A gridded dataset can be either 'area average' where the value assigned to the grid cell represents the average measurement value over the area of the grid cell or a point value where the grid cell measurement value is actually a discrete measurement value at the center point of the grid cell. Gridded radar data is typically supplied as an average over area grid. Other types of data may be some sort of sampled data at discrete point locations at the grid center points. When evaluating a gridded dataset as a TIN, the program assigns the measurement value for the grid cell to the location at the center of the TIN. For gridded datasets where the grid cell value represents an average over area, this representation as a TIN is an approximation of the surface. Some algorithms in the program, such as the Depth-Area-Duration tool have the capability to process the TIN as either a TIN or grid cell average. Under normal circumstances, this approximation is insignificant. However, in the case where an area being analyzed is smaller than the area of just a few grid cells, the combination of the modelling grid cell average data as point values and the resolution of a degenerate dataset with grid cells bisected in an arbitrary direction could give inaccurate results.

For example, consider a group of 9 grid cells where the center cell has a measurement value of 10 and the surrounding 8 grid cells have a 0 measurement value. If a polygon, shown as a thick black outline, occupied the same area as the center grid cell, considering the polygon as a basin average would obviously give an average over the area of 10. However when modeled as a TIN, the interconnected point values now would give a surface that resembles a pyramid and results in average of 5.9. Also note that the direction of the degeneracy resolution could impact the results if the surrounding points were not all the same value.

By the time the polygon encompasses a total of 4 grid cells made up of the center cell and ½ of the surrounding grid cells, the TIN approximation of the grid and the computation of the grid as cell averages give the same value of 2.5. This leads to the conclusion that any results for polygons with an area smaller than about 4 grid cells should be inspected closely.