CrossRef] [Green Version]
  13. Tucker, C.J.; Fung, I.Y.; Keeling, C.D.; Gammon, R.H. Relationship between atmospheric CO2 variations and a satellite-derived vegetation index. Nature 1986, 319, 195–199. [Google Scholar] [CrossRef]
  14. Keeling, C.D.; Chin, J.F.S.; Whorf, T.P. Increased activity of northern vegetation inferred from atmospheric CO2 measurements. Nature 1996, 382, 146–149. [Google Scholar] [CrossRef]
  15. McPherson, R.A. A review of vegetation-atmosphere interactions and their influences on mesoscale phenomena. Prog. Phys. Geogr. 2007, 31, 261–285. [Google Scholar] [CrossRef]
  16. Richardson, A.D.; Keenan, T.F.; Migliavacca, M.; Ryu, Y.; Sonnentag, O.; Toomey, M. Climate change, phenology, and phenological control of vegetation feedbacks to the climate system. Agric. For. Meteorol. 2013, 169, 156–173. [Google Scholar] [CrossRef]
  17. Guan, K.; Medvigy, D.; Wood, E.F.; Caylor, K.K.; Li, S.; Jeong, S.J. Deriving vegetation phenological time and trajectory information over africa using seviri daily LAI. IEEE Trans. Geosci. Remote Sens. 2014, 52, 1113–1130. [Google Scholar] [CrossRef]
  18. Zhang, X.; Friedl, M.A.; Schaaf, C.B.; Strahler, A.H.; Hodges, J.C.F.; Gao, F.; Reed, B.C.; Huete, A. Monitoring vegetation phenology using MODIS. Remote Sens. Environ. 2003, 84, 471–475. [Google Scholar] [CrossRef]
  19. Zhu, J.; Zeng, X. Influences of the seasonal growth of vegetation on surface energy budgets over middle to high latitudes. Int. J. Climatol. 2017, 37, 4251–4260. [Google Scholar] [CrossRef]
  20. Parmesan, C. Ecological and Evolutionary Responses to Recent Climate Change. Annu. Rev. Ecol. Evol. Syst. 2006, 37, 637–669. [Google Scholar] [CrossRef] [Green Version]
  21. IPCC. Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change; Core Writing Team, Pachauri, R.K., Meyer, L.A., Eds.; IPCC: Geneva, Switzerland, 2014; ISBN 978-92-9169-143-2. [Google Scholar]
  22. Churkina, G.; Running, S.W. Contrasting Climatic Controls on the Estimated Productivity of Global Terrestrial Biomes. Ecosystems 1998, 1, 206–215. [Google Scholar] [CrossRef]
  23. Garonna, I.; de Jong, R.; Stöckli, R.; Schmid, B.; Schenkel, D.; Schimel, D.; Schaepman, M.E. Shifting relative importance of climatic constraints on land surface phenology. Environ. Res. Lett. 2018, 13, 024025. [Google Scholar] [CrossRef] [Green Version]
  24. Nemani, R.R.; Keeling, C.D.; Hashimoto, H.; Jolly, W.M.; Piper, S.C.; Tucker, C.J.; Myneni, R.B.; Running, S.W. Climate-Driven Increases in Global Terrestrial Net Primary Production from 1982 to 1999. Science 2003, 300, 1560–1563. [Google Scholar] [CrossRef] [Green Version]
  25. Zhang, X.; Friedl, M.A.; Schaaf, C.B. Global vegetation phenology from Moderate Resolution Imaging Spectroradiometer (MODIS): Evaluation of global patterns and comparison with in situ measurements. J. Geophys. Res. Biogeosci. 2006, 111, G04017. [Google Scholar] [CrossRef]
  26. Walden, A.T.; Percival, D.B.; McCoy, E.J. Spectrum estimation by wavelet thresholding of multitaper estimators. IEEE Trans. Signal Process. 1998, 46, 3153–3165. [Google Scholar] [CrossRef]
