Analysis of vegetation-activity trends in a global land degradation framework

R. de Jong

Research output: Thesisinternal PhD, WU

Abstract

 Land degradation is a global issue on a par with climate change and loss of biodiversity, but its extent and severity are only roughly known and there is little detail on the immediate processes – let alone the drivers. Earth-observation methods enable monitoring of land resources in a consistent, physical way and on global scale by making use of vegetation activity and/or cover as proxies. A well-known spectral proxy is the normalized difference vegetation index (NDVI), which is available in high temporal resolution time series since the early 1980s. In this work, harmonic analyses and non-parametric trend tests were applied to the GIMMS NDVI dataset (1981–2008) in order to quantify positive changes (or greening) and negative changes (browning). Phenological shifts and variations in length of growing season were accounted for using analysis by vegetation development stage rather than by calendar day. This approach does not rely on temporal aggregation for elimination of seasonal variation. The latter might introduce artificial trends as demonstrated in the chapter on the modifiable temporal unit problem. Still, a major assumption underlying the analysis is that trends were invariant, i.e. linear or monotonic, over time. However, these monotonic trends in vegetation activity may consist of an alternating sequence of greening and/or browning periods. This effect and the contribution of short-term trends to longer-term change was analysed using a procedure for detection of trend breaks. Both abrupt and gradual changes were found in large parts of the world, especially in (semi-arid) shrubland and grassland. Many abrupt changes were found around large-scale natural influences like the Mt Pinatubo eruption in 1991 and the strong 1997/98 El Niño event. This marks the importance of accounting for trend changes in the analysis of long-term NDVI time series. These new change-detection techniques advance our understanding of vegetation variability at a multi-decadal scale, but do not provide links to driving processes. It is very complex to disentangle all natural and human drivers and their interactions. As a first step, the spatial relation between changes in climate parameters and changes in vegetation activity was addressed in this work. It appeared that a substantial proportion (54%) of the spatial variation in NDVI changes could be associated to climatic changes in temperature, precipitation and incident radiation, especially in forest biomes. In other regions, the lack of such associations might be interpreted as human-induced land degradation. With these steps we demonstrated the value of global satellite records for monitoring land resources, although many steps are still to be taken.

Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Wageningen University
Supervisors/Advisors
  • Schaepman, Michael, Promotor
  • de Bruin, Sytze, Co-promotor
Award date11 Jun 2012
Place of PublicationS.l.
Publisher
Print ISBNs9789461733122
Publication statusPublished - 2012

Fingerprint

land degradation
vegetation
NDVI
time series
climate change
shrubland
monitoring
resource
biome
trend
analysis
long-term change
growing season
spatial variation
volcanic eruption
seasonal variation
grassland
biodiversity
climate

Keywords

  • vegetation
  • vegetation indices
  • land degradation
  • mapping
  • monitoring
  • observation
  • exploration
  • climate
  • seasonal variation
  • satellite imagery
  • remote sensing

Cite this

@phdthesis{5cec9078f2344ed2bbf9736e31b0f0a9,
title = "Analysis of vegetation-activity trends in a global land degradation framework",
abstract = " Land degradation is a global issue on a par with climate change and loss of biodiversity, but its extent and severity are only roughly known and there is little detail on the immediate processes – let alone the drivers. Earth-observation methods enable monitoring of land resources in a consistent, physical way and on global scale by making use of vegetation activity and/or cover as proxies. A well-known spectral proxy is the normalized difference vegetation index (NDVI), which is available in high temporal resolution time series since the early 1980s. In this work, harmonic analyses and non-parametric trend tests were applied to the GIMMS NDVI dataset (1981–2008) in order to quantify positive changes (or greening) and negative changes (browning). Phenological shifts and variations in length of growing season were accounted for using analysis by vegetation development stage rather than by calendar day. This approach does not rely on temporal aggregation for elimination of seasonal variation. The latter might introduce artificial trends as demonstrated in the chapter on the modifiable temporal unit problem. Still, a major assumption underlying the analysis is that trends were invariant, i.e. linear or monotonic, over time. However, these monotonic trends in vegetation activity may consist of an alternating sequence of greening and/or browning periods. This effect and the contribution of short-term trends to longer-term change was analysed using a procedure for detection of trend breaks. Both abrupt and gradual changes were found in large parts of the world, especially in (semi-arid) shrubland and grassland. Many abrupt changes were found around large-scale natural influences like the Mt Pinatubo eruption in 1991 and the strong 1997/98 El Ni{\~n}o event. This marks the importance of accounting for trend changes in the analysis of long-term NDVI time series. These new change-detection techniques advance our understanding of vegetation variability at a multi-decadal scale, but do not provide links to driving processes. It is very complex to disentangle all natural and human drivers and their interactions. As a first step, the spatial relation between changes in climate parameters and changes in vegetation activity was addressed in this work. It appeared that a substantial proportion (54{\%}) of the spatial variation in NDVI changes could be associated to climatic changes in temperature, precipitation and incident radiation, especially in forest biomes. In other regions, the lack of such associations might be interpreted as human-induced land degradation. With these steps we demonstrated the value of global satellite records for monitoring land resources, although many steps are still to be taken.",
keywords = "vegetatie, vegetatie-indexen, landdegradatie, cartografie, monitoring, observatie, exploratie, klimaat, seizoenvariatie, satellietbeelden, remote sensing, vegetation, vegetation indices, land degradation, mapping, monitoring, observation, exploration, climate, seasonal variation, satellite imagery, remote sensing",
author = "{de Jong}, R.",
note = "WU thesis 5256",
year = "2012",
language = "English",
isbn = "9789461733122",
publisher = "s.n.",
school = "Wageningen University",

}

de Jong, R 2012, 'Analysis of vegetation-activity trends in a global land degradation framework', Doctor of Philosophy, Wageningen University, S.l..

