The SarraH SarraO Crop Models
Transcript of The SarraH SarraO Crop Models
An
oth
erS
ahel
is
po
ssib
le !
Coupling remote sensing data with crop models for
crop monitoring and yield forecasting in West Africa
Training course on the use of Satellite based data and crop monitoring and forecasting tools for drought monitoring and agro-m eteorological applications
06-10 May 2019, Nairobi, Kenya
Christian Baron, Agnès Bégué, Mathieu Castets, Camille Jahel, Danny Lo Seen
CIRAD – TETIS Research UnitMontpellier, France
Seydou B. Traoré, Alhasane Agali, Henri Songoti,
AGRHYMETNiamey, Niger
An
oth
erS
ahel
is
po
ssib
le !
2
Agricultural campaign monitoring
• Use of the DHC crop simulation model
• Identification of zones where rainfall at the start of the season allowed sowing
• Later verification of successful sowing
• At the end of July, analysis of the sowing situation across CILSS member countries
– Focus on zones with late sowing relatively to the average and the previous year
Crop Yield Forecasting at AGRHYMET
An
oth
erS
ahel
is
po
ssib
le !
3
Agricultural campaign monitoring
• Water requirement satisfaction indices
• Soil water reserves at the end of the dekad
• Water requirements for remaining crop cycle
• Issuing alerts in case of prolonged water deficit (2 successive dekads)
• First estimation of potential crop yields at the end of August
• Update every dekad thereafter
Crop Yield Forecasting at AGRHYMET
An
oth
erS
ahel
is
po
ssib
le !
4
Possible Sowing dates
Potential crop yields anomalies
Crop water requirementsatisfaction indices
Outputs of the DHC model
Crop Yield Forecasting at AGRHYMET
An
oth
erS
ahel
is
po
ssib
le !
Some deficiencies of the DHC model
– Valid only for millet in the sahelian zone
– Underestimation of yields under optimal wateringconditions (< 1000 kg/ha all the time)
– Does not account for photoperiod sensitivity of local cropvarieties still predominantly used by farmers
Crop Yield Forecasting at AGRHYMET
An
oth
erS
ahel
is
po
ssib
le !
• The SARRA -H model
– Outcome of researches conducted at CIRAD incollaboration with African partners in the framework ofinternational projects (CERAAS, PROMISE, AMMA)
– Simulate both water and carbon balances (morephysiologically oriented)
– Can be used for several crop types and agroclimatic zones
– Good results with on station experimental data
The Advent of SARRA -H
An
oth
erS
ahel
is
po
ssib
le !
Maintenance respirationRain
ETo
T °
Sarra-H: a determinist model, simple & robust
Maintenancerespiration
7
An
oth
erS
ahel
is
po
ssib
le !
Impact of water stress on biomass
b a m b e y P lu ie
B i o m A e ro (b a m b e y P l u i e ) N u m P h a se (b a m b e y P l u i e ) B i o m A e ro (B a m b e y E t m ) N u m P h a se (B a m b e y E t m )
N b J a s8 07 57 06 56 05 55 04 54 03 53 02 52 01 5
1 3 0 0 0
1 2 0 0 0
1 1 0 0 0
1 0 0 0 0
9 0 0 0
8 0 0 0
7 0 0 0
6 0 0 0
5 0 0 0
4 0 0 0
3 0 0 0
2 0 0 0
1 0 0 0
0
7
6
5
4
3
2
1
0
Above ground biomass and grain yield (kg/ha)
PanicleInitiation
Anthesis
Water Stress
PotentialYield
Day after sowing8
Water Stress
An
oth
erS
ahel
is
po
ssib
le !
• On station trials to characterize the most used local or about to be released millet, sorghum and maize varieties for the parameterization of the SARRA-H model
• On farm agronomic surveys to evaluate the model
• Several sites with contrasted agroclimatic conditions and cropping systems
Adaptation and Evaluation of SARRA -H
An
oth
erS
ahel
is
po
ssib
le !
10
• Two millet varieties
• Two sowing dates,
• Two nitrogen fertilization levels (N0, N1)
0500
100015002000250030003500400045005000550060006500700075008000
Dry
wei
ght (
kg h
a-1 )
Observation date
HKP x N1
0500
100015002000250030003500400045005000550060006500700075008000
Observation date
Total abovegroundLeavesGrains
MTDO x N1
0500
100015002000250030003500400045005000550060006500700075008000
Dry
wei
ght (
kgha
-1)
Observation date
HKP x N0
0500
100015002000250030003500400045005000550060006500700075008000
Observation date
Total abovegroundLeavesGrains
MTDO x N0
Adaptation and Evaluation of SARRA -H
AGRHYMET SiteAgali’s PhD thesis
An
oth
erS
ahel
is
po
ssib
le !
