The RAYTRACE command in the ZPL will always trace the Hx/Hy normalized field coordinates.  The actual Field Type is controlled in the Field Data Editor and if you want Hx/Hy to reflect image heights, then you can change the Field Type to either Paraxial Image Height or Real Image Height.  Manually changing this before you run your ZPL code will now change the image height as you cycle through your normalized field coordinates.  

@MichaelH 

Thank you for your sincere answer. In fact, before running the RAYTRACE macro below, I had already modified the Fields Type to Real Image Height in the optic studio.
If so, can I regard the result of the figure 1 below as meaning that the RAYTRACE function moved to image height and generated Ray?

@MichaelH

Thank you for your continuous reply.
If I reverse the lens system in the optic studio and run the RAYTRACE macro below, I could get the figure 2 above. The following macro is not written by me, but shows part of the macro written for the experiment at Stanford University.
When the lens is reversed, I do not understand why figure 2 is possible when executing the macro below. 
Could you explain it to me? Below, the ImgHeight variable means Real Image Height. 

ImgHeight = 5
NumOfHeightStep = 50   
NumOfRadiusStep = 30
NumOfAnglesStep = 18

ImgHgtDelta = 1/NumOfHeightStep
RadiusDelta = 1/NumOfRadiusStep
AnglesDelta = 2*PI/NumOfAnglesStep

FOR hIndex, 0, NumOfHeightStep, 1
    hx = 0
    hy = hIndex * ImgHgtDelta
    FOR rIndex, 0, NumOfRadiusStep, 1
           FOR aIndex, 0, rIndex*NumOfAnglesStep-1, 1
                px = rIndex * RadiusDelta * COSI( aIndex * AnglesDelta / rIndex)
                py = rIndex * RadiusDelta * SINE( aIndex * AnglesDelta / rIndex)
                RAYTRACE hx, hy, px, py, 1
                PRINT  RAGX(0),  hy*ImgHeight, RAGZ(0), RAGL(0),  RAGM(0),  RAGN(0),  RAGX(ns),  RAGY(ns),  RAGZ(ns),  RAGL(ns),  RAGM(ns),  RAGN(ns)