English
 
Privacy Policy Disclaimer
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT
  A Spark-Based Platform to Extract Phenological Information from Satellite Images

Bakayov, V., Goncalves, R., Zurita-Milla, R., Izquierdo-Verdiguier, E. (2018): A Spark-Based Platform to Extract Phenological Information from Satellite Images - Proceedings, 2018 IEEE 14th International Conference on e-Science (e-Science) (Amsterdam, Netherlands 2018).
https://doi.org/10.1109/eScience.2018.00095

Item is

Files

show Files

Locators

show

Creators

show
hide
 Creators:
Bakayov, Viktor1, Author
Goncalves, Romulo2, Author              
Zurita-Milla, Raul1, Author
Izquierdo-Verdiguier, Emma1, Author
Affiliations:
1External Organizations, ou_persistent22              
20 Pre-GFZ, Departments, GFZ Publication Database, Deutsches GeoForschungsZentrum, ou_146023              

Content

show
hide
Free keywords: Measurement; Vegetation mapping; Spatial resolution; Meteorology; Time series analysis; Satellites; Land surface
 Abstract: Phenology is the study of periodic plant and animal life cycle events and how these are influenced by seasonal and inter-annual variations in weather and climate, as well as in other environmental factors. Time series of remote sensing (RS) images can be used to characterize land surface phenology at continental to global scales. For this, the RS images are typically transformed into various vegetation indices (VI) such as the normalized difference vegetation index (NDVI) or the enhanced vegetation index (EVI). These indices can then be used to extract various phenological metrics. In our previous work we used cloud computing to generate temperature-based phenological indices [1], [2], and to relate one phenological metric, namely the Start-of-Season (SOS), with those indices [3], [4]. Here we present an extension of our work where we use a Spark-based platform to efficiently extract phenological metrics from time series of NDVI and EVI. This platform allows obtaining and analyzing high spatial resolution metrics (in this case 1km) from 10-day composites. The platform uses the same architecture as in [3], i.e., it is organized into three layers: a storage layer, a processing layer, and JupyterHub services for user-interaction. It is designed to store the data in well-known file formats like GeoTiffs and Hierarchical Data Format (HDF). For the data analysis the user expresses the operations in Jupyter notebooks as Python, R, or Scala code (Fig. 1). Hence, with a browser and remote connection, the user can express a research question and/or collect insights from large data sets. All computations are pushed down to the computational platform, and results fetched back for data visualization. To extract the phenological metrics, we rely on TimeSat [5]. TimeSat is a software package that can be used to fit a function (e.g. double logistic) to time series of VIs. After that, it uses various approaches to extract vegetation seasonality metrics such as SOS. The programs numerical and graphical routines are coded in Matlab and Fortran. These routines are highly vectorized and efficient for use with large data sets. However, distributed processing is required to determine SOS at continental scales. Through an efficient partition of the data, and Spark’s scheduling policies, these single-core routines are scheduled for parallel execution over multiple machines. The study evaluates which VIs and fitting functions are most suitable for certain vegetation types by comparing the SOS metrics to volunteered phenological observations curated by the USA national phenological network [6]. Our preliminary results show there can be up to 20-30 days differences in the SOS depending on the fitting function, the VI and the approach used to extract the SOS metric. In the South, SOS is around mid-February or March whereas in mountainous regions and the North, the SOS can be as late as June-July. We are to further evaluate how our results compare to the ground volunteered observations. This work is then a first stepping stone towards being able to systematically analyze and map the impact of climate change on the seasonality of plants. Our tests show that the platform is scalable and can be extended to work with even higher resolution VIs, such as those that can be derived from Sentinel-2 images (10 m resolution). Because of this, our work opens the door to studies at continental to global scales, and to the use of high and very high spatial resolution data.

Details

show
hide
Language(s): eng - English
 Dates: 2018-12-272018
 Publication Status: Finally published
 Pages: -
 Publishing info: -
 Table of Contents: -
 Rev. Type: -
 Identifiers: DOI: 10.1109/eScience.2018.00095
 Degree: -

Event

show
hide
Title: 2018 IEEE 14th International Conference on e-Science (e-Science)
Place of Event: Amsterdam, Netherlands
Start-/End Date: 2018-10-29 - 2018-11-01

Legal Case

show

Project information

show

Source 1

show
hide
Title: Proceedings
Source Genre: Proceedings
 Creator(s):
Affiliations:
Publ. Info: -
Pages: - Volume / Issue: - Sequence Number: - Start / End Page: - Identifier: -