11
Sowing June 17 Sowing July 17 Sowing August 17
M. Kouressy & al., 2008, Adaptation to diverse semi-arid environments of sorghum genotypes having different plant type and sensitivity to photoperiod
a
V1
Calendar date
1/7 1/8 1/9 1/10 1/11 1/12 1/1
Dry
wei
ght (
Mg.
ha-1
)
0
5
10
15
20c
V3
1/7 1/8 1/9 1/10 1/11 1/12 1/1
Date 1, abovegroundDate 2, abovegroundDate 3, abovegroundDate 1, grainDate 2, grainDate 3, grain
b
V2
1/7 1/8 1/9 1/10 1/11 1/12 1/1
Same Variety
Same Guy DifferentSowing dates
Parameterizing photosensitive varieties
An
oth
erS
ahel
is
po
ssib
le !
Adaptation and Evaluation of SARRA -H
SENEGAL
Diourbel (450 mm)
Tambacounda (800 mm)
MALI
Cinzana (550 mm)
Koutiala (700 mm)
BURKINA FASO
Tougou (600 mm)
Dano (900 mm)
NIGERNiamey (500 mm)Bengou (700 mm)
On-farm surveys & Experimental trials
Millet Varieties
Sorghum Sowing dates
Maize Planting densities
An
oth
erS
ahel
is
po
ssib
le !
0,00
0,10
0,20
0,30
0,40
0,50
0,60
0,70
0,80
0 5000 10000 15000 20000
ratio
leav
es /
(leav
es+s
tem
s)
leaves + stems biomass (kg ha-1)
Souna
Thialack
Sanio
HKP
MTDO
Zatib
Choho
Toroniou
SNTC
Adaptation and Evaluation of SARRA -H
Duration of the sowing to flag leaf stages of different local sorghum varieties in Mali
Allometric relationships of different local millet varieties in Senegal, Mali, and Niger
An
oth
erS
ahel
is
po
ssib
le !
Simulation effectuée avec SarraH v3.2 - Modèle SarrahV3.2 - http://ecotrop.cirad.fr
Lai(DMR_S1_V3.2) Lai(DMR_S1_V3.2) BiomasseAerienne(DMR_S1_V3.2)BiomasseFeuil les(DMR_S1_V3.2) Rdt(DMR_S1_V3.2) BiomasseAerienne(DMR_S1_V3.2)BiomasseFeuil les(DMR_S1_V3.2) Rdt(DMR_S1_V3.2)
Date10/07/1225/06/1210/06/1226/05/1211/05/1226/04/12
m²/m
²
3
2
1
0
kg/ha
10 500
10 000
9 500
9 000
8 500
8 000
7 500
7 000
6 500
6 000
5 500
5 000
4 500
4 000
3 500
3 000
2 500
2 000
1 500
1 000
500
0
LAI Above ground Biomass
Leaf Biomass Yield
14
Sarra-H performances 1
Thanks to Ulrich, Cirad PHD student, (Maïze experimentation in Benin, 2012)
An
oth
erS
ahel
is
po
ssib
le !
Simulation effectuée avec SarraH v3.2 - Modèle SarrahV3.2 - http://ecotrop.cirad.fr
Lai(B2BresIrrigV3.2)Lai(B2BresIrrigV3.2)BiomasseAerienne(B2BresIrrigV3.2)BiomasseFeuil les(B2BresIrrigV3.2)Rdt(B2BresIrrigV3.2)BiomasseAerienne(B2BresIrrigV3.2)BiomasseFeuil les(B2BresIrrigV3.2)Rdt(B2BresIrrigV3.2)
Date24/01/0425/12/0325/11/0326/10/03
m²/
m²
6
5
4
3
2
1
0
kg/ha
26 000
24 000
22 000
20 000
18 000
16 000
14 000
12 000
10 000
8 000
6 000
4 000
2 000
0
Maize varieties inMali, Benin, Brazil, Tanzania, USA, France
Simulation effectuée avec SarraH v3.2 - Modèle SarrahV3.2 - http://ecotrop.cirad.fr
Lai(Souna96PluieV3.2)Lai(Souna96PluieV3.2)BiomasseAerienne(Souna96PluieV3.2)BiomasseFeuil les(Souna96PluieV3.2)Rdt(Souna96PluieV3.2)BiomasseAerienne(Souna96PluieV3.2)BiomasseFeuil les(Souna96PluieV3.2)Rdt(Souna96PluieV3.2)
Date02/10/9602/09/9603/08/96
m²/
m²
4
3
2
1
0
kg/ha
8 000
7 000
6 000
5 000
4 000
3 000
2 000
1 000
0
Pearl Millet varieties inMali, Niger, Senegal, Burkina Faso…(Photoperiodic and non photoperiodic)
Simulation effectuée avec SarraH v3.2 - Modèle SarrahV3.2 - http://ecotrop.cirad.fr