Analysis of vegetation-activity trends in a global land degradation framework. / de Jong, R.

S.l. : s.n., 2012. 147 p.

Research output: Thesisinternal PhD, WU

TY - THES

T1 - Analysis of vegetation-activity trends in a global land degradation framework

AU - de Jong, R.

N1 - WU thesis 5256

PY - 2012

Y1 - 2012

N2 -  Land degradation is a global issue on a par with climate change and loss of biodiversity, but its extent and severity are only roughly known and there is little detail on the immediate processes – let alone the drivers. Earth-observation methods enable monitoring of land resources in a consistent, physical way and on global scale by making use of vegetation activity and/or cover as proxies. A well-known spectral proxy is the normalized difference vegetation index (NDVI), which is available in high temporal resolution time series since the early 1980s. In this work, harmonic analyses and non-parametric trend tests were applied to the GIMMS NDVI dataset (1981–2008) in order to quantify positive changes (or greening) and negative changes (browning). Phenological shifts and variations in length of growing season were accounted for using analysis by vegetation development stage rather than by calendar day. This approach does not rely on temporal aggregation for elimination of seasonal variation. The latter might introduce artificial trends as demonstrated in the chapter on the modifiable temporal unit problem. Still, a major assumption underlying the analysis is that trends were invariant, i.e. linear or monotonic, over time. However, these monotonic trends in vegetation activity may consist of an alternating sequence of greening and/or browning periods. This effect and the contribution of short-term trends to longer-term change was analysed using a procedure for detection of trend breaks. Both abrupt and gradual changes were found in large parts of the world, especially in (semi-arid) shrubland and grassland. Many abrupt changes were found around large-scale natural influences like the Mt Pinatubo eruption in 1991 and the strong 1997/98 El Niño event. This marks the importance of accounting for trend changes in the analysis of long-term NDVI time series. These new change-detection techniques advance our understanding of vegetation variability at a multi-decadal scale, but do not provide links to driving processes. It is very complex to disentangle all natural and human drivers and their interactions. As a first step, the spatial relation between changes in climate parameters and changes in vegetation activity was addressed in this work. It appeared that a substantial proportion (54%) of the spatial variation in NDVI changes could be associated to climatic changes in temperature, precipitation and incident radiation, especially in forest biomes. In other regions, the lack of such associations might be interpreted as human-induced land degradation. With these steps we demonstrated the value of global satellite records for monitoring land resources, although many steps are still to be taken.

AB -  Land degradation is a global issue on a par with climate change and loss of biodiversity, but its extent and severity are only roughly known and there is little detail on the immediate processes – let alone the drivers. Earth-observation methods enable monitoring of land resources in a consistent, physical way and on global scale by making use of vegetation activity and/or cover as proxies. A well-known spectral proxy is the normalized difference vegetation index (NDVI), which is available in high temporal resolution time series since the early 1980s. In this work, harmonic analyses and non-parametric trend tests were applied to the GIMMS NDVI dataset (1981–2008) in order to quantify positive changes (or greening) and negative changes (browning). Phenological shifts and variations in length of growing season were accounted for using analysis by vegetation development stage rather than by calendar day. This approach does not rely on temporal aggregation for elimination of seasonal variation. The latter might introduce artificial trends as demonstrated in the chapter on the modifiable temporal unit problem. Still, a major assumption underlying the analysis is that trends were invariant, i.e. linear or monotonic, over time. However, these monotonic trends in vegetation activity may consist of an alternating sequence of greening and/or browning periods. This effect and the contribution of short-term trends to longer-term change was analysed using a procedure for detection of trend breaks. Both abrupt and gradual changes were found in large parts of the world, especially in (semi-arid) shrubland and grassland. Many abrupt changes were found around large-scale natural influences like the Mt Pinatubo eruption in 1991 and the strong 1997/98 El Niño event. This marks the importance of accounting for trend changes in the analysis of long-term NDVI time series. These new change-detection techniques advance our understanding of vegetation variability at a multi-decadal scale, but do not provide links to driving processes. It is very complex to disentangle all natural and human drivers and their interactions. As a first step, the spatial relation between changes in climate parameters and changes in vegetation activity was addressed in this work. It appeared that a substantial proportion (54%) of the spatial variation in NDVI changes could be associated to climatic changes in temperature, precipitation and incident radiation, especially in forest biomes. In other regions, the lack of such associations might be interpreted as human-induced land degradation. With these steps we demonstrated the value of global satellite records for monitoring land resources, although many steps are still to be taken.

KW - vegetatie

KW - vegetatie-indexen

KW - landdegradatie

KW - cartografie

KW - monitoring

KW - observatie

KW - exploratie

KW - klimaat

KW - seizoenvariatie

KW - satellietbeelden

KW - remote sensing

KW - vegetation

KW - vegetation indices

KW - land degradation

KW - mapping

KW - monitoring

KW - observation

KW - exploration

KW - climate

KW - seasonal variation

KW - satellite imagery

KW - remote sensing

M3 - internal PhD, WU

SN - 9789461733122

PB - s.n.

CY - S.l.

ER -