Dear Ziad,

I apologize for the delay, various events outside my control have
conspired.

Please find attached sample qced processed JWD files from Kwajalein
the output format is as follows:

yyyy mon day hour min sec [20 categories of N(D) in units of mm^-3 mm^-1] dBZ R W waldlambda waldNo Dm Nostar totalrawcts accumperiodsec elapsedsec

2003 10 11 15 58 0  654.86 1150.91 1072.79 821.27 397.11 311.40 330.66 220.37 155.61 111.76 81.21 40.80 12.72 4.16 1.40 0.20 0.15 0.03 0.00 0.00  38.54 10.46 507.04 2.53 6649.90 1.73 4523.84 16192 1800 1800

R in mm/hr
W in mm^3/m^3
waldlambda in mm^-1 calculated using eqn 5 in Waldvogel (1974)
waldNo in mm^-3 mm^-1 calculated using eqn 4 in Waldvogel (1974)
Dm in mm and is the ratio of 4th to 3rd moment of DSD
Nostar in mm^-3 mm^-1 calculated using eqn 8 in Testud et al. (2001) 
rawcount = integer total number of drops hitting disdrometer over time period
accumperiodsec = total rainy seconds in accumulation period
elapsedsec = total sec elapsed time from start to end of accumulation period

I'm looking for feedback on if these parameters are what you need and
if you would like me to add anything else.

The processing accumulates the raw counts over the designated time
period.  For this example, there are two files, for the first the time
period is 10 min consecutive, i.e. it has to rain for 10 consecutive
minutes to be included in the sample.  The second file uses a time
period of 30 consecutive minutes. I can also do elapsed time but these
will be of lower quality.

Since the deadtime correction is non-linear, the raw counts are first
summed for each size category and then the dead time correction is applied.
The last step in the computation of nD utilizes the fall velocity
for each size bin to determine n(D) in mm^-3 mm^-1.

There are bias in the integrated parameters and in the shape of the
DSD which decrease as rawcounts increases (see Smith et al. 1993).  I
highly advise against using any accumulation less than 10 min as they
have enormous sampling errors.

My suggestion is to filter on rawcounts and to focus on data where
rawcounts >=3000 to minimize these errors.  Joss has commented to me
that the strength of his device is the calculation of the integrated
parameter R rather than n(D).

Let me know if you have any questions. If you want any changes
please let me know. Once we have settled on format I can
quickly churn out the data for you.

Also note that we are getting about 200-300 rainy 10 min samples per
rainy season month. 

Please also I would ask you to confer with me if you want to present
or write something up on these data as I am preparing a paper and it
would be better not to duplicate effort (or disagree without knowing
ahead of time regarding interpretations of the same data set).

Sandra