How to mosaic several spatially referenced raster tiles or arrays

I have several geolocated small tiles (obtained from satellite data) which I want to combine to conduct a global-scale analysis/plotting. What would be the best way to combine them into a single mosaic? What I want to do is similar to the example posted below, but I have several tiles, not only two, and they are not adjacent. Please advise a efficient way to combine the tiles into a single dataset. I have the original tiles in the form of arrays as well so I can use both cells and arrays. Each tile/array has associated lat/long arrays.

https://www.mathworks.com/help/map/mosaic-spatially-referenced-rasters.html

Thank you!

[KSSV]

Do you have (x,y) i.e. locations of the data?

[Sagar Parajuli]

Yes, I have lat/long (degrees) arrays associated with the data

Answer 1

[KSSV]

Convert all your tile data into three column data i.e. (x,y,z) and then use griddata. Read about gridata. If you are not able to get it share the data.

[Sagar Parajuli]

Thank you, I have attached sample data. The data are in aod_final.mat, and associated lat/long are in lat_final.mat/long_final.mat. The third dimension represents the number of tiles in a single day (160). I want to combine all the tiles in a single day to create a global mosaic. The fourth dimension is the time (2 days currently). The data have lots of NaNs in the first few tiles so please ignore the content. I want to mosaic the data for each day and finally put them in a global array of size 180*360*2 (lat*long*time).

[KSSV]

You have to proceed like below:

load lon_final.mat ;
load lat_final.mat ;
load aod_final.mat ; 
% select first day
lon = lon_final(:,:,:,1) ; 
lat = lat_final(:,:,:,1) ;
aod = aod_final(:,:,:,1) ; 

lon = lon(:) ; lat = lat(:) ; aod = aod(:) ; 
idx = ~isnan(lon) ; 
lon = lon(idx) ; lat = lat(idx) ; aod = aod(idx) ; 

m = 500 ; n = 500 ; 
x = linspace(min(lon),max(lon),m) ;
y = linspace(min(lat),max(lat),n) ; 
[X,Y] = meshgrid(x,y) ; 

Z = griddata(lon,lat,aod,X,Y) ; 

Your data aod_final for the first day has hardly any data. Out of all the numbers only 8 are non-nans. Rest everything is nan. So Z has no values. You need to check your aod values.

[Sagar Parajuli]

Thanks, this is promising. I will have to use a loop to process all the data for the entire year (365 days) but I believe I can do that. This is very helpful.

[KSSV]

Yes...you run a loop.... If you feel the indices always remain same....you can get them once and run a loop.

I think this is a satelite pass data.