  27. Schuster, A. On the investigation of hidden periodicities with application to a supposed 26 day period of meteorological phenomena. J. Geophys. Res. 1898, 3, 13–41. [Google Scholar] [CrossRef]
  28. Schuster, A. On the Periodicities of Sunspots. Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 1906, 206, 69–100. [Google Scholar] [CrossRef]
  29. Nawaz, S.; Khan, A.A.; Mahmood, A.; Javed, C.F. Performance Analysis of the Time-Based and Periodogram-Based Energy Detector for Spectrum Sensing. Int. J. Electr. Comp. Energ. Electr. Commun. Eng. 2017, 11, 586–590. [Google Scholar]
  30. Kayham, A.S.; El-Jaroudi, A.; Chaparro, L.F. Evolutionary Periodogram for Nonstationary Signals. IEEE Trans. Signal Process. 1994, 42, 1527–1536. [Google Scholar] [CrossRef]
  31. Zakharov, Y.V.; Baronkin, V.M.; Tozer, T.C. DFT-based frequency estimators with narrow acquisition range. IEE Proc. Commun. 2001, 148, 1–7. [Google Scholar] [CrossRef]
  32. Bloomfield, P. Fourier Analysis of Time Series: An Introduction, 2nd ed.; John Wiley & Sons: New York, NY, USA, 2001; ISBN 0-471-88948-2. [Google Scholar]
  33. Lieth, H. (Ed.) Phenology and Seasonality Modeling; Springer Science + Business Media: New York, NY, USA, 1974; ISBN 978-3-642-51865-2. [Google Scholar]
  34. Van Vliet, A.J.H.; De Groot, R.S.; Bellens, Y.; Braun, P.; Bruegger, R.; Bruns, E.; Clevers, J.; Estreguil, C.; Flechsig, M.; Jeanneret, F.; et al. The European Phenology Network. Int. J. Biometeorol. 2003, 47, 202–212. [Google Scholar] [CrossRef]
  35. Zalamea, P.C.; Munoz, F.; Stevenson, P.R.; Timothy Paine, C.E.; Sarmiento, C.; Sabatier, D.; Heuret, P. Continental-scale patterns of Cecropia reproductive phenology: Evidence from herbarium specimens. Proc. R. Soc. B Biol. Sci. 2011, 278, 2437–2445. [Google Scholar] [CrossRef]
  36. Adamescu, G.S.; Plumptre, A.J.; Abernethy, K.A.; Polansky, L.; Bush, E.R.; Chapman, C.A.; Shoo, L.P.; Fayolle, A.; Janmaat, K.R.L.; Robbins, M.M.; et al. Annual cycles are the most common reproductive strategy in African tropical tree communities. Biotropica 2018, 50, 418–430. [Google Scholar] [CrossRef]
  37. Bush, E.R.; Abernethy, K.A.; Jeffery, K.; Tutin, C.; White, L.; Dimoto, E.; Dikangadissi, J.T.; Jump, A.S.; Bunnefeld, N. Fourier analysis to detect phenological cycles using long-term tropical field data and simulations. Methods Ecol. Evol. 2017, 8, 530–540. [Google Scholar] [CrossRef]
  38. Chapman, C.A.; Wrangham, R.W.; Chapman, L.J.; Kennard, D.K.; Zanne, A.E. Fruit and flower phenology at two sites in Kibale National Park, Uganda. J. Trop. Ecol. 1999, 15, 189–211. [Google Scholar] [CrossRef] [Green Version]
  39. Tucker, C.J. Red and photographic infrared linear combinations for monitoring vegetation. Remote Sens. Environ. 1979, 8, 127–150. [Google Scholar] [CrossRef] [Green Version]
  40. Baret, F.; Guyot, G. Potentials and limits of vegetation indices for LAI and APAR assessment. Remote Sens. Environ. 1991, 35, 161–173. [Google Scholar] [CrossRef]
  41. Pinzon, J.E.; Tucker, C.J. A non-stationary 1981-2012 AVHRR NDVI3g time series. Remote Sens. 2014, 6, 6929–6960. [Google Scholar] [CrossRef]
