Details of estimating these approximate PPDs can be found in Padhi and Mallick (2014) and Li and Mallick (2015). The dashed curves, shown in Figure 3 denote the search windows used for estimation each model parameter. Notice from Figure 3 that the peaks of these estimated PPDs were connected to obtain an estimate for each model parameter. Also notice although we used a wide window for estimating each model parameter, the estimated PPD functions are very wellconstrained. The sliding window inversion allowed good constraints on the model parameter estimates from shallow to deep, which, in turn, is clearly reflected in these PPD estimates. Figure 2: An example of the 3-component VSP data used in this inversion study.