Handling missing values when the value is not applicable (spatial regression)


I'm preparing data to run regression analysis in a geo-spatial context.

I am adding spatial analysis data to housing characteristics, i.e. distance to railway station. I have limited this to a search radius of 1km around each house. The issue is with some variables like close to water bodies, some houses are further away than 1km, so the numeric value is 0. If I replace all the blanks with 0, will the regression treat that 0 as a water body being really close to the house and therefore it will skew the result. Some houses do have water bodies within 1km and so have a value and this will have an impact in the regression and the water body will have some impact on the house price, but when the result isnt just 0, its just not applicable, is where I am getting confused?

Any help much appreciated.