  42. Huete, A.; Didan, K.; Miura, T.; Rodriguez, E.; Gao, X.; Ferreira, L. Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sens. Environ. 2002, 83, 195–213. [Google Scholar] [CrossRef]
  43. Park, T.; Ganguly, S.; Tømmervik, H.; Euskirchen, E.S.; Høgda, K.A.; Karlsen, S.R.; Brovkin, V.; Nemani, R.R.; Myneni, R.B. Changes in growing season duration and productivity of northern vegetation inferred from long-term remote sensing data. Environ. Res. Lett. 2016, 11, 084001. [Google Scholar] [CrossRef]
  44. White, M.A.; de Beurs, K.M.; Didan, K.; Inouye, D.W.; Richardson, A.D.; Jensen, O.P.; O’Keefe, J.; Zhang, G.; Nemani, R.R.; van Leeuwen, W.J.D.; et al. Intercomparison, interpretation, and assessment of spring phenology in North America estimated from remote sensing for 1982–2006. Glob. Chang. Biol. 2009, 15, 2335–2359. [Google Scholar] [CrossRef]
  45. White, M.A.; Nemani, R.R. Real-time monitoring and short-term forecasting of land surface phenology. Remote Sens. Environ. 2006, 104, 43–49. [Google Scholar] [CrossRef]
  46. Yu, L.; Liu, T.; Bu, K.; Yan, F.; Yang, J.; Chang, L. Monitoring the long term vegetation phenology change in Northeast China from 1982 to 2015. Sci. Rep. 2017, 7, 14770. [Google Scholar] [CrossRef]
  47. Atzberger, C.; Klisch, A.; Mattiuzzi, M.; Vuolo, F. Phenological metrics derived over the European continent from NDVI3g data and MODIS time series. Remote Sens. 2013, 6, 257–284. [Google Scholar] [CrossRef]
  48. Adole, T.; Dash, J.; Atkinson, P.M. Characterising the land surface phenology of Africa using 500 m MODIS EVI. Appl. Geogr. 2018, 90, 187–199. [Google Scholar] [CrossRef] [Green Version]
  49. Ding, M.; Chen, Q.; Xiao, X.; Xin, L.; Zhang, G.; Li, L. Variation in cropping intensity in northern China from 1982 to 2012 based on GIMMS-NDVI data. Sustainability 2016, 8, 1123. [Google Scholar] [CrossRef]
  50. Estel, S.; Kuemmerle, T.; Levers, C.; Baumann, M.; Hostert, P. Mapping cropland-use intensity across Europe using MODIS NDVI time series. Environ. Res. Lett. 2016, 11, 024015. [Google Scholar] [CrossRef] [Green Version]
  51. Jönsson, P.; Eklundh, L. Seasonality extraction by function fitting to time-series of satellite sensor data. IEEE Trans. Geosci. Remote Sens. 2002, 40, 1824–1832. [Google Scholar] [CrossRef]
  52. Kontgis, C.; Schneider, A.; Ozdogan, M. Mapping rice paddy extent and intensification in the Vietnamese Mekong River Delta with dense time stacks of Landsat data. Remote Sens. Environ. 2015, 169, 255–269. [Google Scholar] [CrossRef]
  53. Sakamoto, T.; Van Nguyen, N.; Ohno, H.; Ishitsuka, N.; Yokozawa, M. Spatio-temporal distribution of rice phenology and cropping systems in the Mekong Delta with special reference to the seasonal water flow of the Mekong and Bassac rivers. Remote Sens. Environ. 2006, 100, 1–16. [Google Scholar] [CrossRef]
  54. Vrieling, A.; De Leeuw, J.; Said, M.Y. Length of growing period over africa: Variability and trends from 30 years of NDVI time series. Remote Sens. 2013, 5, 982–1000. [Google Scholar] [CrossRef]
