Downsample and reconstruct a subsection of a vector?

Hi all,

I have vectors of lonitude and latitude (and other equally sized metadata vectors) from some geospatial data over a region at a uniform resolution. Within this I have defined an area of interest using a bounding box (geo.FineSamplingBox) defining top left and bottom left coordinates of the area of interest. I have used this to index the points inside and outside this box. The points inside the box I would like to retain the original sampling rate, and the points outside the box I would like to downsample by a factor of 10 (geo.SS_Factor).

I have managed to extract the respective points, but I am unsure how to re-insert them into the original matrix without losing the original order. I have attempted to solve it by appending the fine matrix with the coarse matrix and then sorting them but I am not sure the best way of doing this.

Any thoughts, answers or comments would be greatly appreciated, thank you! Apologies I cannot provide any data - the fine sampled vectors are 1888x1, and the coarse sampled vectors are 1370x1.

Code extract below:

 %% ADDED - APPLY COARSE/FINE SUBSAMPLING BASED ON BOUNDING BOX
    if geo.BB ==1
        FineSampLon = loadedData.Lon > geo.FineSamplingBox(1) & loadedData.Lon < geo.FineSamplingBox(3);
        FineSampLat = loadedData.Lat < geo.FineSamplingBox(2) & loadedData.Lat > geo.FineSamplingBox(4);
        FineSamp = FineSampLon & FineSampLat;
        CoarseSamp = find(loadedData.Lon < geo.FineSamplingBox(1) | loadedData.Lon > geo.FineSamplingBox(3) | loadedData.Lat > geo.FineSamplingBox(2) | loadedData.Lat < geo.FineSamplingBox(4));
        if sum(CoarseSamp)>0
            PhaseCoarse= downsample(loadedData.Phase(CoarseSamp),geo.SS_Factor);
            LatCoarse = downsample(loadedData.Lat(CoarseSamp),geo.SS_Factor);
            LonCoarse = downsample(loadedData.Lon(CoarseSamp),geo.SS_Factor);
            HeadCoarse = downsample(loadedData.Heading(CoarseSamp),geo.SS_Factor);
            IncCoarse = downsample(loadedData.Inc(CoarseSamp),geo.SS_Factor);
            
            PhaseFine = loadedData.Phase(FineSamp);
            LatFine = loadedData.Lat(FineSamp);
            LonFine = loadedData.Lon(FineSamp);
            HeadFine = loadedData.Heading(FineSamp);
            IncFine = loadedData.Inc(FineSamp);
        end
    end     
    %Reorder lat and lon matrix and apply that to the other matricies?
    %Find when lat is above or below the thresholds of the bounding box and
    %add coarse sampled data back in??

Answer 1

[dpb]

If they're just vectors, I think the simplest is probably the best and likely the quickest...

LO=geo.FineSamplingBox(1); HI=geo.FineSamplingBox(3);      % convenient shorthand names
N=geo.SS_Factor;
dataPrime=[decimate(data(:,1:LO),N) data(:,LO+1:HI-1) decimate(data(:,HI:end),N)];                 % columns of 2D
dataPrime=[decimate(dataPrime(1:LO,:),N) dataPrime(LO+1:HI-1,:) decimate(dataPrime(HI:end,:),N)];  % rows of 2D

I didn't try to construct it, an alternative would be to create a 2D logical addressing array that matches the areas and uses it as a logical address expression.