Chairman: Dr. T. Aoki

• Dr. T. Aoki introduced the session, explaining that as there are only 2 PIs involved, the meeting was likely finish before given schedule. He explained that Dr. Hori was to give a presentation, including comments on the launch status of ADEOS-II. Dr. Aoki then introduced each member of the cryosphere group attending the session.
• Dr. Hori talked about the agenda of the cryosphere group meeting, starting with the impact of the postponement of ADEOS-II launch. Unofficially he has heard that there will be a one-year postponement, leaving more time to implement and develop algorithms and carry out cal/val. His presentation then went on to cover the following
  
  • GLI standard products flow
    (Question by Dr. Aoki about SNGI data and timing of data. Dr. Hori responded in detail about SNGI data but was not sure about timing. Dr. James Simpson pointed out that it may take up to a year after launch to obtain products. Dr. Hori said that data timing may be around 10 days to 2 weeks)
  • input/output files (CTSK1)
  • input/output files (CTSK2b1_s)
  • input/output files (CTSK2b1_g) (global)
  • two types of map production for dataset (56 files and 4 files/zones)
  • examples of test input/output data
  • Concluded with implementation status of GLI standard algorithms for cryosphere
• Dr. Aoki mentioned other research products (snow/sea ice covered areas, snow grain size and impurity algorithm etc) and expressed the need for more accurate snow BRDF to deal with non-spherical snow grains. Dr. Aoki presented various aspects of the issue.
• Another problem is surface is the need to model surface roughness. Results of a simple geometrical model were not so good. Also good bi-directional distribution observation must be made. He referred to development of polar nepherometer for snow particles and measurement for each snow type, and to different BRDF measurement methods. Since there is time to the ADEOS launch, there is time to prepare such methods. Monte Carlo method is most effective and maybe only method available which is suitable at this stage.
• Suggestion received to model snow surface whilst considering defraction for near field snow particles.
  (Comment from James Simpson that a model is needed which combines atmospheric properties and surface data to properly address the problem.)

Top of this page

• The discussion started on the issue of aerosols. GLI now not able to assess ground aerosols (assumes aerosol type from global data) and it was agreed that aerosol issue is biggest uncertainty to be addressed - methods of aerosol correction were also discussed.
• Another issue was that of wavelength and snow penetration - if both grain size and impurity retrieval are being carried out, need to be penetrating same layer. Two wavelengths have to be probing to the same depth (therefore chose shorter wavelengths) On the issue of near field, it was commented that far field approximation may be more accurate for near field than expected.
• It was commented that there could be scale effects in moving from meter scale to kilometer scale GLI pixel.
• There was a lot of discussion of snow shapes and smooth phase function, and need was stressed to average size, orientation and shape - the latter being the cause of smooth phase function. There was a comparison of methods used on clouds and discussion of whether this could be applicable to snow. Dr. Stamnes suggested a spheroid for averaging shape.
• Dr. Simpson pointed out that the level of input should be appropriate for retrieval required as detailed input could be smeared out on retrieval. Sensitivity tests with many constants are required.
• The point was reiterated that aerosol impact must be grasped as impurities which are in aerosols may be read as being in snow.
• Dr. Aoki then began to summarise each item:
  • aerosol correction is high priority.
  • Vertical profile of snow grain size will try to use more channels. Make a dataset of some retrieved results for snow grain size using different channels.
• Further discussion ensued, including the use of 250m channel and direct downloading of data.
• Dr. Simpson asked about snow water equivalent, which is probably being carried out by the microwave group. He pointed out that may be problematical as the microwave signature can easily disappear. He planned to check with the land group.
• There was a discussion of using MODIS data - which is available from next year to JAXA via NASA. It was agreed that selecting scenes was more useful than global processing. It was also pointed out that validation work done on MODIS may save time and effort at a later stage with GLI data.
• The delay in launch may allow up to 3 years more time for preparation on instruments.
  Validation was widely discussed, the validation site in Japan is not big enough. Dr. Simpson pointed out advantages of Alaskan site - a lot of measurements are already made routinely and can be very helpful. Dr. Stamnes wondered if it would be possible to convince JAXA it is useful to carry out validation against MODIS before GLI? It was agreed to ask Dr. King what MODIS field experiments are. Dr. Aoki explained that current plans were to prepare for experiment in Alaska in 2001, or observation in Hokkaido with scientists from US
• There was more detailed discussion about methods for snow grain size and vertical profile, and an explanation by Dr. Aoki of other algorithms proposed but without concrete programs as yet. He explained that preliminary survey problem will be worked on with Dr. Nishio of Chiba University, who plans to make a dataset for snow covered areas.
• Dr. Simpson pointed out the benefits of blending infrared and NIR with microwave measurement in order to both see through cloud and not be affected with wet surface. He said that as JAXA already have a neural network already in place it could be used to get hybrid/blending approach started.
• Dr. Aoki explained that the number of members is limited to carry out new experiments and he is hoping for new members from Chiba University. He welcomes collaboration with other groups.
• On the question of ancillary data, it was agreed that aerosol data is most important and that atmospheric pressure would be needed for aerosol correction. The present algorithm only uses background aerosols, using 5 different models and compare reflectance differences to get closest model to observation, not very sophisticated - more is needed, this is just a start.
• Dr. Simpson pointed out that water vapour concentration doesn't affect things very much, so is not a high priority.
• Finally, it was agreed that there was 2 or 3 years available for upgrading systems and for fine tuning. Dr. Hori said that upgrading would only take less than half a year.
• Drs Stamnes and Simpson pointed out that spending some time on initial preparation (e.g. database for ancillary data) pays dividends at later stages, so allow time for upgrading.
• Dr. Aoki closed the evening's meeting, and asked that any comments be passed to him for inclusion in report tomorrow. He asked that any corrections to his report be made