  55. Guan, X.; Huang, C.; Liu, G.; Meng, X.; Liu, Q. Mapping rice cropping systems in Vietnam using an NDVI-based time-series similarity measurement based on DTW distance. Remote Sens. 2016, 8, 19. [Google Scholar] [CrossRef]
  56. Zhang, Y.; Hepner, G.F. The Dynamic-Time-Warping-based k -means++ clustering and its application in phenoregion delineation. Int. J. Remote Sens. 2017, 38, 1720–1736. [Google Scholar] [CrossRef]
  57. Olsson, L.; Eklundh, L. Fourier Series for analysis of temporal sequences of satellite sensor imagery. Int. J. Remote Sens. 1994, 15, 3735–3741. [Google Scholar] [CrossRef]
  58. Verhoef, W.; Menenti, M.; Azzali, S. Cover a colour composite of NOAA-AVHRR-NDVI based on time series analysis (1981-1992). Int. J. Remote Sens. 1996, 17, 231–235. [Google Scholar] [CrossRef]
  59. Menenti, M.; Azzali, S.; Verhoef, W.; van Swol, R. Mapping agroecological zones and time lag in vegetation growth by means of fourier analysis of time series of NDVI images. Adv. Space Res. 1993, 13, 233–237. [Google Scholar] [CrossRef]
  60. Andres, L.; Salas, W.A.; Skole, D. Fourier analysis of multi-temporal AVHRR data applied to a land cover classification. Int. J. Remote Sens. 1994, 15, 1115–1121. [Google Scholar] [CrossRef]
  61. Moody, A.; Johnson, D.M. Land-Surface Phenologies from AVHRR Using the Discrete Fourier Transform. Remote Sens. Environ. 2001, 75, 305–323. [Google Scholar] [CrossRef]
  62. Azzali, S.; Menenti, M. Mapping vegetation-soil-climate complexes in southern Africa using temporal Fourier analysis of NOAA-AVHRR NDVI data. Int. J. Remote Sens. 2000, 21, 973–996. [Google Scholar] [CrossRef]
  63. Jakubauskas, M.E.; Legates, D.R.; Kastens, J.H. Crop identification using harmonic analysis of time-series AVHRR NDVI data. Comput. Electron. Agric. 2002, 37, 127–139. [Google Scholar] [CrossRef]
  64. Canisius, F.; Turral, H.; Molden, D. Fourier analysis of historical NOAA time series data to estimate bimodal agriculture. Int. J. Remote Sens. 2007, 28, 5503–5522. [Google Scholar] [CrossRef]
  65. Huesca, M.; Litago, J.; Palacios-Orueta, A.; Montes, F.; Sebastián-López, A.; Escribano, P. Assessment of forest fire seasonality using MODIS fire potential: A time series approach. Agric. For. Meteorol. 2009, 149, 1946–1955. [Google Scholar] [CrossRef]
  66. Vrieling, A.; Meroni, M.; Mude, A.G.; Chantarat, S.; Ummenhofer, C.C.; de Bie, K.C.A.J.M. Early assessment of seasonal forage availability for mitigating the impact of drought on East African pastoralists. Remote Sens. Environ. 2016, 174, 44–55. [Google Scholar] [CrossRef] [Green Version]
  67. Palacios-Orueta, A.; Huesca, M.; Whiting, M.L.; Litago, J.; Khanna, S.; Garcia, M.; Ustin, S.L. Derivation of phenological metrics by function fitting to time-series of Spectral Shape Indexes AS1 and AS2: Mapping cotton phenological stages using MODIS time series. Remote Sens. Environ. 2012, 126, 148–159. [Google Scholar] [CrossRef]
  68. Sayre, R.; Dangermond, J.; Frye, C.; Vaughan, R.; Aniello, P.; Breyer, S.; Cribbs, D.; Hopkins, D.; Nauman, R.; Derrenbacher, W.; et al. A New Map of Global Ecological Land Units—An Ecophysiographic Stratification Approach; Association of American Geographers: Washington, DC, USA, 2014; p. 46. [Google Scholar]
