Advanced

Analysing the vegetation condition during the 2017 and 2018 growing seasons using indices derived from Sentinel-2 data : a case study over southern Sweden

Persson, Camilla LU (2019) In Student thesis series INES NGEM01 20191
Dept of Physical Geography and Ecosystem Science
Abstract
The summer of 2017 was climatically close to normal in southern Sweden, whereas the following summer was unusually warm and dry. These two years therefore make an interesting case study for investigating the impact of severe drought on vegetation, particularly when considering that climate change is predicted to lead to an increased frequency of drought events in the study area.

The comparison was done by calculating vegetation indices (VI) based on satellite imagery. The calculated indices are Normalised Vegetation Index (NDVI), Enhanced Vegetation Index (EVI) and Normalised Difference Water Index (NDWI) based on four Sentinel-2 images from 2017 and four images from 2018. Rain Use Efficiency (RUE) was calculated based on NDVI and EVI... (More)
The summer of 2017 was climatically close to normal in southern Sweden, whereas the following summer was unusually warm and dry. These two years therefore make an interesting case study for investigating the impact of severe drought on vegetation, particularly when considering that climate change is predicted to lead to an increased frequency of drought events in the study area.

The comparison was done by calculating vegetation indices (VI) based on satellite imagery. The calculated indices are Normalised Vegetation Index (NDVI), Enhanced Vegetation Index (EVI) and Normalised Difference Water Index (NDWI) based on four Sentinel-2 images from 2017 and four images from 2018. Rain Use Efficiency (RUE) was calculated based on NDVI and EVI and a precipitation data set from PERSIANN-CCS, in an attempt to measure drought resilience. The data was extracted for different land covers and crop types to determine where the largest differences were seen in the study area.

It was found that the land cover with the smallest decreases (and some increases) in VI values between the two years was mixed forest. Both coniferous and deciduous forest had more negative changes between the two years; this suggests that the combination of tree types could increase the drought resilience. The land cover with the largest decrease was agricultural land which had decreasing values for all VI’s, suggesting that this might be the least drought resilient land cover.

The break-down of the changes by crops showed that part of the decrease in VI values between 2017 and 2018 in agricultural land could be explained by a shift from winter crops to spring crops due to the late harvest in 2017. It could also be a result of the satellite imagery from the different years being taken at different dates, and capturing different stages of the growing cycle which would affect the results of cereal crops and rapeseed in particular. The largest decrease in VI values were seen in winter wheat and spring barley which indicates that these might be more drought sensitive. Both RUE and the standard deviation of RUE increased for the whole study area between the two years, which is likely a result of the large amount of precipitation amounts in 2017. (Less)
Popular Abstract (Swedish)
Sommaren 2017 uppfattades av många som ovanligt blöt och kall, men var i själva verket nära det normala i södra Sverige, medan sommaren efter (2018) var ovanligt varm och torr. Dessa två år utgör därför en intressant fallstudie för att undersöka hur extrem torka påverkar vegetationen i södra Sverige. Jämförelsen är särskilt intressant eftersom klimatförändringarna bland annat förväntas leda till en ökad frekvens av torka i området.

Jämförelsen mellan de två åren gjordes genom att räkna ut vegetations index (VI) baserat på satellitbilder för att mäta vegetationens hälsa och vätskeinnehåll. Uträkningarna baserades på åtta bilder från Sentinel-2, fyra från 2017 och fyra från 2018. De index som användes är Normalised Vegetation Index... (More)
Sommaren 2017 uppfattades av många som ovanligt blöt och kall, men var i själva verket nära det normala i södra Sverige, medan sommaren efter (2018) var ovanligt varm och torr. Dessa två år utgör därför en intressant fallstudie för att undersöka hur extrem torka påverkar vegetationen i södra Sverige. Jämförelsen är särskilt intressant eftersom klimatförändringarna bland annat förväntas leda till en ökad frekvens av torka i området.

