You can ignore that graph entirely. Unless explicitly stated, 'adjustable' in the context of voltage regulators (regardless of the type) does NOT mean adjustable while in operation. It means you can adjust the fixed output voltage to a preset value, and no more. In other words, it is still a fixed regulator, the fixed regulation point happens to also be adjustable (allowing regulation to any single fixed voltage).
That chart is intended for picking a single, fixed output voltage along with a maximum current and maximum input voltage. This is not the same was your intended application.
But, in case you still want to know how to decipher the chart, the equation to convert diode drop, switch saturation voltage, input voltage, and output voltage into a corresponding magnetic flux (aka Webers/volt-seconds) value, it is on page 27 of the datasheet, listed under equation 4.
As you can see however, it is quite apparent that this is specific to a fixed output voltage.
I am not sure why there seems to be no mention of this in TI's datasheet considering they are the original creators, but the LM2596 is heavily second sourced, and On semi's version of the datasheet, page 21, figure 30 (the part is, in theory, identical in performance and behavior across different vendors) for the LM2596 just shows you exactly the schematic for what you want.
Sure, it's to a higher voltage, but who cares. If it works for 0-30V, then it will work for 0-12V since 0-12V is contained within 0-30V.
One caveat: you cannot regulate down below the reference voltage, so you're minimum voltage will be about 1.5V, not 0V. Properly regulating down to 0V will require, most likely, creating your own voltage regulator using an op amp and pass transistor, as well as a negative voltage rail.... but that is all well outside the scope of this answer and question. My advice is: if you actually need 0V, then just buy a bench power supply. If you can live with 1.5V, cool.
In the case of a power supply you want to adjust during operation over a wide range, it is (and treasure this because it happens so rarely)... way easier. You just use a 33µH inductor rated for the current (make sure its magnetic core saturation current is high enough, not just the max winding current) you want. 47µH also works, and 100µH. You get slightly different efficiency curves for various input voltages, with larger inductors improving efficiencies at light loads, but at the cost of heavy load efficiency.
So its really not even that big a deal. Anything between 33µH and 100µH will definitely work fine, just with slightly different performance across the other variables.
I say 47µH because almost all of the now-ubiquitous adjustable DC/DC converters (just search amazon for LM2596 if you live in the states - the thing you want is almost commodity at this point) have little 47µH shielded inductors on them, and if its good enough for them, its probably good enough for you.
Good luck and mind the circuit layout!