How can I create an image of satellite pixel using 4 latitude/longitude corner points instead of a 1 center lat/long coordinate

If the satellite data is just non-linearly gridded, you can supply those coordinates to an Igor image plot using auxillary X and Y waves.

For more info, enter the following on Igor's command line:

Hello,

Thank you for this reply, which made me realize I was not sufficiently clear in my description. The pixel are are not rectangles, but something more like not fully linear parallelograms that change size across the image (field of view). In short, I would need to be able to supply individual x and y scalings for each row and column of the image matrix. I am attaching an image to illustrate what I would like to plot. The example image I have available is from the center of the satellite's field of view, where the pixel almost all have the same size, but across the whole image each image has a different size that is only determined by its corner latitude/longitude points. Perhaps wanting to plot it as as image is not the right approach?

Thanks and Best, Barbara 

You should probably use the Drawing Tools to create this, then. It might not render very quickly, though, and the recreation macro would be huge. Still, it should work. If the drawing tools are new to you, start with this help topic:

The useful thing about drawing tools is that you can write code to create the drawings, and you can use the graph coordinate system to place lines, rectangles and polygons.

You can set the color of the poly before you draw it, too.

I imagine code that loops through your loaded data and executes SetDrawEnv fillRGB commands to set the color and then DrawPoly commands.

Like this:

Thank you very much, this is extremely helpful. I will check it out and if the rendering is too time consuming I will at least have it available for select cases.

Much appreciated, Cheers and Best,

Barbara

I think that there is another approach that may or may not have coding or rendering advantages. It is a little complicated, but not much more than programming the drawing tools. The suggestion is to use Gizmo, and construct a Group using quad objects for the irregular pixels. All quads would have the same fixed z-coordinate, each with its own four x,y vertex values. To get started

 To give a flavor of a coding example, here is how I altered the color of one quad, in a two-quad group, setting it to green:

You could use 2 Nx4 waves to hold all your vertex 'x' and 'y' values for each pixel ('z' is an arbitrary constant for the plane), and write a loop to construct each set of quad coordinates in the group. Coloring each quad would proceed as shown above, using an Nx4 wave of your color values (R,G,B,alpha) for each pixel (quad). Here is the recreation macro for my test Gizmo with two quads; converting to a loop structure over pixels should not be too difficult.

(Edit: Macro transcription error corrected)

Here is a Gizmo comparison to Jim's previous graph. In practice the user supplies the wNx and wNy vertex values and color scale values, which I have concocted. Prior to executing the function below I opened a Gizmo window with initial box axes, translate, and scale operations. to simplify the example, I also opted not to use a group of quads

The attached figure shows a vertical (home) view of the gap-less pixels' surface in the z=0 plane

After helpful discussions with AG, my current opinion is that a Gizmo Surface plot using sequential quads for the source wave is the easiest way to go.

All you have to do is put your data into the specified source wave format. Here is a code snippet for one kind of conversion:

I am attaching a picture of a Gizmo projected view using simulated vertices (the color wave shows the data values along the projection axes, taken from one coordinate of vertex data). There are only two Gizmo objects involved, axes and surface plot.