Ongoing projects
All the tests of the sensitivity study have been conducted using a computer system with the following specifications:
Processor: 2.4 GHz Opteron processor AMD64;
RAM : 4GB (4x1 Gb) DDR-400 ECC
Hard Disk: 2 x 250GB/150, SATA, 7,200 rpm
KLIMA-IASI Accelerated Forward Model
Configuration of the AFM
In this section, the conclusions of the sensitivity tests conducted for the definition of suitable trade-off between efficiency and accuracy of the KLIMA-IASI FM are drawn and the impact of the different approximations and the results of simulations performed using a fully optimised FM are presented. These will show the overall impact on the accuracy of the synthetic spectrum due to the combined approximations that we individually discussed in the previous section.
The accelerated configuration of KLIMA-IASI FM (see Table 3) has been set up by implementing the following options:
- Fine frequency grid with steps of 0.00390625 cm-1 (equal to 1/64 the spectral sampling of the IASI instrument).
- Contribution of CO2 line mixing modelled only in the spectral bands [650 - 800] cm-1 and [2000 - 2500] cm-1.
- Spectroscopic data are taken from the HITRAN 2004 catalogue for all species other than CO2 and for each species all spectral features with intensity smaller than 10-5 the most intense line in the selected frequency band are skipped. For CO2 the dedicated spectroscopic database used in the line-mixing model is adopted.
- The atmospheric line shape is modelled up to ±25 cm-1 from the line centre for the target species (H2O, CO2 and O3) and ±10 cm-1 from the line centre for all other constituents.
- An optimised vertical grid is used consisting of 44 pressure levels from 1050.0 hPa up to 0.005 hPa (21 levels between 1050 and 478.4 hPa and 23 levels between 478.4 hPa and 0.005 hPa).
- The FOV has not been modelled.
|
|
Reference FM |
Accelerated FM |
|
Fine Frequency Grid |
0.001953125 cm-1 (1/128 of IASI spectral sampling) |
0.00390625 cm-1 (1/64 of IASI spectral sampling) |
|
Line-Mixing |
modelled in the band []645 - 2760] cm-1 |
modelled in the bands [650 -800] and [2000 -2500] cm-1 |
|
Spectroscopic Database |
HITRAN 2004 (All lines) |
HITRAN 2004 (skipping all lines smaller than 10-5 the largest line in the selected band) |
|
Atmospheric Line Shape |
All lines modelled up to ±25 cm-1 from line centre |
lines modelled up to: ±25 cm-1 from centre (target species) ±10 cm-1 from centre (other species) |
|
Vertical Grid |
90 levels from 1050 to 0.005 hPa |
44 levels from 1050 to 0.005 hPa |
|
Field of View |
Not modelled |
Not modelled |
Approximations adopted for the KLIMA-IASI AFM with respect to the reference configuration
Figure 13 shows the difference between the spectrum simulated with the reference FM and the accelerated FM.
Figure 13
Difference between the spectrum simulated with the reference FM and the AFM.
The red line is the nominal instrumental noise;
the green line in the nominal instrumental noise reduced by a factor 10
Performances of the KLIMA-IASI Accelerated Forward Model
The level of performances achieved by the AFM must be compared with the requirements specified for the size of the code and for the computing time necessary to perform a single run of the FM without Jacobian calculation.
Table 4 shows that despite the significant reduction obtained with respect to the reference FM both in terms of code size (~50%) and computing time (75%), the AFM is not yet fully compliant with the requirements.
A further reduction can be obtained by limiting the spectral band to be analysed. By restricting the spectral coverage of the FM calculations to the bands already selected for modelling the line-mixing effects ([650 -800] cm-1 and [2000 -2500] cm-1), a 66% reduction of the spectral band to be analyzed is achieved. As shown in Table 4, this is sufficient to satisfy the requirements for integration on G-POD.
|
|
G-POD integration requirements |
Reference FM |
Optimised FM |
Optimised FM with band selection |
|
Program Size |
1.0 Gbyte |
3.7 Gbyte |
1.9 Gbyte |
0.9 Gbyte |
|
Computing time |
180 min |
800 min |
200 min |
150 min |
Table 4
Comparison between the performances of the reference and optimised FMs and G-POD integration requirements.
The reported estimate of the computing time are based on the results
of tests carried out using a computer with the specifications given here.
Sensitivity test for the optimization
of the atmospheric vertical grid
The vertical grid adopted for the reference FM is the same that is used to represent IASI L2 operational products for temperature and water vapour, consisting of 90 levels in the pressure range from 1050.0 hPa to 0.005 hPa. In the accelerated configuration of the forward model, the vertical resolution of the pressure grid was degraded: the 21 lowermost levels from 1050.0 hPa to 478.4 hPa (2 km) were the same as in the reference vertical grid. For the upper levels, only one every third pressure level was maintained in the new grid.