Jämförelsen mellan de två åren gjordes genom att räkna ut vegetations index (VI) baserat på satellitbilder för att mäta vegetationens hälsa och vätskeinnehåll. Uträkningarna baserades på åtta bilder från Sentinel-2, fyra från 2017 och fyra från 2018. De index som användes är Normalised Vegetation Index (NDVI), Enhanced Vegetation Index (EVI) och Normalised Difference Water Index (NDWI). Rain Use Efficiency (RUE) räknades ut baserat på NDVI och EVI samt nederbördsdata från PERSIANN-CCS, i ett försök att mäta hur resistent vegetationen var mot torka. Data extraherades för olika marktäcken och grödor för att undersöka var de största skillnaderna mellan åren kunde ses.

Marktäcket med minst minskning i (och viss ökning) i vegetations index (VI) värden mellan de två åren var blandad barr- och lövskog. Både barrskog och lövskog hade större negativa skillnader än blandad skog; vilken pekar mot att kombinationen av barr- och lövskog skulle kunna öka resistensen mot torka. Marktäcket med den största minskningen i VI värden var odlad mark, där värdena var lägre för alla VI, vilket indikerar att det är det marktäcke som är mest känsligt mot torka.

Vidare undersökning av skillnaderna i VI värden för olika grödor visade att en del av de skillnader som sågs i odlad mark kan förklaras med att det odlades mer vårgrödor 2018 än 2017 på grund av den sena skörden 2017. Men skillnaderna kan också bero på att satellitbilderna är tagna vid olika tidpunkter under de två åren, och därför kan ha fångat växterna i olika utvecklingsskeden. Det här skulle framförallt kunna påverka resultaten för spannmål och raps. Den största minskningen av VI värden kunde ses i höstvete och vårkorn, vilket skulle kunna indikera att de är mer känsliga mot torka än andra grödor. Både RUE och standardavvikelsen i RUE ökade i hela studieområdet, vilket troligen beror på den stora mängden nederbörd under 2017. (Less)
Please use this url to cite or link to this publication:
author
Persson, Camilla LU
supervisor
organization
course
NGEM01 20191
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Physical Geography and Ecosystem analysis, remote sensing, vegetation indices, drought, Geomatics
publication/series
Student thesis series INES
report number
496
language
English
id
8996782
date added to LUP
2019-10-18 17:43:41
date last changed
2019-10-18 17:43:41
@misc{8996782,
  abstract     = {The summer of 2017 was climatically close to normal in southern Sweden, whereas the following summer was unusually warm and dry. These two years therefore make an interesting case study for investigating the impact of severe drought on vegetation, particularly when considering that climate change is predicted to lead to an increased frequency of drought events in the study area.

The comparison was done by calculating vegetation indices (VI) based on satellite imagery. The calculated indices are Normalised Vegetation Index (NDVI), Enhanced Vegetation Index (EVI) and Normalised Difference Water Index (NDWI) based on four Sentinel-2 images from 2017 and four images from 2018. Rain Use Efficiency (RUE) was calculated based on NDVI and EVI and a precipitation data set from PERSIANN-CCS, in an attempt to measure drought resilience. The data was extracted for different land covers and crop types to determine where the largest differences were seen in the study area. 

It was found that the land cover with the smallest decreases (and some increases) in VI values between the two years was mixed forest. Both coniferous and deciduous forest had more negative changes between the two years; this suggests that the combination of tree types could increase the drought resilience. The land cover with the largest decrease was agricultural land which had decreasing values for all VI’s, suggesting that this might be the least drought resilient land cover.

The break-down of the changes by crops showed that part of the decrease in VI values between 2017 and 2018 in agricultural land could be explained by a shift from winter crops to spring crops due to the late harvest in 2017. It could also be a result of the satellite imagery from the different years being taken at different dates, and capturing different stages of the growing cycle which would affect the results of cereal crops and rapeseed in particular. The largest decrease in VI values were seen in winter wheat and spring barley which indicates that these might be more drought sensitive. Both RUE and the standard deviation of RUE increased for the whole study area between the two years, which is likely a result of the large amount of precipitation amounts in 2017.},
  author       = {Persson, Camilla},
  keyword      = {Physical Geography and Ecosystem analysis,remote sensing,vegetation indices,drought,Geomatics},
  language     = {eng},
  note         = {Student Paper},
  series       = {Student thesis series INES},
  title        = {Analysing the vegetation condition during the 2017 and 2018 growing seasons using indices derived from Sentinel-2 data : a case study over southern Sweden},
  year         = {2019},
}