I would add to Ken's observations that across crop species, it is very
hard to model yield accurately under extreme stress. Among other things
in real-world situations, you often have loss of plant stand and with
severely stunted growth, a harsher microclimate than with a closed
canopy.
- Jeff White
-----Original Message-----
From: DSSAT - Crop Models and Applications
[mailto:[log in to unmask]] On Behalf Of Boote,Kenneth J
Sent: Tuesday, April 26, 2011 1:14 PM
To: [log in to unmask]
Subject: Re: Unrealistic maize yields
Lyndon,
I'm a bit surprised too, unless you start with too much soil moisture.
You said 20% starting soil moisture. If so, that is higher than your
DUL. Or did you mean 20% of available PAW, set in File X with XBUILD?
Another possibility is that you have genetic coefficients set too high
(starting with a very productive cultivar): P5=850.0 G2=800.0 G3=10.00.
Values of G2 and G3 this high should give 12 mt yield potential if you
have no water or N limitations. The G3 is higher than I've ever seen it
for CERES-Maize. In addition, you have an SLPF of 1.00 in the soil
file. A value of 1.00 should be reserved for the best soils in the
world (Midwestern USA or Argentine pampas). Your soil seems
unremarkable, with low SOC and should have a lower SLPF.
Ken Boote
-----Original Message-----
From: DSSAT - Crop Models and Applications
[mailto:[log in to unmask]] On Behalf Of Abraham Singels
Sent: Tuesday, April 26, 2011 12:12 AM
To: [log in to unmask]
Subject: Re: Unrealistic maize yields
Hi Lyndon
The late Andre du Toit made several changes to the Ceres Maize model to
get it to simulate maize growth and yield in South Africa better. Some
of these were published in the S.A. Journal of Plant and Soil. They may
have info that could help you.
To me the answer (at least an important part of it) lies in the weather
data. I could comment if you send me the daily data, especially
rainfall and reference evaporation. What are the lat and long of the
site?
Abraham Singels
-----Original Message-----
From: DSSAT - Crop Models and Applications
[mailto:[log in to unmask]] On Behalf Of Lyndon Estes
Sent: Sunday, April 24, 2011 10:01 PM
To: [log in to unmask]
Subject: Unrealistic maize yields
Dear DSSAT users,
I am busy with a country-scale (South Africa) maize study, and I am
getting
some unrealistically optimistic yield estimates in arid areas of the
country
where dryland maize is not viable. For example, here are a few summary
statistics (averaged over 19 years of runs, with reinitialization each
year)
for one location where mean annual rainfall is 276 mm.
YEARS PRCM yield cv.yield N failures days to maturity
1980-1999 159 1127 89 3 101
This was done with a 20% starting soil moisture content measured each
July
15th (soil NO3 = 0.9; NH4 = 0.1), with auto-planting occurring anytime
between the end of October and mid-January provided soil moisture in the
top
25 cm exceeds 70%. I am working with fairly minimal inputs,
particularly
for soil, since I couldn't get or extrapolate information on KSat, Ph,
CEC,
etc.
Here's the soil:
*LTAf000404 ARC LT -99 120 Landtype LTAf000404
@SITE COUNTRY LAT LONG SCS FAMILY
-99 S.Africa -99 -99 Landtype group Af
@ SCOM SALB SLU1 SLDR SLRO SLNF SLPF SMHB SMPX SMKE
-99 0.13 6.0 0.80 61.5 1.00 1.00 -99 -99 -99
@ SLB SLMH SLLL SDUL SSAT SRGF SSKS SBDM SLOC SLCL SLSI SLCF
SLNI SLHW SLHB SCEC SADC
13 A1 0.064 0.158 0.308 1.000 -99 1.83 0.09 2.0 -99 -99
-99 -99 -99 -99 -99
26 A2 0.064 0.158 0.308 1.000 -99 1.83 0.09 2.0 -99 -99
-99 -99 -99 -99 -99
40 A3 0.064 0.158 0.308 0.517 -99 1.83 0.08 2.0 -99 -99
-99 -99 -99 -99 -99
67 B1 0.062 0.169 0.335 0.343 -99 1.76 0.06 3.3 -99 -99
-99 -99 -99 -99 -99
94 B2 0.062 0.169 0.335 0.200 -99 1.76 0.02 3.3 -99 -99
-99 -99 -99 -99 -99
120 B3 0.062 0.169 0.335 0.118 -99 1.76 0.00 3.3 -99 -99
-99 -99 -99 -99 -99
And cultivar coefficients
KR0169 KROONSTAD(MICH) . IB0001 200.0 0.750 850.0 800.0 10.00 45.00
I throw down 32 kg/ha of ammonium nitrate at planting, and have a
planting
density of 2.0 and row-spacing
of 200 cm.
I have done some sensitivity analyses on certain variables:
X file
SH20 (doubles yield if increased to 50% soil moisture),
MESEV (changed to R--increases yield by 300 kg/ha)
EVAPO (changed to F--increased yield by 300 kg/ha)
Soil SLNF, SLPF, SALB, SLU1, SLDR (negligible effects)
Although I wasn't worried as much about them, I also changed SDUL, SLLL,
and
SAT values by 10% in the
direction that would produce less PAW. These had small effects (~ 100
kg/ha
or less--oddly, lowering SDUL
and SSAT produces small increases in yields). SLOC is already low, so I
wouldn't expect this to
be the culprit behind these unrealistic results.
I also tested different cultivar coefficients. These, together with
cultivars, together with SH20, had the
largest effects on yields, but none of these increased the number of
years
in which crops failed, which is
the value I was expecting to move. These remain constant at 3 no matter
how
I tweak the model.
There are of course other parameters to test in DSSAT, but these to me
seemed the most obvious places to look.
Since all these tweaks still end up with production in 16 out of 19
years in
a place that probably
shouldn't grow any maize, I was hoping for some pointers on where else I
might look to bring the simulations
closer to reality. My only other thought is that the model is
overestimating
soil moisture retention in the deep sandy soil, but I am not sure how I
would fix this outside of making the soil shallower. Perhaps I am
missing
something rather obvious in my settings?
Thanks in advance for any help.
Cheers, Lyndon
*********************************************************************
This email and all contents are subject to the following disclaimer:
http://www.sasa.org.za/emaildisclaimer619.aspx
*********************************************************************
|
