NASA GISS: ANN FRIDLIND: TWP-ICE/ACTIVE CASE STUDY

TWP-ICE/ACTIVE MONSOON CASE STUDY

Precipitation rate measurements (white bars) versus simulations (colored bars):

Comments:

Frequency distributions are calculated for six time ranges. The first three are longer time ranges: (1) all times following an initial 36 hours for model spin-up, (2) six days of monsoon, and (3) six days of suppressed conditions. The second three are sequential monsoon events of varying intensity.
Precipitation rate retrievals from the C-Pol radar at Darwin are supplied by the Australian Bureau of Meteorology. Retrievals represent instantaneous values reported at 10-min and 2.5-km resolution at an elevation of 2.5 km over an area of approximately 31,000 km².
All measurement data have been downloaded from the DOE ARM data archive (www.arm.gov) and use in publications is subject to the policies described there. Any errors in plotting or representation shown here should be blamed on Ann Fridlind.
Model results are instantaneous values at an elevation of 2.5 km, averaged to 2.5-km horizontal resolution, reported at 10-min frequency over a domain size of 31,000 km².

Preliminary summary:

During active periods, the lowest rain rates are systematically predicted too often and rates in the 2-20 mm/h range are predicted too infrequently. A closer look at the fractional amount of precipitation occurring in each rain rate range indicates that while model domain-mean precipitation rates closely match C-pol measurements (see previous page), the fractional amount of precipitation occurring in the highest rate bin is overestimated at the expense of precipitation occurring in the mid-range bins. Overestimates are similar in magnitude to the degree of overestimate of domain-wide instantaneous peak rain rates (see previous page). Three-hour sample frequency of both radar and model data is used for the following comparisons (precipitation fraction not available at 10-minute frequency; 3-h sampling is grossly insufficient for analysis of cirrus periods, but appears adequate for rough analysis of active periods):