unprojected MODIS with LAT/LON rasters

[TimSzeliga]

We're doing a mixed-resolution comparison between MODIS and TM data.
I'm required to do the analysis using the original unprojected
MODIS 1km satellite raster.  The MODIS comes with a pair
of lat/lon rasters, in floating point, same size as the image
bands, indicating the latitude and longitude of each image pixel.
I'm working from raw MOD03, MOD021KM, MOD02QKM, etc from
LAADS through rapidfire.sci.gsfc.nasa.gov and with TM from
glovis.usgs.gov.

Is there a mechanism in ArcGis 9.3.1 to use this information
to navigate the image, without first reprojecting to geographic
or UTM?   There is a warp function, and I will eventually need to
pick matching landmarks.  The goal is to warp projected TM data
into the "satellite-projection" MODIS space. The TM is in UTM,
in meters, while the MODIS has lat/lon given (although is not in
an actual "geographic" projection).  The lat/lon does not progress
regularly across the image.  If I stick to a TM scene falling in the
center of the MODIS swath, the distortion is not too bad.

Do I need to transform the target lat/lons to some kind of relative
meter-based extents?  Can warp work from meters->degrees?
The snow classifications don't have any detail, so I'd have
to use the MODIS and TM images to pick landmarks and build
a file of correspondences, then apply this to the two snowmaps
to do the final warp.  I have ENVI and the MODIS-SWATH_MRT
available. 

We're investigating a MODIS snow cover classification scheme,
and using the TM-derived snow as ground truth to examine fractional
snow cover.  The scene looks like a postage stamp affixed to
a poster, but blown up to 30m TM resolution, the MODIS drapes
over like a 1km checkerboard.  Counting the ratio of Snow/No_Snow TM pixels
falling in each MODIS box gives a good assessment of fractional snow cover.

The big problem will be warping the TM scene (in UTM-17N) to the
MODIS (in satellite projection).  I plan to build a fishnet at 1km the
size of the smaller TM coverage, convert to polygon and use Zonal
Statistics to make the count.   Do any of the standard raster/math/feature
operations refuse to work in an unprojected background?  Is there a
list anywhere?

Yes, I know this would be a trivial matter if I could project the MODIS
to match the TM in UTM, but I am required to do it in this ass-backwards
manner.

Tim Szeliga
timothy.szeliga@noaa.gov

[TimSzeliga]

Writing it out helped clarify the problem.  Since the MODIS comes with
LAT and LON rasters, I contoured them at whole degrees to clearly
identify the crossing points at (41, 90), (42, 90),..., (45, -99), (46, -99),
then noted down the associated raster (line, sample), still in the unprojected space.

Then, in another window, I took the UTM-projected TM data and displayed
the whole degree lat/lon grid.  This gave me a set of corresponding (x,y) points.
From there I could throw away the actual latitude and longitude values
and build a table of MODIS_x, MODIS_y, TM_x, TM_y.
Should the TM be expressed as pixel-locations or UTM-meters?
Should all projection info for the TM be scrapped, treating it as a
raw raster?

A five-minute MODIS section covers a much greater area (at lower spatial
resolution), so I needed work from a smaller clip that fully contains the TM,
to minimize distortion, adjusting the numbers accordingly. 

Mapping the TM straight to the MODIS projection is a problem, as the TM is 30m
and the MODIS is around 1000m.  It won't do to merely warp TM to MODIS,
since the result would select one TM pixel out of a thousand to represent the value.

I need to fudge a raster 33.33 times the size of the original, increasing the
resolution, but keeping the same extents.  The new MODIS would look like a
checkerboard, with one value covering a large square and the TM filling the block
at normal res.  The goal is to count the pixels within each block, comparing the
MODIS fractional snow-covered area with the calculated FCSA from the TM.
Set snow as 1, bare-ground as 0 and cloud as NO_DATA, then a simple mean
calculates the FSCA.

I'll try fishnet to build a grid surrounding each big MODIS pixel, then
Zonal stats to find the mean.  The first attempt made a polyline grid
and only computed stats along the lines.  Need to convert the grid to polygon squares,
but it just runs for hours.  A smaller section completed in 34 hours.  I may have to
chop this into much smaller pieces and merge.