The dynamic context server has an entry for monitoring ice-out conditions on Minnesota lakes. The ice-out time is a sensitive indicator of local climate and the date has been arriving earlier in a statistical average sense each year.
The Minnesota DNR hosts a server which provides JSON data on a number of Minnesota lakes organized by year. So to store a record for a particular year simply amounts to querying the server for that year.
minnesota_dnr_ice_out( 'http://www.dnr.state.mn.us/services/climatology/ice_out_by_year.html?year=' ). %% get_ice_out(-Year) % % Retrieve data from remote server for given year and save get_ice_out(Year) :- minnesota_dnr_ice_out(URL), atom_concat(URL, Year, U), http_client:http_get(U, R, ), atom_json_term(R, J, ), J=json([status='OK', results=L, message='']), maplist(store_record,L).
The web interface to the server looks like the following. The data download is executed once for a range of years, and then the graph is selected for a particular latitude.
Scoping the data to a latitude is useful in reducing the variance, as the ice-out dates obviously will arrive later the farther north one goes.
Since I last looked at this data and posted at the old blog, we had a spring with a rather late ice-out date. On average, this reduced the rate of change in ice-out date per year by about 23%. This points to the extreme variability of the data and the importance of long-term statistics needed to wean out trends.
I have never seen an ice-out incident in real-time, but the following is a short time-lapse of a pre-ice-out from earlier this year, taken April 28 on Maple Lake at 45.2N latitude. As you can see, some of the chunks of ice started to peel off from the perimeter, which is a precursor to the complete ice breakup, which likely occurred in the next day or so. This particular example was the 118th day of the year, and one can see that the average ice-out for that latitude was clustered around day 120.
We will keep on monitoring this data over the coming years. Like the Arctic ice coverage, enough variability exists that nothing beats keeping statistical records to monitor the actual trend.