Search

link to homepage

Institute for Energy and Climate Research

Navigation and service


JURASSIC (Inverse radiance modeling)

The analysis of limb and nadir sounders in the infrared region is an inverse problem. It requires an accurate radiation transport model and associated inversion procedures. JURASSIC is a software package developed by IEK-7 for analysing the measured data from instruments of this type.

The contribution of different trace gases to spectrally resolved infrared radiation as measured by limb sounders such as GLORIA.The contribution of different trace gases to spectrally resolved infrared radiation as measured by limb sounders such as GLORIA.

HNO3 concentrations derived with JURASSIC2 from CRISTA-NF measurements during the RECONCILE campaign.HNO3 concentrations derived with JURASSIC2 from CRISTA-NF measurements during the RECONCILE campaign.

The original version was created by Dr. Lars Hoffmann to interpret MIPAS and CRISTA-NF data. Currently, special AIRS retrievals are carried out to investigate gravity waves, and three ESA studies are performed to select the PREMIER mission for ESA Earth Explorer 7. At the heart of the model written in C is a radiation transport model based on the emissivity growth approximation and the Curtis-Godson approximation. These band models calculate the reference tables in advance and are therefore up to 1000 times faster than more precise models based on an analysis of all spectral lines. Thanks to a 3D ray tracing procedure, the model can be adapted to the view geometries of a variety of instruments. The inversion of the measured data is performed using the Bayesian optimal estimation approach.

The new version of the software, JURASSIC2, is being developed further by several PhD students and postdocs at the institute. JURASSIC2 is used in retrieval studies for the (tomographic) derivation of 3D trace gas distributions from PREMIER and GLORIA data. At the moment, CRISTA-NF measurements from the RECONCILE campaign are also being analysed with JURASSIC2.
JURASSIC2 is based on Python/C++ and additionally contains a generic instrument model, which is able to accurately simulate the field of vision and correlated instrument effects, even for different instruments in the same analysis. Linear algebra can use sparse matrices and associated algorithms, which means considerable advantages, in particular for 2D and 3D tomographical analysis. In addition to Bayesian estimation, Tikhonov regularization is also used with different norms. In cooperation with STCE der RWTH Aachen, IEK-7 is working on an analytical derivation using algorithmic differentiation, which can already calculate the derivative matrices when the model is simulated, causing little extra cost. This reduces the typical computing time required for an inversion by one to three orders of magnitude, depending on the problem size.





Servicemeu

Homepage