These features are known as crossover peaks. They arise whenever there are enough atoms whose Doppler shifts are exactly half the frequency difference between two transitions. Because the pump and probe beams are counterpropagating , the atom will see them Doppler shifted by equal and opposite amounts. If the laser frequency in the lab rest frame is exactly halfway between the two transitions, the atom will experience both pump and probe as being on resonance. The rest is just like the 'normal' peaks.

Note that some atoms like potassium and lithium can also have crossover features which are negative -- that is, they show increased rather than reduced absorption. The principle is the same, but the details rely on optical pumping between different levels. This doesn't happen in rubidium or caesium because the splittings between ground state hyperfine levels are much bigger than the Doppler width.

These features are known as crossover peaks. They arise whenever there are enough atoms whose Doppler shifts are exactly half the frequency difference between two transitions. Because the pump and probe beams are counterpropagating , the atom will see them Doppler shifted by equal and opposite amounts. If the laser frequency in the lab rest frame is exactly halfway between the two transitions, the atom will experience both pump and probe as being on resonance. The rest is just like the 'normal' peaks.

Note that some atoms like potassium and lithium can also have crossover features which are negative -- that is, they show increased rather than reduced absorption. The principle is the same, but the details rely on optical pumping between different levels. This doesn't happen in rubidium or caesium because the splittings between ground state hyperfine levels are much bigger than the Doppler width.

The peaks which I got for Rubidium and the simulation for Saturated Absorption Spectroscopy were found to be different.I expect it to be due to earth's magnetic field.How far do you think it is right?

I would need to know more about how they differ. As an estimate, note that the most sensitive states of rubidium atoms have frequency shifts due to small magnetic fields of 1.4 MHz/Gauss. Given that the Earth's field is well below 1 G, this means the shifts are in the 100s of kHz. I doubt you are this sensitive.

There are a lot of subtle effects. For example, if the beams are not perfectly counter-propagating, then there will be residual Doppler shifts which cause broadening. There is also broadening due to the fact that atoms only spend a finit time in the beam (as they fly through it), or due to collisions. There are phase shifts associated with birefringence of the cell windows which can cause etaloning effects which distort the signal. If the vapour pressure is too high then the signal is no longer linear; similarly if the laser intensity is too high Beer's law will not apply (absorption not linear), and optical pumping and population transfer between different states will occur.

In short, it is very hard to get quantitative agreement between saturated absorption theory and measurements. The main uses of saturation absorption are to identify the central locations of resonances, or to stabilise lasers in their vicinity.

The setup which I made was for actually locking lasers used in laser cooling experiments.I made a small portable setup and the spectrum found was good(as seen in the oscilloscope). But of course I did not lock any laser with it.

will using a magnetic filed isolator help?

Usually saturated absorption is not very sensitive to magnetic fields, as long as you mean weak fields (like the Earth's field). There are sometimes slight polarisation effects, but these shouldn't be a problem for locking a laser.

I appreciate this good discussion be Jon Goldwin and Ashwin Kumar. Thank you, it is very useful to me even I am working in different regime using LIBS spectroscopy.