Advanced

Comparison and validation of five land cover products over the African continent

Bai, Ling (2010) In Lunds universitets Naturgeografiska institution - Seminarieuppsatser
Dept of Physical Geography and Ecosystem Science
Abstract
Earth surface has always been an interesting scientific study area. it is tightly
connected with other researches such as hydrology, ecosystem, atmosphere and
climate change. Among other options, satellite collected earth surface information has
several advantages including its objectivity, synoptic view, large scale and relatively
short time intervals. A series of satellite derived land cover products were generated to
provide for the science community, some of them are free of access to all users. This
study chooses five free land cover products over the African continent to find out the
similarities (differences) among them, and to find out their classification accuracy by
using Google Earth derived reference data. The five land... (More)
Earth surface has always been an interesting scientific study area. it is tightly
connected with other researches such as hydrology, ecosystem, atmosphere and
climate change. Among other options, satellite collected earth surface information has
several advantages including its objectivity, synoptic view, large scale and relatively
short time intervals. A series of satellite derived land cover products were generated to
provide for the science community, some of them are free of access to all users. This
study chooses five free land cover products over the African continent to find out the
similarities (differences) among them, and to find out their classification accuracy by
using Google Earth derived reference data. The five land cover products are: MODIS,
USGS, UMD, GLC2000 and GlobCover. The results of this study show
understandable agreement across the five land cover products. Discrepancies mainly
attributed to the different data source and generation process used by each land cover
product, furthermore the data pre-process procedures applied to these products in this
study may also share some responsibility. Accuracy assessment shows relatively low
accuracy of about 54% as an average. Reasons for the low accuracy could be the
quality and availability of reference data, and possible biases introduced during visual
interpretation and labeling of validation scenes. The findings of this study suggest
users be cautious about any particular land cover product, aware of their weakness
and strength in their applications. (Less)
Abstract
Scientific abstract: As the increasing attentions drawn to the environment and climate change issue, the
profound impact on the earth system made known to nearly every corner of the world.
A great variety of global protocols and conventions has been established aiming to
reduce global environmental risks and maintain a healthy earth system. In order to
monitor, supervise these protocols and conventions, science communities have
expressed their urgent need for accurate and up-to-date land cover database. Land
cover products play a very important role in various global change researches such as
climate change, biodiversity conservation, ecosystem assessment, environment
modeling and is in many aspects a foundation platform. This... (More)
Scientific abstract: As the increasing attentions drawn to the environment and climate change issue, the
profound impact on the earth system made known to nearly every corner of the world.
A great variety of global protocols and conventions has been established aiming to
reduce global environmental risks and maintain a healthy earth system. In order to
monitor, supervise these protocols and conventions, science communities have
expressed their urgent need for accurate and up-to-date land cover database. Land
cover products play a very important role in various global change researches such as
climate change, biodiversity conservation, ecosystem assessment, environment
modeling and is in many aspects a foundation platform. This study compares five
freely available and in common used satellite derived land cover products: MODIS
map, USGS-IGBP map, UMD map, GLC2000 map, GlobCover map. They are
produced by different working groups using different sources, characterization
algorithms, and different classification schemes for the same purpose namely to
provide accurate land surface data for the scientific communities. This study put these
land cover products together, compare them with each other and assess them against
an independent set of validation data aiming to find out how different they are, and
how well these land cover product represent the land cover of Africa. The result
shows reasonable agreement among the five land cover products, and reasonable over
all accuracies of 54%. Although substantial discrepancies do exist, to some extent
they can be explained by the data origins and data processing applied in this study.
From analyzing possible reasons that cause the differences,the aim of this study is to
try to provide an insight in these products, reveal their characteristics, explore their
strengths and weaknesses. Hopefully, it could serve as indications to potential users
and provide for future improvement of similar product. (Less)
Please use this url to cite or link to this publication:
author
Bai, Ling
supervisor
organization
year
type
H2 - Master's Degree (Two Years)
subject
keywords
land cover product validation, land cover products, geography, classification algorithm, classification legend
publication/series
Lunds universitets Naturgeografiska institution - Seminarieuppsatser
report number
195
language
English
id
1969905
date added to LUP
2011-05-26 12:05:51
date last changed
2011-12-21 10:54:13
@misc{1969905,
  abstract     = {Scientific abstract: As the increasing attentions drawn to the environment and climate change issue, the
profound impact on the earth system made known to nearly every corner of the world.
A great variety of global protocols and conventions has been established aiming to
reduce global environmental risks and maintain a healthy earth system. In order to
monitor, supervise these protocols and conventions, science communities have
expressed their urgent need for accurate and up-to-date land cover database. Land
cover products play a very important role in various global change researches such as
climate change, biodiversity conservation, ecosystem assessment, environment
modeling and is in many aspects a foundation platform. This study compares five
freely available and in common used satellite derived land cover products: MODIS
map, USGS-IGBP map, UMD map, GLC2000 map, GlobCover map. They are
produced by different working groups using different sources, characterization
algorithms, and different classification schemes for the same purpose namely to
provide accurate land surface data for the scientific communities. This study put these
land cover products together, compare them with each other and assess them against
an independent set of validation data aiming to find out how different they are, and
how well these land cover product represent the land cover of Africa. The result
shows reasonable agreement among the five land cover products, and reasonable over
all accuracies of 54%. Although substantial discrepancies do exist, to some extent
they can be explained by the data origins and data processing applied in this study.
From analyzing possible reasons that cause the differences,the aim of this study is to
try to provide an insight in these products, reveal their characteristics, explore their
strengths and weaknesses. Hopefully, it could serve as indications to potential users
and provide for future improvement of similar product.},
  author       = {Bai, Ling},
  keyword      = {land cover product validation,land cover products,geography,classification algorithm,classification legend},
  language     = {eng},
  note         = {Student Paper},
  series       = {Lunds universitets Naturgeografiska institution - Seminarieuppsatser},
  title        = {Comparison and validation of five land cover products over the African continent},
  year         = {2010},
}