# Missing values in large spatial data

I have a groundwater dataset (4000 locations). For this data, the depth is very important to coordinate the data. I found that there are about 415 missing values in the depth column. Some of these missing values are also matching the missing values of other variables. So, I decided to remove these missing values from my data as I have already had a large number of spatial locations. In addition, the depth cannot be predicted from its neighbourhood (there are various depths for near locations). Furthermore, the missing values are almost for the same location. That is, one location has multiple measurements. For example, one location has 10 measurements for the depth and other variables. Then, 8 of these (depth) measurements are missing. The corresponding variables (may also missing) or close to the non-missing values (which may mean they will have the same depth. However, some of them are very different from the non-missing values which mean the depth is not the same). My model is based on predicting the unobserved location based on its neighbourhood. So, I think it is ok to remove these variables. That is because the most missing values are for locations (that are measured multiple time and may for the same depth). that are already measured and have non-missing values. However, I am not sure if this will be ok or not. Any idea, please?

• When you say 'remove these values/variables' do you mean that you ignore the entire data point whenever the depth is missing? – user20160 Aug 17 '19 at 19:28
• @user20160 Yes. I mean I remove the values from the dataset (delete them). For example, if my data is 400 with 50 missing values, then, my data becomes 350. – Mary Aug 18 '19 at 8:59