1. ASSESSING VARIABILITY IN CLIMATE DATA: A SIGNIFICANT EVENT VIEWER TOOL
Iryna Rozum, Baudouin Raoult and Dick Dee i
European Centre for Mid-Range Weather Forecast, Shinfield Park, Reading RG2 9AX, UK
ABSTRACT
The EU-funded CHARMe project addresses the major
difficulty faced by users of climate data when judging
whether data are fit for purpose, by introducing the concept
of ‘Commentary’ metadata. The Significant Event Viewer
tool, together with a database of events, is being developed
as part of CHARMe tools. It will allow users to interactively
browse and visualize time series climate data with their
associated events and to determine whether the variability
and features seen in the dataset are likely to be artefacts of
the measurement of processing steps, or real changes in the
environment. The tool will also link events and climate
datasets to the commentary metadata via the CHARMe
system.
Index Terms— Climate services, metadata, linked data,
data sharing, significant events, visualization, graphical tool
1. INTRODUCTION
Climate variability and change exert considerable influences
on human and natural systems. This drives the scientific
quest for understanding of how climate behaved in the past
and how it will behave in the future. The users of climate
data are not just scientific researchers but also resource
managers and policy-makers who rely on climate data to
develop their strategies. With this comes the responsibility
to curate and share climate data more freely and usefully
than ever before.
There are many different types of climate data,
encompassing in-situ and remotely sensed observations,
output of numerical models, and the combination of both in
the reanalysis. The volume of worldwide climate data is
expanding rapidly, creating challenges not only for physical
archiving and sharing, but also for access and finding what’s
needed especially for users who are not climate scientists.
Another major difficulty faced by users of climate data
today is to be able to judge on the quality of this data due to
a lack of comprehensive and easily accessible supportive
information about datasets. The EU-funded CHARMe
project (“Characterization of metadata to enable high-
quality climate applications and services”) addresses this by
creating a repository of commentary metadata and a robust
framework (CHARMe node, CHARMe plugin, CHARMe
Maps and the Significant Event Viewer) for linking climate
datasets with commentary metadata. The CHARMe system
is described elsewhere [1] and will not be repeated here.
This presentation will describe the concept and
development of the Significant Event Viewer as part of the
CHARMe tools.
2. MOTIVATION
One of the important types of climate data is reanalysis -
hybrid model-observational data sets created by assimilating
observations into a global or regional forecast model for a
given time period. Input data for reanalysis comes from all
possible sources, including the ever growing amount of data
from satellite instruments. The use of satellite data in
climate observations present it’s own challenges as
individual satellites and their instruments have relatively
short life spans over which their orbits and sensitivities can
change. Degradation of sensors from exposure to space
environment, problems with telemetry and ground stations
availability can all affect the quality of data. Satellite data
assimilation requires the use of advanced data-processing
techniques. The resulting data are prone to being re-
processed as previously unknown problems are discovered
over time. In addition, gaps in the record and systematic
errors between satellites or a lack of overlapping calibration
periods make the construction of climate data a challenge.
At the same time the ever increasing use of satellites in
climate observations has made the predictions much better
[3].
The ECMWF’s1
ERA-CLIM reanalysis project [2]
produced several climate data products including the 111
years Observation Feedback Archive which contains all the
Earth observations processed by the reanalysis, as well as,
for each of them, feedback information generated by the
data assimilation system. Figure 1 shows global temperature
changes in ERA-Interim and ERA-40 datasets. The strong
signals indicated by circles are not a signature of a global
climate shift but are caused by: dashed line - volcanic
eruptions El Chichon and Pinatubo, dotted line - start of
AMSU-A instrument on NOAA-15 satellite, solid line -
large El Niño event.
This has motivated us to complement the reanalysis data
with a database of significant events, such as volcanic
1
European Centre for Mid-Range Weather Forecast
2. Figure 1: Temperature changes in ERA-CLIM
reanalysis datasets at different levels. Circles show
prominent signals - sharp temperature increases -
caused by external events.
eruptions, the launch of new satellites or El Niño phases,
etc. A web-based graphical tool (the Significant Event
Viewer) will link climate signals to significant event data,
and will allow users to interactively browse and visualize
time series climate data with their associated events. This
will provide users with the opportunity to become more
familiar with the variety of observations that feed into the
reanalysis, and to determine whether the variability and
features seen in the dataset were likely to be artefacts of the
measurements or processing steps, or real changes in the
environment.
3. WHAT ARE SIGNIFICANT EVENTS
Significant events are any external events that can affect
climate data. For the purpose of CHARMe project
significant events are categorized into:
1. Climate events, e.g. hurricanes, volcanic eruptions, El
Niño, etc.
2. Software events, e.g. ECMWF’s Integrated Forecast
System (IFS) software cycle upgrades.
Figure 2: Example of reanalysis time series data with the
timeline of significant events plotted using the
Significant Event Viewer.
3. Operational events, e.g. satellite or instrument failure,
operational changes to satellite orbit calculations, etc.
4. Data/Observing system events, e.g. data source, alarms,
etc.
More categories can be added based on user feedback.
Significant events are stored in a database which is
constantly evolving and updated with new data.
CHARMe commentary metadata are modeled on the
W3C Open Annotation data model [1]. Within this model
significant events are not annotations, they can be either
targets or bodies and themselves require a separate data
model. The PROV-O data model’s Activity class
(http://www.w3.org/TR/prov-o/) complemented with SKOS
vocabulary was used to model significant events.
4. THE SIGNIFICANT EVENT VIEWER
The significant event viewer tool is a web based graphical
tool for visualizing climate time series data with their
associated events. Although this tool is focusing on
reanalysis datasets, it is designed to be general enough to be
extended to other datasets and user needs.
Using this tool the user will be able to visually correlate
climate signals with significant events in order to determine
the cause of these signals. The user will be able to zoom in
on a particular feature in time series data, to obtain more
information about events and to list all events. The tool will
provide a mean to link significant events and climate
datasets to commentary metadata via the CHARMe plugin,
3. and, hence, to comment on the events and datasets. Figure 2
is the example of a climate time-series plot with the event
timeline produced by the tool.
5. CONCLUSIONS
Climate data volumes have increased dramatically in recent
years, from terabytes to many petabytes. Such large volumes
of information have many issues related to Big Data, such as
data volumes, data quality, variety and complexity. The
CHARMe project is concentrating on understanding variety
and veracity (data quality) by providing users with the
system to create additional information about climate
datasets. The Significant Event Viewer tool is harnessing
modern technologies to help users understand signals
present in climate data products thus contributing to
understanding data quality, variety and complexity.
6. REFERENCES
[1] D. Clifford, J. Blower, R. Alegre, and R. Phipps, “Annotating
Climate Data with Commentary: The CHARMe Project” Big Data
From Space conference proceedings, 2014, submitted
[2] Towards the Consistent Reanalysis of the Climate System, SAC
41st Session report, ECMWF, Reading UK, 14 September 2012
[3] ] M. Dowell, P. Lecomte, R. Husband, J. Schultz, T. Mohr, Y.
Tahara, E. Lindstrom, C. Wooldridge, S. Hilding, J. Bates, B.
Ryan, J. Lafeuille and S. Bojinski, “2013: Strategy Towards an
Architecture for Climate Monitoring from Space”, ESA, pp. 39,
2013
i
This paper was prepared on behalf of the CHARMe
consortium