Where the data is dense, and where it isn't. Prices cluster around at-the-money strikes in near-dated expiries and thin out fast moving deep in- or out-of-the-money. MID_PRICE covers far more of the grid than TRDPRC_1: 23.9% of contract-days have a quoted mid with no trade behind it. The trade cloud is a sparse subset sitting inside a much denser quote cloud.
Why interpolating the gaps is dangerous here. On a $0.50 grid for a name this thin, an empty cell usually means nobody quoted or traded that contract — not that a price existed and went unrecorded. Interpolating invents a market that wasn't there, and on a ~$14 underlying each $0.50 step is a real move in moneyness, so a smooth surface can imply a fill you never could have gotten.
Mark vs. evidence of a trade. MID_PRICE is the mark — it's the closing NBBO midpoint, it exists on far more series, and LSEG exposes no true settlement price for expired US equity options. TRDPRC_1 is evidence someone actually transacted: one print, not a valuation. Median gap between them is $0.035 where both exist, so the mid is a reasonable stand-in — but I'll price off the mid and use trades to check it, not the other way around.