Nice to see that you are using ARTS. Yes, rigorous=0 should be safe for your case. This check is mainly relevant for limb sounding.

Depending on how much you care about the surface, DISORT is an alternative that could be faster.

Looks like you are doing stuff very similar to what we are doing. We should get an article accepted very soon involving GMI, the onion peeling algorithm, LPA and Evans! We will send it when it's out.



On 2022-03-22 12:04, daniele.case...@artov.isac.cnr.it wrote:
Dear all,

I am trying to simulate GMI with ARTS, some standard ice habits, using DOIT. However, I get an error from DoitGetIncoming: ERROR: Radiance difference between interpolation points is too large (factor 100) to safely interpolate. This might be due to za_grid being too coarse or the radiance field being a step-like function.
Happens at boundary 104 between zenith angles 92.3077 and 96.9231deg
for frequency #2, where radiances are 3.90337e-19 and 4.14798e-17 W/(sr m2 Hz).

I tryed to increase the N_za_grid in DOAngularGridsSet however the error remains, unless I set N_za_grid to a very large number (about 100) that crashes arts. These happens with all the ice habits that I tested (EvansSnowAggregates and LargePlateAggregate). Moreover, I used doit_scat_fieldCalc in the doit_scat_field_agenda (however using doit_scat_fieldCalcLimb does not change much), and I am using ice water contents derived from OnionPeeling code (with CPR profiles). My intention is to simulate a down looking view geometry, so I do not expect that such a fine za grid is needed.

I set the sensor geometry to:
   sensor_pos=[ 1, 1, 817e3]
   sensor_los=[1, 1, 135.0]
If I run DoitGetIncoming with rigorous=0 the simulation does not give an error, however is it safe? Other option could be to write a za_grid_opt_file like the one in testdata/testdoit_za_grid_opt.xml.
Any hint on how to proceed, would be very much appreciated.

Thank you very much,

Daniele Casella

arts_users.mi mailing list

Reply via email to