  69. University of East Anglia Climatic Research Unit; Harris, I.C.; Jones, P.D. CRU TS4.00: Climatic Research Unit (CRU) Time-Series (TS) Version 4.00 of High-Resolution Gridded Data of Month-by-Month Variation in Climate (Jan. 1901–Dec. 2015); Centre for Environmental Data Analysis: Didcot, UK, 2017. [Google Scholar]
  70. Box, G.E.P.; Jenkins, G.M.; Reinsel, G.C. Time Series Analysis. Forecasting and Control, 3rd ed.; Prentice Hall: Englewood Cliff, NJ, USA, 1994; ISBN 0-13-060774-6. [Google Scholar]
  71. Fuller, W.A. Introduction to Statistical Time Series; John Wiley and Sons: New York, NY, USA, 1976. [Google Scholar]
  72. Ahdesmaki, M.; Fokianos, K.; Strimmer, K. GeneCycle: Identification of Periodically Expressed Genes, R package version 1.1.2. 2015; Available online: http:CRAN.R-project.org/package=GeneCycle (accessed on 1 August 2016).
  73. R Core Team. R: A Language and Environment for Statistical Computing. Available online: http://www.r-project.org/ (accessed on 15 October 2018).
  74. Maignan, F.; Bréon, F.M.; Bacour, C.; Demarty, J.; Poirson, A. Interannual vegetation phenology estimates from global AVHRR measurements. Comparison with in situ data and applications. Remote Sens. Environ. 2008, 112, 496–505. [Google Scholar] [CrossRef]
  75. Julien, Y.; Sobrino, J.A. Global land surface phenology trends from GIMMS database. Int. J. Remote Sens. 2009, 30, 3495–3513. [Google Scholar] [CrossRef] [Green Version]
  76. Beck, P.S.A.; Atzberger, C.; Høgda, K.A.; Johansen, B.; Skidmore, A.K. Improved monitoring of vegetation dynamics at very high latitudes: A new method using MODIS NDVI. Remote Sens. Environ. 2006, 100, 321–334. [Google Scholar] [CrossRef]
  77. Zhang, X.; Tan, B.; Yu, Y. Interannual variations and trends in global land surface phenology derived from enhanced vegetation index during 1982–2010. Int. J. Biometeorol. 2014, 58, 547–564. [Google Scholar] [CrossRef]
  78. Forkel, M.; Migliavacca, M.; Thonicke, K.; Reichstein, M.; Schaphoff, S.; Weber, U.; Carvalhais, N. Codominant water control on global interannual variability and trends in land surface phenology and greenness. Glob. Chang. Biol. 2015, 21, 3414–3435. [Google Scholar] [CrossRef]
  79. Rodriguez-Galiano, V.F.; Dash, J.; Atkinson, P.M. Characterising the land surface phenology of Europe using decadal MERIS data. Remote Sens. 2015, 7, 9390–9409. [Google Scholar] [CrossRef]
  80. De Jong, R.; de Bruin, S.; de Wit, A.; Schaepman, M.E.; Dent, D.L. Analysis of monotonic greening and browning trends from global NDVI time-series. Remote Sens. Environ. 2011, 115, 692–702. [Google Scholar] [CrossRef] [Green Version]
  81. Verger, A.; Filella, I.; Baret, F.; Peñuelas, J. Vegetation baseline phenology from kilometric global LAI satellite products. Remote Sens. Environ. 2016, 178, 1–14. [Google Scholar] [CrossRef] [Green Version]
  82. Yan, D.; Zhang, X.; Yu, Y.; Guo, W.; Hanan, N.P. Characterizing land surface phenology and responses to rainfall in the Sahara desert. J. Geophys. Res. Biogeosci. 2016, 121, 2243–2260. [Google Scholar] [CrossRef] [Green Version]
