Gridded Mean Areal Precipitation Processing

This section explains how MAP time series are created from a field of gridded precipitation estimates (usually derived from radar data) using the MAPX preprocessor. The input to MAPX is a temporal sequence of grids covering the entire RFC area. A precipitation value is associated with each grid cell.

Several methods for producing gridded precipitation fields have been used in the NWS, including Stage III, P1, and RFCWide Multisensor Precipitation Estimation (MPE). Gridded future precipitation, or QPF, are also available. RFCWide, the current nationally-supported precipitation processing systems (PPS), takes raw radar/rainfall fields, gage only fields, and satellite-based rainfall fields and combines them in a single step to generate a merged precipitation field for an RFC area. For training on precipitation processing, see the 'Assimilating Hydrometeorological Data PDS' at: http://www.nwstc.noaa.gov/d.ntp/hydro/asipds.html

So long as gridded precipitation data is provided in the required input file format, the MAPX preprocessor does not care about which PPS was used to develop the precipitation field. For a given basin, the grid for a one hour period is superimposed on the basin and the grid cells that are in the basin are identified.

The gridded field should be thought of as lines and columns of precipitation values.  For each line, the processor adds the precipitation value in each grid to a running sum, if the grid cell is in the basin.  The final sum is then divided by the total number of grid cells in the basin to get an average value or MAP value.  Consider the case of a hypothetical storm with a grid and watershed outline superimposed, as shown below.  In this example illustrated in Figure 1, color represents rainfall depth.

Figure 1

We must determine which grid cells are in the basin and which are out.  In other words, we must define the basin boundary with grid cell.  We cannot represent the boundary at partial grid cells -

Figure 2

 – but rather it must be total grids cells.  The only grid cells considered to be in the basin are those cells which have their center in the basin. Using the center of each grid cell as the determining factor, we would define the basin as in Figure 3:

Figure 3

We now see the storm with the original basin and the grid based basin superimposed in Figure 4.  You can see that the first grid cell that is contained in the basin is in line #2 and column #4.

Figure 4

The mean areal precipitation is then computed by summing the precipitation value of all grid boxes in the basin and dividing by the total number of grid boxes. This process is repeated on each basin for each hour of precipitation being processed until the end of available gridded data. The resulting time series of hourly precipitation is written to the NWSRFS processed database. If missing data are encountered at any grid point, a missing value is written for the hour in which it occurs. Note that if a three or six hour time series is required for modeling, this will have to be generated in a separate step (such as CHANGE-T) in the forecast simulation. As previously indicated, the MAPX preprocessor operates independent of the source of the gridded precipitation fields.