cluster capture-recapture - trouble shooting

[Annie]

Hello, 

I am running cluster capture-recapture models on double observer aerial surveys with uncertain ID (https://onlinelibrary.wiley.com/doi/10.1111/biom.12983). Surveys are on ice-associated seals. I have 5 strata. 

The model requires manual setting of parameter values (time lag, half-width of the detection area, half-width of the survey area, truncation distance between possible 'matches' i..e, duplicate observations), among others.

The challenge seems to be that very minor tweaks in parameter values results in wildly different density estimates, which is ultimately the metric I am after. I've run through a range of parameter values to see where D may stabilize (below), but very little comes out. 

My question is if anyone here has run similar types of models, or with the palm intensity function - and has suggestions on stabilizing models. I haven't included a lot of detail in this post, but happy to provide more if anyone has tips.