Lai(GuineaAmD104SarV3.2)Lai(GuineaAmD104SarV3.2)BiomasseAerienne(GuineaAmD104SarV3.2)BiomasseFeuilles(GuineaAmD104SarV3.2)Rdt(GuineaAmD104SarV3.2)BiomasseAerienne(GuineaAmD104SarV3.2)BiomasseFeuilles(GuineaAmD104SarV3.2)Rdt(GuineaAmD104SarV3.2)
Date19/11/0420/10/0420/09/0421/08/0422/07/04
m²/
m²
5
4
3
2
1
0
kg/ha
14 000
12 000
10 000
8 000
6 000
4 000
2 000
0
Sorghum varieties inMali, Kenya, Burkina Faso…(Photoperiodic and non photoperiodic)
Simulation effectuée avec SarraH v3.2 - Modèle SARRAHMil2 - http://ecotrop.cirad.fr
Lai(SarMil2AntsiE933)Lai(SarMil2AntsiE933)BiomasseAerienne(SarMil2AntsiE933)BiomasseFeuilles(SarMil2AntsiE933)
Date23/04/0423/02/0425/12/03
m²/
m²
3
2
1
0kg/ha
12 000
11 000
10 000
9 000
8 000
7 000
6 000
5 000
4 000
3 000
2 000
1 000
0
Rainfed Rice variety in Madagascar
SARRA-Hcatches the variability
There is also wheat & Soy bean
(France)… and
last scoop Coton (Cameroun)
Thanks to Seydou, Agali, Michel , Mamoutou, Bertrand, Fernando….
15
An
oth
erS
ahel
is
po
ssib
le !
Predictive Capacity of the SARRA -H model Maize in Burkina Faso
16
0 1000 3000 5000
010
0020
0030
0040
0050
0060
00
observed yield (kg/ha)
Sim
ulat
ed y
ield
(kg/
ha)
R²= 0.8054
P<0,05
(2014)
0 1000 3000 5000
010
0020
0030
0040
0050
0060
00
Observed yield (kg/ha)
R²=0,8561
P<0,05
(2016)
An
oth
erS
ahel
is
po
ssib
le !
Adaptation to crop monitoring needs
• User interface allowing easy execution of routine tasks (coupling with R ),
• Execution at the dekadal time step of simulations on crop status and yield prediction
• Combined use of historical and current climate data for the prediction
• Combination of several scenarios (crop species and varieties, sol types and depth)
• Mapping of results
An
oth
erS
ahel
is
po
ssib
le !
Adaptation to crop monitoring needs
Crop water requirement satisfaction indices
Soil water reserves
Past dekad
Average since sowing
An
oth
erS
ahel
is
po
ssib
le !
Adaptation to crop monitoring needs
Expected crop yields relatively to average
8 years out of 10
5 years out of 10
2 years out of 10
An
oth
erS
ahel
is
po
ssib
le !
Need for a Spatialized crop model
• SarraH– Dry cereal crops (millet , maize, sorghum)– Water, carbon budgets, phenelogy– Daily rainfall, meteorological input data from stat ions
• Crop model spatialization– Ocelet spatial dynamics modelling environment– Same SarraH processes– Gridded and shape GIS input data– Rainfall estimates from satellite (e.g. TAMSAT, CHIRPS)
An
oth
erS
ahel
is
po
ssib
le !
Simulation protocol
• Input maps– Soil (FAO, 13 soil types) for estimating available wate r
capacity– Daily TAMSAT rainfall estimates– 10 days ECMWF data (PET, Rg, Tmin, Tmax)– Onset and End of rainy season (map kriged from histo rical
time series)
• Parameters– 4 fertility levels: very low to potential – 2 soil depths: shallow (60 cm), deep (2 m) – 8 Crop varieties (2 photoperiodic, 3 fixed cycle x 2 cycle
length)– 3 sowing date search methods (first rains, early, from end)
An
oth
erS
ahel
is
po
ssib
le !
Input images and maps
22
Rainfall Radiation Temp Min Temp Max Evapotranspiration
Soil Map
Onset dates
Cessation dates
An
oth
erS
ahel
is
po
ssib
le !
