Shading on the VIP freeze maps

 The freeze maps in VIP are shaded using a different method than many other MRCC maps. Typical shading methods use some type of interpolation to infer values between points with known values. For many situations, interpolation is a valid way to create shading. In the case of freeze dates, the episodic nature of the events can lead to situations where stations have freeze dates that vary considerably from those of nearby stations, and the intervening dates may not be valid for the variable being mapped.

 Therefore, the shading on VIP products uses Thiessen polygons. Rather than interpolate data between stations, the shading is determined by mapping to the nearest station with a known value. In this scheme, it is possible to have shading jump from one value to another without the intervening colors appearing on the map. This is the desired behavior so that values that did not occur will not appear on the map.

 As an example, two neighboring stations (A and B) have a first fall freeze date that differs by several weeks. An early freeze occurs at A on 9/28 while B remains above freezing. A period of warmer weather keeps the area above freezing until 11/3 when the entire area experiences a freeze. In this example, the first freeze date for A is 9/28 and for B is 11/3. Shading methods that interpolate would give values throughout October to the area between the stations despite the temperatures remaining above freezing throughout the month. Using Thiessen polygons, the areas of the map closest to A would be shaded as 9/28 and the areas closest to B would be shaded as 11/3 switching at the midpoint between the two stations. This method does a better job of representing what happened in the example by not introducing October freeze dates where no October freezes occurred.

About Station Data

 Stations with excessive missing data could lead to spatial discrepancies in several of the ViP maps that are created at MRCC. Because the maps created are national in scope, it is impractical to check stations individually and determine if the missing data would affect the output. Therefore we have implemented a screening process that is run each time the data is accessed to omit stations with insufficient data. The process has multiple criteria that could lead to a station being omitted from the maps. The first criteria is to ensure the most recent data is complete such that each station must have 14 consecutive days of data for this period and must end no earlier than the day before yesterday. A second criteria is to ensure that there is sufficient data throughout the year. A station is omitted if the missing data for the year-to-date, or any consecutive 30-day period, tops 10%.