  83. Zhao, L.; Dai, A.; Dong, B. Changes in global vegetation activity and its driving factors during 1982–2013. Agric. For. Meteorol. 2018, 249, 198–209. [Google Scholar] [CrossRef]
  84. Schwinning, S.; Sala, O.E.; Loik, M.E.; Ehleringer, J.R. Thresholds, memory, and seasonality: Understanding pulse dynamics in arid/semi-arid ecosystems. Oecologia 2004, 141, 191–193. [Google Scholar] [CrossRef] [PubMed]
  85. Vicente-Serrano, S.M.; Gouveia, C.; Camarero, J.J.; Begueria, S.; Trigo, R.; Lopez-Moreno, J.I.; Azorin-Molina, C.; Pasho, E.; Lorenzo-Lacruz, J.; Revuelto, J.; et al. Response of vegetation to drought time-scales across global land biomes. Proc. Natl. Acad. Sci. USA 2013, 110, 52–57. [Google Scholar] [CrossRef] [PubMed]
  86. Noy-Meir, I. Desert Ecosystems: Environment and Producers. Annu. Rev. Ecol. Syst. 1973, 4, 25–51. [Google Scholar] [CrossRef]
  87. Kim, D.; Chin, M.; Remer, L.A.; Diehl, T.; Bian, H.; Yu, H.; Brown, M.E.; Stockwell, W.R. Role of surface wind and vegetation cover in multi-decadal variations of dust emission in the Sahara and Sahel. Atmos. Environ. 2017, 148, 282–296. [Google Scholar] [CrossRef]
  88. Anyamba, A.; Small, J.L.; Tucker, C.J.; Pak, E.W. Thirty-two Years of Sahelian Zone Growing Season Non-Stationary NDVI3g Patterns. Remote Sens. 2014, 6, 3101–3122. [Google Scholar] [CrossRef]
  89. Anyamba, A.; Tucker, C.J. Analysis of Sahelian vegetation dynamics using NOAA-AVHRR NDVI data from 1981–2003. J. Arid Environ. 2005, 63, 596–614. [Google Scholar] [CrossRef]
  90. Abdi, A.M.; Boke-Olén, N.; Tenenbaum, D.E.; Tagesson, T.; Cappelaere, B.; Ardö, J. Evaluating water controls on vegetation growth in the semi-arid sahel using field and earth observation data. Remote Sens. 2017, 9, 294. [Google Scholar] [CrossRef]
  91. Brandt, M.; Hiernaux, P.; Tagesson, T.; Verger, A.; Rasmussen, K.; Diouf, A.A.; Mbow, C.; Mougin, E.; Fensholt, R. Woody plant cover estimation in drylands from Earth Observation based seasonal metrics. Remote Sens. Environ. 2016, 172, 28–38. [Google Scholar] [CrossRef] [Green Version]
  92. Broich, M.; Huete, A.; Tulbure, M.G.; Ma, X.; Xin, Q.; Paget, M.; Restrepo-Coupe, N.; Davies, K.; Devadas, R.; Held, A. Land surface phenological response to decadal climate variability across Australia using satellite remote sensing. Biogeosciences 2014, 11, 5181–5198. [Google Scholar] [CrossRef] [Green Version]
  93. Adams, D.K.; Comrie, A.C. The North American Monsoon. Bull. Amer. Meteor. Soc. 1997, 78, 2197–2213. [Google Scholar] [CrossRef] [Green Version]
  94. Myneni, R.B.; Yang, W.; Nemani, R.R.; Huete, A.R.; Dickinson, R.E.; Knyazikhin, Y.; Didan, K.; Fu, R.; Negron Juarez, R.I.; Saatchi, S.S.; et al. Large seasonal swings in leaf area of Amazon rainforests. Proc. Natl. Acad. Sci. USA 2007, 104, 4820–4823. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  95. Yan, D.; Zhang, X.; Yu, Y.; Guo, W.A. Comparison of Tropical Rainforest Phenology Retrieved from Geostationary (SEVIRI) and Polar-Orbiting (MODIS) Sensors Across the Congo Basin. IEEE Trans. Geosci. Remote Sens. 2016, 54, 4867–4881. [Google Scholar] [CrossRef]
