Hi Monica,


A few things:

If you start the simulation many months before the planting date, you are letting the soil get full of water (many months receiving rainfall and not having any evapotranspiration).  Depending on how you are simulating Nitrogen you may also be accumulating a lot of available Nitrogen.  Both of these situations may be different from reality if you are simulating a sequence of crops (e.g., soil profile may not be full of water, may have low values of NO3-N).

On the other hand, if you are using “Initial Conditions” in your X File, the model will use the soil water and Nitrogen content at the date that you set Initial Conditions (regardless of what happened before that date)

CERES is very sensitive to starting conditions (mainly initial water and available nitrogen [NO3-N]).  If you have measured starting conditions (soil water and NO3-N) you should use the date of the sampling (and the soil water and soil N values) as Initial Conditions.  If you do not have initial conditions measurement, I would use a starting date that is close to planting and set initial soil water and soil N values to levels that are reasonable for the production system that you are simulating (e.g., crop sequences, soil ability to mineralize N, etc.)

I hope this helps…

Walter Baethgen
IRI, Columbia University




> On Jul 13, 2015, at 11:26 AM, Monica Ortiz <[log in to unmask]> wrote:
> 
> Dear DSSAT community,
> 
> Hello, I was wondering if anyone could kindly explain if there was an 'optimal' start date for simulations? As I am familiarizing myself with the DSSAT (CERES-wheat) model I realize that there is some sensitivity to when I start simulations. 
> 
> I am simulating wheat production and for example, I have the following dates that I am using with DSSAT:
> 
> Planting date: (DDMMYYY): 13/10/2009
> Fertilizer application date: 21/04/2009
> 
> Therefore, my simulation starting date cannot be earlier than these two. I am arbitrarily setting all my simulation start dates to be 01/01/2009 (and other years) to be consistent, but I have noticed that the further away the date is the yield is higher (similar to this question: https://groups.google.com/forum/#!msg/dssat/yL-3JztB7kw/tktqrDw_fvEJ) 
> 
> Would it be possible for one of the model developers, or those well-versed in DSSAT to give me a bit of insight as to what "simulation start date" really means / signifies? I'd like to be able to quantify this uncertainty.
> 
> Thank you!
> 
> 
> Sincerely,
> 
> Monica Ortiz
> 
> University of Sheffield
> 
>