High Resolution Precipitation:
CMORPH, PERSIANN, TRMM, CPC_Famine

High Resolution Precipitation

Several high resolution precipitation datasets have beome available. These datasets should be viewed as experimental. The three datasets included in this suite of examples are all freely available. Some are in binary format and others in HDF.

Details descriptions of the datasets are available at:

1. CMORPH: Climate Prediction Center Morphing Technique
2. PERSIANN: Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks
3. TRMM: Tropical Rainfall Measuring Mission
4. CPC Famine Early Warning System daily estimates (New_Algo).

NCL Comments

Generally, the examples below process only one file. To process multiple files would require only minor additions to the sample codes.

        dir  = "./"               ; input directory 
        fili = systemfunc("cd "+dir+" ; ls root*") ; 'root' is unique
        nfil = dimsizes(fili)
    
        do nf=0,nfil-1
            :
            ; change all fili to fili(nf)
            :
        end do

Reading "big endian" and "little endian" binary file types is readily accomplished via

        setfileoption("bin","ReadByteOrder","BigEndian")
    or
        setfileoption("bin","ReadByteOrder","LittleEndian")

Interpolating high spatial/temporal resolution precipitation fields, is best accomplished via area_hi2lores rather than (say) linint2.

If netCDF creation is desired and file space is a concern, it may be best to "pack" the precipitation values. Using pack_values will create a file half the size of those created using float values. Some precision is lost but is not important here.

------------------------------------------------------------------------

Often, precipitation variables contain N-hourly accumulated totals, where N=1 or 3 or 6 or 12. A common question: Given hourly precipitation (eg, 0Z, 1Z, 2Z, ..., 23Z, 0Z, ...), how can (say) 6-hourly (0Z, 6Z, 12Z, 18Z) totals be calculated? Please see Example 5 of the dim_sum_n function.