  96. Gond, V.; Fayolle, A.; Pennec, A.; Cornu, G.; Mayaux, P.; Camberlin, P.; Doumenge, C.; Fauvet, N.; Gourlet-Fleury, S. Vegetation structure and greenness in Central Africa from Modis multi-temporal data. Philos. Trans. R. Soc. B Biol. Sci. 2013, 368, 20120309. [Google Scholar] [CrossRef] [Green Version]
  97. Hilker, T.; Lyapustin, A.I.; Tucker, C.J.; Hall, F.G.; Myneni, R.B.; Wang, Y.; Bi, J.; Mendes de Moura, Y.; Sellers, P.J. Vegetation dynamics and rainfall sensitivity of the Amazon. Proc. Natl. Acad. Sci. USA 2014, 111, 16041–16046. [Google Scholar] [CrossRef] [Green Version]
  98. Samanta, A.; Ganguly, S.; Vermote, E.; Nemani, R.R.; Myneni, R.B. Why Is Remote Sensing of Amazon Forest Greenness So Challenging? Earth Interact. 2012, 16, 1–14. [Google Scholar] [CrossRef] [Green Version]
  99. Evans, J.; Geerken, R. Discrimination between climate and human-induced dryland degradation. J. Arid Environ. 2004, 57, 535–554. [Google Scholar] [CrossRef]
  100. Chauhan, B.S.; Mahajan, G.; Sardana, V.; Timsina, J.; Jat, M.L. Productivity and Sustainability of the Rice-Wheat Cropping System in the Indo-Gangetic Plains of the Indian Subcontinent: Problems, Opportunities, and Strategies. In Advances in Agronomy, 1st ed.; Sparks, D.L., Ed.; Elsevier: Amsterdam, The Netherlands, 2012; Volume 117, pp. 315–369. ISBN 978-0-12-394278-4. [Google Scholar]
  101. Nguyen, T.T.H.; De Bie, C.A.J.M.; Ali, A.; Smaling, E.M.A.; Chu, T.H. Mapping the irrigated rice cropping patterns of the Mekong delta, Vietnam, through hyper-temporal SPOT NDVI image analysis. Int. J. Remote Sens. 2012, 33, 415–434. [Google Scholar] [CrossRef]
  102. Kotb, T.H.S.; Watanabe, T.; Ogino, Y.; Tanji, K.K. Soil salinization in the Nile Delta and related policy issues in Egypt. Agric. Water Manag. 2000, 43, 239–261. [Google Scholar] [CrossRef]
Remotesensing 11 02497 g001 550
Figure 1. (a) Bioclimates derived from the global ecological land units map [68]. The lighter and darker colors indicate drier and more humid bioclimates respectively. (b) The number of pixels in percentage per bioclimate.
Figure 1. (a) Bioclimates derived from the global ecological land units map [68]. The lighter and darker colors indicate drier and more humid bioclimates respectively. (b) The number of pixels in percentage per bioclimate.
Remotesensing 11 02497 g001
Remotesensing 11 02497 g002 550
Figure 2. (a) Spatial distribution of the three seasonality modes in an RGB color composition. (b) Proportion of each type seasonality mode within bioclimates. Green, blue and red colors represent ecosystems with one, two and three intra-annual cycles.
Figure 2. (a) Spatial distribution of the three seasonality modes in an RGB color composition. (b) Proportion of each type seasonality mode within bioclimates. Green, blue and red colors represent ecosystems with one, two and three intra-annual cycles.
Remotesensing 11 02497 g002
Remotesensing 11 02497 g003 550
Figure 3. (a) Seasonality amplitude, (b) seasonality stability and (c) pluri-annual cycles indicator of NDVI time series.
Figure 3. (a) Seasonality amplitude, (b) seasonality stability and (c) pluri-annual cycles indicator of NDVI time series.
