Lyndon,

The only other thing I can think of (other than water), would be that you have more N mineralized by the time that sowing occurs, and that is what increases yield of maize.  Look for the nitrate and ammonium levels in the soil on the day of sowing.  This would be especially true if you use CENTURY for SOM, as CENTURY tends to mineralize more N than the PAPRAN (Godwin) version.

What do I recommend?  I dislike having any difference between start of simulation and the sowing date.  So, I would make the two exactly the same.  Why?  Because if I want to look at crop ET, then pre-sowing E confounds the ET.  Also, you have the issues of N leaching or N mineralization during the pre-sowing period that you may be seeing.

So, I would always use start of simulate to be the same as the sowing date.  The only exception is when I want to use automatic sowing option in the model (based on window of sufficient soil water or soil temperature, for example).

I hope this helps.

Ken Boote

-----Original Message-----
From: DSSAT - Crop Models and Applications [mailto:[log in to unmask]] On Behalf Of Lyndon Estes
Sent: Wednesday, September 08, 2010 6:41 PM
To: [log in to unmask]
Subject: Sensitivity to simulation starting date

Dear DSSAT users,

Hello, I had two quick questions related to simulation starting date.

The first is related to the fact that the model seems quite sensitive to the
starting date variable (SDATE)--the earlier you start, the higher yields
tend to be. I had assumed this was because more water was being stored in
the soil (rainfall accumulates for longer), but when I compared two maize
runs on the same soil, both with planting date of 80320, one started on
80001, the other on 80244, the 80001 start had a higher yield (5784 vs 4547)
but lower values for SWTD, SWXD, SW1D, etc. on the day of planting.

I am thus unclear as to what causes this fairly large yield discrepancy? I
was thus hoping someone could inform me as to the reason for this
discrepancy.

The second question I had relates to the above. Given these differences,
what is the "soundest" starting date to work with when running simulations?
What I am doing is running simulations year after year, but not in
sequeuntial mode (i.e. I am not doing rotations or interested in seeing how
soil properties change over time).  I thus code the X file to begin on 1st
year, e.g. 1980, and then let it run for 20 years or so (this is mostly to
avoid writing enormous numbers of X files).  e.g.

@N GENERAL     NYERS NREPS START SDATE RSEED SNAME....................
SMODEL
 1 GE             19     1     S 80001  2150 Sname

I had figured I would let it start at the beginning of each year, but given
the sensitivity, I was hoping for advice as to a good rule of thumb for
starting dates (if one exists)? Where I am working, the growing season
starts anywhere from October to December, with spring rains following a dry
winter starting Sept-November.

Thanks in advance for any advice.

Best, Lyndon