I have reviewed your questions as well as your designs and have a few comments and suggestions. Before I get to that based on your questions I would strongly recommend you watch the introductory webinars hosted on the Optiwave website (<a href=”https://optiwave.com/category/resources/webinars/”>https://optiwave.com/category/resources/webinars/</a&gt;), specifically look for the webinar week post which will have a collection of webinars. The OptiFDTD basics contains an example of a reflectance calculation for a dielectric half space.

1. The observation area performs fourier expansions at each of the mesh points for specific wavelengths. What you are seeing in the observation area is the coefficients of the expansions. You can control what wavelength you are examining along the toolbar. So the results you see are the cumulative effect of the entire simulation.
2. The simulation summary is incorrect, if you setup observation points (I strongly encourage you to put one observation point at the center of each observation area) you will see that the sources are in fact pulses.
3. Your sketch of your desired time pulse and its spectrum is off. The Fourier transform of rect is in fact a sinc function not a delta. The fourier transform of a infinite sine or cosine is the delta. Are you looking for a pulse (something with a finite spectral width) or are you looking for something with delta function spectral form? If you are looking for a spectral delta then the source should be a CW (continuous wave) simulation. If you want a pulse but for it to almost be a delta you should simply increase the FWHM on your pulse, as this gets larger the width in frequency will decrease.
4. The y axis of power plots is either W or a normalized quantity. The normalized value is the power normalized by the input which was set to be the key input (Input Plane 1 in your case). Because you have multiple sources you either need to run three simulations and superimpose them or export unnormalized data and normalize it yourself.
5. Your spatial resolution is set to the auto which is the largest possible and still be numerically stable. With the dimensions of your structure this is not enough. You can see this by examining the quality of your spheres in the XY and XZ planes of the 3D Refr_Idx_RE(y) viewer within OptiFDTD designer.
6. You simulation was not run long enough, see my comment above about placing observation points at each observation area. This lets you see the time evolving field at one grid point. If you do this you will see that the pulse only just barely reaches your transmission observation area. So the pulse has not completely propagated through the structure so your reflectance measurements are incomplete.
7. You will need to clarify your comment that “the reflectance changes with time”. What time? How long the simulation runs or are you saying each simulation has a different reflectance?

Before proceeding much further I strongly encourage you to explore the introductory webinars I mention above as they establish much of the basics of setting up an FDTD simulation.

Scott