The total number of pressure levels in this approximation is 44.
Figure 12 shows the difference between the spectrum simulated with the downgraded and the reference vertical grid.
The largest differences are observed around 670 cm-1 in the middle of the CO2 band, where the CO2 content in the upper tropospheric and lower stratospheric levels provide a significant contribution to the observed radiance and where a coarser vertical resolution of the atmospheric model at higher altitude may affect the accuracy of the simulation.
Figure 12
Difference between the spectrum simulated with and without the vertical grid reduction.
The red line is the nominal instrumental noise;
the green line in the nominal instrumental noise reduced by a factor 10
KLIMA-IASI Reference Forward Model
In order to test the sensitivity of the KLIMA-IASI algorithms to individual approximations applied to the FM module, a reference configuration of the forward model was identified.
The approximate values of spectral radiance simulated using different optimisation options were compared with the synthetic spectrum generated by the reference FM configuration and the resulting differences were evaluated with respect to the nominal NESR of the IASI instrument.
The list of forward model approximations tested in the sensitivity study included the following aspects:
- Fine frequency grid used to simulate the high-resolution spectrum
- Modelling of the Line-Mixing Effects
- Spectroscopic Database
- Modelling of Atmospheric Line Shape
- Vertical grid
- Instrument Field of View
With respect to these aspects, the reference configuration set-up for the KLIMA-IASI FM can be described as follows:
- Fine frequency grid - a regular grid with steps of 0.001953125 cm-1 - equal to 1/128 the spectral sampling of the IASI instrument (0.25 cm-1) is adopted. A multiple of the frequency sampling of the ILS is used in order to avoid interpolation, when the convolution between the ILS and the simulated high-resolution spectrum is computed. Tests have shown that the use of a finer frequency grid results in systematic effects smaller than the nominal instrumental noise and affecting the uppermost altitude levels, where the information provided by the IASI measurements is negligible. Figure 6 shows the difference between the simulated spectrum in the spectral interval [650 - 800] cm-1 calculated using a fine frequency grid with steps of 0.0009765635 cm-1 (equal to 1/256 of IASI spectral sampling) and the reference spectrum. In the figure is also reported the nominal instrumental noise as reference for the approximation. The simulation has been performed in the CO2 from 650 cm-1 to 800 cm-1 where the higher levels contribute and more narrow lines are observed
- Line-mixing Effects - CO2 line-mixing effects are included and their contribution is estimated according to the model by Niro et al. (2005a and 2005b). are taken into account. Spectroscopic data for carbon dioxide are taken from the dedicated database used by the CO2 line-mixing model.
- Atmospheric Line Shape - All atmospheric lines are modelled up to ±25 cm-1 from the line centre.
- Vertical Grid - The vertical grid is the same of the IASI L2 products, i.e. consists of 90 pressure levels from 1050.0 hPa to 0.005 hPa.
- Instrument Field of View - The effects due to the field of view of the instrument are not taken into account (see the section on KLIMA-IASI upgrades for more details about the impact of this approximation on the accuracy of the FM).
Figure 7 shows the spectrum simulated using the reference FM.
Figure 6
Difference between the spectrum simulated with 0.0009765635 cm-1 (1/256 of IASI spectral sampling)
and the reference fine frequency grid. The red line is the nominal instrumental noise;
the green line is the nominal instrumental noise reduced by a factor of 10.
Figure 7
Synthetic spectral radiance in the wave number range [650-2670] cm-1,
simulated using the KLIMA-IASI Reference Forward Model
Sensitivity test for the optimization
of the atmospheric line shape
In the reference FM the atmospheric line shape was modelled up to ±25 cm-1 from the line centre.
Sensitivity tests were conducted to evaluate the impact on computing time and FM accuracy of modelling the atmospheric line shape for non-target species (i.e., all the simulated species except H2O, CO2 and O3) up to ±10 cm-1 from the line centre.
Figure 11 shows the difference between the spectrum simulated with and without this approximation in the model of atmospheric line shape. The simulation was performed on the full spectral range covered by IASI. In the figure is also reported the nominal instrumental noise as reference for the approximation. The larger differences are observed around 1250 cm-1, but they never exceed the nominal instrumental noise.
We conclude that the error introduced by this approximation does not affect the CO2 information.
Figure 11
Difference between the spectrum simulated with and without atmospheric line shape approximation.
The red line is the nominal instrumental noise;
the green line in the nominal instrumental noise reduced by a factor
