2013_Feb05_AM_AsefiNajafabady_202

Report
High-resolution global CO2 emissions from fossil fuel
inventories for 1992 to 2010 using integrated in-situ
and remotely sensed data in a fossil fuel data
assimilation system
Salvi Asefi1, K. R. Gurney1, P. Rayner2, Y. Song1,
K. Coltin1, C. D. Elvidge3, K. Baugh3, A. Mcrobert2
1- Arizona State University, School of Life Sciences
2- School of Earth Sciences, University of Melbourne
3- NOAA-NESDIS National Geophysical Data Center
Introduction

accurate global quantification of FFCO2 with high space/time
resolution accompanied by uncertainty is a critical need within
the carbon cycle science community.

There is a need for functional or process-based quantification.
• This provides better space/time resolution (can avail of sectorspecific space/time proxies)
• Potential for multiple uses (energy analysis, growth morphology)
Our answer: Fossil Fuel Data Assimilation (FFDAS) system to
create a global high temporal/spatial resolution fossil fuel
CO2 emission inventory with uncertainties
See Rayner et
al., 2010
Current FFCO2 emission datasets
Vulcan
Vulcan data product:
• Gridded to 10 km x 10 km, hourly, year 2002
• Includes process detail for all sectors of the
U.S economy (on-road, non-road, industrial,
commercial, residential, cement production,
airport, power production, aircraft).
……detailed bottom-up info is rarely available at
global scale………….
Other global data products have employed
population and nightlights to downscale national
emissions. These efforts have begun to use other
datasets such as power plants emissions and spatial
proxies such as road maps.
ODIAC
Fossil Fuel Data Assimilation System (FFDAS)
In contrast to downscaling national emissions we utilize the Fossil Fuel Data
Assimilation System (FFDAS) which has a dynamical model at its
core………….the Kaya Identity:
F = emissions,
P = areal population density
g = per capita economic activity
e = energy intensity of economic activity
f = carbon intensity of energy consumption
F=Pgef
Data assimilation is applied to constrain components of Kaya with a number of
observational operators.
Advantages of data assimilation to downscaling techniques:
 Process-based dynamical model at core
 Smoother spatial distribution
 The ability to integrate the range of observations
 The ability to include prior uncertainty and estimate posterior uncertainties
 Ability to perform at different spatial and temporal scales
Inputs
National emissions:

National and global FFCO2 are constrained by FFCO2 sectoral
emissions reported by International Energy Agency IEA and Carbon
Dioxide Information and Analysis Center (CDIAC).

Prior uncertainties for national emissions were also objectively
estimated and included in FFDAS (see next talk).
Per Country CO2 Emissions (CDIAC)
Inputs
Population:

SEDAC global gridded population dataset (0.04° resolution, 1995, 2000,
2005 & 2010) combined with LandScan global gridded population dataset
(30 arc second resolution, 2004, 2006, 2007, 2008, 2010)

Result: population dataset from 1997 to 2010 at 30 arc second
resolution.
SEDAC population density
LandScan population density
Germany
Germany
France
Spain
France
Spain
Inputs
Nightlights:


Nightlight is a global remote sensing
product provided by NOAA-NGDC at 30 arc
second resolutions (1992-2010).
However this dataset is subject to
instrumental saturation meaning areas of
bright nightlights, such as urban cores are
often underestimated.

Saturation has been addressed by NGDC
and a new unsaturated dataset has been
created for five years (1997, 1999, 2003,
2006 and 2010) at 30 arc second resolution.

Linear interpolation applied to estimate
unsaturated values for all years from 1997 to
2010.
Nightlights
Nightlight(unsaturated)
(saturated) - 0.1deg
Nightlight
- 0.1deg
Inputs
Power plant point sources:

Currently the only available global
dataset is CARMA. That includes
more than 60000 power plants
worldwide.

CARMA provides plant location and
estimated CO2 emission for each
power plant.

We are finding sizeable
biases…….will discuss in next talk
FFDAS Results
Results represent annual emissions, 1997 - 2010 at the global scale and
spatial resolutions of 0.1° x 0.1° (FFDAS v.2)
Can produce any resolution – land/sea mask is critical - coastal shuffling.
FFDAS fossil fuel emissions in 2010 at 0.1°
Year=2001
Year=2002
Year=2003
Year=2009
Year=1998
Year=1999
Year=
2000
Year=2007
Year=2006
Year=2004
Year=2005
Year=2008
Year=2010
FFDAS fossil fuel emissions
FFDAS Results
Comparison between 0.1° resolutions and 0.25°
FFCO2 emission 0.25°
FFDAS v1
FFCO2 emission 0.1°
FFDAS v2
FFDAS Results
Inclusion of power plant emission. A major improvement from
FFDAS v.1*

Power plants are major global CO2 emission sources (40% of
global emissions).
No Power plants
FFDAS v.1
Power plants included
FFDAS v.2
*(Rayner et al. 2010)
New power plant data product - Ventus
Given the importance of power plants to the results (they have no
spatial proxy & they are a large component of total)…………….
We are building a new power plant CO2 data product:

Improving locations & emissions via national datasets and GE search.
 New predictive model utilizing multiple national datasets
 Providing uncertainty for each individual power plant
Poster 249,
Wednesday (in
pavilion)
Ventus crowd sourcing
effort:
An interactive website engaging
individuals and institutions to
help us improve our knowledge
of the power plant emissions
and locations.
Comparisons with Vulcan
At 0.5° resolution
FFDAS v.1 with VULCAN
Correlation =0.74
Difference map between FFDAS v.2
and Vulcan at 0.1°
FFDAS v.2 with VULCAN
Correlation =0.92
----------------------------------At 0.1° resolution
FFDAS v.2 with VULCAN
Correlation =0.61
Improvements under development for future versions of FFDAS:
• Other observational operators will be included: roads, airports, industrial
point sources, aviation routes, impervious surface, etc.
• Temporal resolution at hourly timescale using TIMES (Nassar et al.)
among others.
• Spatial resolutions of 1km and higher
Conclusions
Data assimilation is powerful approach to building an optimized fossil
fuel CO2 emission inventory at regional and global scales.
Fossil fuel data assimilation system (FFDAS) approach:




Follows an underlying dynamical model (Kaya identity) that takes into
account the relationship between all the elements that contribute to FFCO2
emissions
Enables the use of prior uncertainties and estimates posterior uncertainties
Has the ability to integrate various layers of observations
Can perform at high temporal & spatial resolutions
Integration with Hestia & Vulcan & satellite RS shows promise
We have a preliminary data product at annual timestep from 1997 to
2010 at 0.1 degrees resolution
Improved data product rolled out in coming months
Acknowledgment:
This project is supported through NASA grant
NNX11AH86G
THANK YOU!
FFDAS fossil fuel emissions, 2010
FFDAS fossil fuel emissions, 2010
FFDAS fossil fuel emissions, 2010
FFDAS fossil fuel emissions, 2010
FFDAS fossil fuel emissions, 2010

similar documents