Yield anomaly evolution in Mali
-50
-40
-30
-20
-10
0
10
202007 2009 2011 2013 2015
Ano
mal
ies
de R
ende
men
t %
Années
2 months before
At harvest
Potential Yield at Anthesis
Potential Yield at Harvest
SARRA-O, UN MODELE ET LOGICIELDU CIRAD, UMR TETIS
23
An
oth
erS
ahel
is
po
ssib
le !
-50
-40
-30
-20
-10
0
10
202007 2009 2011 2013 2015
Ano
mal
ies
de R
ende
men
t %
Années
SARRA-O, UN MODELE ET LOGICIELDU CIRAD, UMR TETIS
24
Yield anomaly evolution in Mali
2 months before
At harvest
Potential Yield at Anthesis
Potential Yield at Harvest
An
oth
erS
ahel
is
po
ssib
le !
-50
-40
-30
-20
-10
0
10
202007 2009 2011 2013 2015
Ano
mal
ies
de R
ende
men
t %
Années
SARRA-O, UN MODELE ET LOGICIELDU CIRAD, UMR TETIS
25
Yield anomaly evolution in Mali
2 months before
At harvest
Potential Yield at Anthesis
Potential Yield at Harvest
An
oth
erS
ahel
is
po
ssib
le !
-50
-40
-30
-20
-10
0
10
202007 2009 2011 2013 2015
Ano
mal
ies
de R
ende
men
t %
Années
SARRA-O, UN MODELE ET LOGICIELDU CIRAD, UMR TETIS
26
Yield anomaly evolution in Mali
2 months before
At harvest
Potential Yield at Anthesis
Potential Yield at Harvest
An
oth
erS
ahel
is
po
ssib
le !
-50
-40
-30
-20
-10
0
10
202007 2009 2011 2013 2015
Ano
mal
ies
de R
ende
men
t %
Années
SARRA-O, UN MODELE ET LOGICIELDU CIRAD, UMR TETIS
27
Yield anomaly evolution in Mali
2 months before
At harvest
Potential Yield at Anthesis
Potential Yield at Harvest
An
oth
erS
ahel
is
po
ssib
le !
-50
-40
-30
-20
-10
0
10
202007 2009 2011 2013 2015
Ano
mal
ies
de R
ende
men
t %
Années
SARRA-O, UN MODELE ET LOGICIELDU CIRAD, UMR TETIS
28
Yield anomaly evolution in Mali
2 months before
At harvest
Potential Yield at Anthesis
Potential Yield at Harvest
An
oth
erS
ahel
is
po
ssib
le !
-50
-40
-30
-20
-10
0
10
202007 2009 2011 2013 2015
Ano
mal
ies
de R
ende
men
t %
Années
SARRA-O, UN MODELE ET LOGICIELDU CIRAD, UMR TETIS
29
2 months before
At harvest
Yield anomaly evolution in Mali
Potential Yield at Anthesis
Potential Yield at Harvest
An
oth
erS
ahel
is
po
ssib
le !
-50
-40
-30
-20
-10
0
10
202007 2009 2011 2013 2015
Ano
mal
ies
de R
ende
men
t %
Années
SARRA-O, UN MODELE ET LOGICIELDU CIRAD, UMR TETIS
30
2 months before
At harvest
Yield anomaly evolution in Mali
Potential Yield at Anthesis
Potential Yield at Harvest
An
oth
erS
ahel
is
po
ssib
le !
-50
-40
-30
-20
-10
0
10
202007 2009 2011 2013 2015
Ano
mal
ies
de R
ende
men
t %
Années
SARRA-O, UN MODELE ET LOGICIELDU CIRAD, UMR TETIS
31
Yield anomaly evolution in Mali
2 months before
At harvest
Potential Yield at Anthesis
Potential Yield at Harvest
An
oth
erS
ahel
is
po
ssib
le !
32
Evolution of above ground biomass in 2013 (Top) and 2015(Bottom)
Daily time step, 3.5 km resolution
2015 had a delayed onset and a below average cumulative rainfall
An
oth
erS
ahel
is
po
ssib
le !
Ongoing and future works
• Prototype testing– Tested at AGRHYMET in 2015– Outputs used in bulletins since 2016– Mid-season simulation with different rainfall scena rios
• Enhancements– Improved rainfall estimates (TAMSAT + SMOS + Mergin g
with Station data)– Start and cessations dates calculation with remote sensing
data
An
oth
erS
ahel
is
po
ssib
le !
Flash info November 2016
Monitoring the 2016 season at AGRHYMET
34
An
oth
erS
ahel
is
po
ssib
le !
Thank you for your attention
http://agrhymet.cilss.int/[email protected]