Satellite data assimilation for air quality forecast

1
Satellite data assimilation for air quality
forecast
Journées ARC Grenoble - 18/10/2006
2
Objectives

• Assimilation of satellite measurements of
  troposphere chemistry, in view of improving the
  quality of forecast.
• Context
• ADOQA INRIA ARC
• Feasibility study concerning the future sensors
  GOME2 and IASI (to be launched in Oct 2006 on
  EPS/MetOp). PI of ESA/Eumetsat project.
• Future mission TRAQ (2012) for the evolution of
  air quality at regional (Europe) and global
  scales.
• Collaboration with IPSL/SA (C. Clerbaux)
  participated to the design of the IASI sensor.

3
Chemical measurements to be assimilated

• Ground-based monitoring network
• Operated locally.
• Irregularly scattered throughout Europe, good
  temporal sampling.
• Non uniform quality.
• Nature ground level concentrations, vertical
  profiles (LIDAR).
• Satellites
• Continuous improvement of satellite measurements
  of troposphere chemistry MOPITT, OMI (NASA),
  GOME, SCIAMACHY (ESA), futurs GOME2 and IASI
  (ESA).
• Regular spatial sampling (12 km), 1
  acquisition/day, uniform quality.
• Nature columns (O3, NOx, CH4, ), vertical
  profiles (O3, NOx), aerosols optical properties.

4
Potential of IASI acquisitions

• Considered measure 0-6km ozone column.
• What information it carries of the boundary
  layer? Contribution of boundary layer ozone to
  the column.
• Sensitivity (via model) of boundary layer O3 to
  modifications of upper troposphere O3.
• IASI assimilation experiments.

5
Contribution of boundary layer ozone

• Computed from a reference atmosphere Polyphemus
  analysis, July 2001.
• Mean contribution 14, larger during day than
  during night.
• Irregularly scattered in space and time.
• Small but not negligible.

0h
15h
6
Sensitivity to modification of upper troposphere
ozone

• Experiment
• Perturbation of reference modification of O3
  above 1500m.
• Perturbation of initial condition, or cyclic
  perturbation (simulating the assimilation of IASI
  data).
• Boundary layer O3 computed from Polyphemus and
  compared to the reference.
• Conclusions
• Sensitivity app. 25.
• Maximum impact on boundary layer observed 27h
  hours after perturbation.
• A better control of upper troposphere ozone
  (obtained by assimilating IASI data) makes it
  possible to improve the ozone forecast in the
  boundary layer.

7
Assimilation of simulated IASI data

• Simulation of IASI data
• Atmosphere description (Polyphemus from 0 to 5km,
  standard atmosphere above).
• Simulation of radiation radiative transfer model
  LBLRTM (AER).
• Simulation of raw measurements (radiances) IASI
  instrument model, provided by IPSL/SA.
• ESA operational inversion algorithm SA-NN,
  developed by IPSL/SA.
• Assimilation by Optimal Interpolation in a
  perturbated model

8
Examples of simulated IASI measurements
Error on 0-6km O3 column mean 27, instead of
expected 20.
Raw measurement IR radiation spectrum
9
Assimilation of simulated IASI data

• -red reference.
• -black perturbated model, mean error 13
• NO2 emissions 30
• O3 deposition -15
• O3 boundary conditions 15
• -green with assimilation, mean error 9

10
Conclusion

• -Yes, IASI can be used for improving air quality
  forecast
• Small but significative contribution on boundary
  layer O3 to the measurement.
• Good sensitivity of boundary layer O3 to a
  control of upper troposphere O3.
• Encouraging assimilation experiments, despites
  the simulation of data and simple assimilation
  method.
• -Better results expected with NOx measurements
  OMI, GOME2.
• -Hot topic in the community.
• -At Clime team ESA projects (EPS MetOp, TRAQ
  proposal), collaboration with IPSL/SA.