An estimator of the underlying size distribution of overlapping impact-craters

The stochastic model of lunar type impact-crater formation which assumes (a) random impacts, (b) circular craters, each obliterating any portions of earlier craters lying within, and (c) a probability P_i(t) that a newly formed crater (primary or secondary) has an area a_i is analyzed to develop a method of estimating P_i from the final overlapping pattern. It is found that if each crater is weighted by the fraction of the rim which is visible and which lies in an observation area A, then the expected value of the weighted sum $\text{Ω}{}_{\mathrm{i}}$ of craters of area a_i is simply proportional to P_i for any degree of coverage under several conditions, including (a) constant P_i for all i, and (b) P_i stepping from a constant early value to zero (for some i's) with otherwise arbitrary bombardment. Furthermore, in the general case, the expected value of the contribution $\text{ΔΩ}{}_{\mathrm{i}}\text{(t}{}_0\text{)}\text{to}\text{Ω}{}_{\mathrm{i}}$ produced during t₀ ± Δt/2 is found to be proportional to P_i(t₀). Thus measurement of $\text{Ω}{}_{\mathrm{i}}$ in the first two cases, or of $\text{ΔΩ}{}_{\mathrm{i}}$ if crater age data is available in the last case, provides an estimate of the desired P_i. Therefore the $\text{Ω}{}_{\mathrm{i}}$ introduce the correct weighting factors that just compensate for the effect of overlap.Expressions for the variances of $\text{Ω}{}_{\mathrm{i}}\text{and}\text{Ω = Σ}{}_{\mathrm{i}}\text{Ω}{}_{\mathrm{i}}$ are derived from which it is shown that under the above conditions, $\text{Ω}{}_{\mathrm{i}}\text{/Ω}\text{or}\text{ΔΩ}{}_{\mathrm{i}}\text{/ΔΩ}$ are consistent estimators of P_i. Formal evaluation of the variances is carried out in the special case of constant P_i and no secondary cratering. A criterion for the degree of coverage is given; in particular it is shown that the expectation of $\text{σ = Σ}{}_{\mathrm{i}}\text{a}{}_{\mathrm{i}}\text{Ω}{}_{\mathrm{i}}$ at saturation is just A.