Remotesensing 11 02497 g003
Remotesensing 11 02497 g004 550
Figure 4. (a) Boxplots of the seasonality amplitude, (b) seasonality stability (c) and the pluri-annual cycles indicator of ecosystems with unimodal seasonal pattern per each bioclimate.
Figure 4. (a) Boxplots of the seasonality amplitude, (b) seasonality stability (c) and the pluri-annual cycles indicator of ecosystems with unimodal seasonal pattern per each bioclimate.
Remotesensing 11 02497 g004
Remotesensing 11 02497 g005 550
Figure 5. (a) Boxplots of the seasonality amplitude, (b) seasonality stability (c) and the pluri-annual cycles indicator of ecosystems with bimodal seasonal pattern per each bioclimate.
Figure 5. (a) Boxplots of the seasonality amplitude, (b) seasonality stability (c) and the pluri-annual cycles indicator of ecosystems with bimodal seasonal pattern per each bioclimate.
Remotesensing 11 02497 g005
Remotesensing 11 02497 g006 550
Figure 6. (a) Boxplots of the seasonality amplitude, (b) seasonality stability (c) and the pluri-annual cycles indicator of ecosystems with trimodal seasonal pattern per each bioclimate.
Figure 6. (a) Boxplots of the seasonality amplitude, (b) seasonality stability (c) and the pluri-annual cycles indicator of ecosystems with trimodal seasonal pattern per each bioclimate.
Remotesensing 11 02497 g006
Table
Table 1. Periodicity indicators for NDVI time series.
Table 1. Periodicity indicators for NDVI time series.
IndicatorStatistical Computation Interpretation
Intra-annual cyclesSeasonality Mode (SM)The period of the maximum amplitude (i.e., 24, 12 or 8)It indicates the occurrence of one, two or three intra-annual cycles.
Seasonality Amplitude (SA)The value of the maximum amplitude (i.e., at 24, 12 or 8 periods). The values range from 0 to 100.High values indicate high amplitude of the NDVI cycle at the dominant seasonality mode.
Seasonality Stability (SS)Ratio between the value of the maximum amplitude (i.e., at 8, 12 or 24 periods) and the sum of the amplitudes at periods lower than or equal to 24 (annual term). Values are in percentage ranging between 0 and 100.High values indicate high stability of a certain seasonality mode. Low values indicate a mix of seasonality modes.
Pluri-annual cycles (PC)Ratio between the sum of all the amplitude values between period 24 (one year) and period 120 (5 years) and the sum of all the amplitudes. Values are in percentage ranging between 0 and 100.It represents the variance explained by the pluri-annual oscillations, from one to five-year cycles. High values indicate the presence of pluri-annual cycles.

Share and Cite

MDPI and ACS Style

Recuero, L.; Litago, J.; Pinzón, J.E.; Huesca, M.; Moyano, M.C.; Palacios-Orueta, A. Mapping Periodic Patterns of Global Vegetation Based on Spectral Analysis of NDVI Time Series. Remote Sens. 2019, 11, 2497. https://0-doi-org.brum.beds.ac.uk/10.3390/rs11212497

AMA Style

Recuero L, Litago J, Pinzón JE, Huesca M, Moyano MC, Palacios-Orueta A. Mapping Periodic Patterns of Global Vegetation Based on Spectral Analysis of NDVI Time Series. Remote Sensing. 2019; 11(21):2497. https://0-doi-org.brum.beds.ac.uk/10.3390/rs11212497

Chicago/Turabian Style

Recuero, Laura, Javier Litago, Jorge E. Pinzón, Margarita Huesca, Maria C. Moyano, and Alicia Palacios-Orueta. 2019. "Mapping Periodic Patterns of Global Vegetation Based on Spectral Analysis of NDVI Time Series" Remote Sensing 11, no. 21: 2497. https://0-doi-org.brum.beds.ac.uk/10.3390/rs11212497

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop