Okay, to answer your first question, there are generally two things that might cause a high price: a generator offers in a high-priced tranche of megawatts that gets cleared and/or there is a transmission constraint.
In the specific example you cite, if we zero in on 21 Feb 2019, we can see that the average price across all nodes in the Central North Island (CNI) is $7,021.49/MWh in trading period 39. If you change the region type filter to Nodes (Central North Island) and then apply some thoughtful trial and error to the series filter, you can quickly figure out that the high average price is caused by a price of $500,000/MWh at ROT1101.
A price of $500k is an infeasibility in the pricing model so that tells me the price you are looking at is not a final price; it is a provisional price. But prices should've been finalised for 21 Feb by now, and indeed when I check our database I find that they have been finalised. What you've 'discovered' is that we've slipped up and failed to load the final prices into the database that drives this EMI report. We will rectify that now or first thing on Monday morning. We've experienced several issues of late with high loads on our data warehouse as we migrate from some legacy processes to new and improved processes. This causes jobs to fail from time to time and in this case we seem not to have picked that up.
The final price at ROT1101 on 21 Feb in trading period 39 is actually $267.24, and that is what EMI will tell you very soon.
Finally, I don't know what you mean when you ask "why one report showed the spike at $2,204.50 /MWh, while another report shows it as $6,917 /MWh".
Hope this helps,