Understanding the climate effects on rice production using BigData

www.ciat.cgiar.org Agricultura Eco-Eficiente para Reducir la Pobreza

Understanding the climate effects on rice production using BigData

Big Data

Daniel Jiménez, Sylvain Delerce, Hugo Dorado, Camila Rebolledo, Edgar Torres

Context •Within the framework “Convenio MADR-CIAT” climate change project •As part of the adaptation strategy – SSA

•Crop sector (FEDEAROZ) holds a lot of information on climate and productivity •Empirical hypothesis of FEDEARROZ needed to be proven

+ + =

Climate Soil Crop management productivity/ha

Objectives

To: •Evaluate multivariate modeling techniques (parametric and non-parametric) to determine their suitability as tools for modeling the response of rice to variation in climate •Provide de crop sector with scientific evidence of the effect of climate on rice productivity •Identify the combination of factors that lead to high productivities

Methods

•Regressions (Linear & Non-linear)

Obs Climate Yield/

Plot

1 X1 X2 X3 X4…Xn

Y1

2 X1 X2 X3 X4…Xn

Y2

3 X1 X2 X3 X4…Xn

Y3

4 Y4

…..

500 X1 X2 X3 X4…Xn

Yn

Ordinary least squares - linear

•ANNs – Non-linear

Yield/Plot= temp (b1) + rainfall (b2) …+ (B)

Sowing Harvest

a cropping event in rice = about 120 days

Climate series for all variables

Plot

time

Hypotesis Yield variation in Saldaña (research station) is associated with climate

Variables profiles

0

1500

3000

4500

6000

7500

9000

34 35 36 37 38 39

Re

nd

imie

nto

Tmax

Tmax

0

1500

3000

4500

6000

7500

9000

52000 53000 54000 55000 56000 57000 58000

Re

nd

imie

nto

Ener_accut

Ener_accu

Multivariate analysis for Saldaña (research station ): cropping events (2010 to 2012), variety FEDEARROZ 733

10.43

6.20 6.03 4.78

3.76 3.74

1.92

0.00

2.00

4.00

6.00

8.00

10.00

12.00

% d

e v

aria

nza

exp

licad

a

Fedearroz 733

Non- lineal

N = 98

Crop sector (FEDEARROZ) -> Sharing information and obtaining new insights

FEDEARROZ 733, 37% of productivity variation explained

Multivariate analysis for Saldaña (research station ): cropping events (2010 to 2012), con por variedad

Lagunas, 22% of productivity variation explained

10.43

6.20 6.03 4.78

3.76 3.74

1.92

0.00

2.00

4.00

6.00

8.00

10.00

12.00

% d

e v

aria

nza

exp

licad

a

Fedearroz 733

8.05

6.57

3.53

1.26 1.03 0.94 0.50

0.00

2.00

4.00

6.00

8.00

10.00

% d

e v

aria

nza

exp

licad

a

Lagunas

N = 98

N = 112

Crop sector (FEDEARROZ) -> Sharing information and obtaining new insights

Climate (%) + Soil (%) + Crop management (%) = productivity/plot

Cómo aumentar la predicción?

Analysis based on phenological stages in Saldaña : multidiciplinary work!

VEG

Ini Pan

FLOR

VEG

Ini Pan

FLOR

Variedad 1 Variedad 2

Siembra Cosecha

Vegetative stage

Panicle initiation

Flowering

Grain filling

Rice

17.76

6.03

3.06 2.74 2.56 1.87 1.56 1.51 1.46 1.38 1.31

0.85 0.69 0.53 0.53 0.50

0

5

10

15

20

Var

ian

za e

xplic

ada

• The crop sector can suggest to farmers the best date for planting • By assessing the same approach in other stations (enviroments) – New insights for

future breeding • Adaptation strategy for climate change

Analysis based on phenological stages in Saldaña (research station) (FEDEARROZ) - N= 329 (cropping events)

We explained more than 40 % of productivity variation of rice

Variable profile (Eneraccu_llen – Radiation)

Crop sector (FEDEARROZ) -> Sharing information -> CIAT working together -> obtaining new insights!!!

Conclusions and perspectives

•The analytical tools used demonstrated that variation of rice productivity in Saldaña can be associated with climate

•Optimization of the crop system- Site-specific conditions (germplasm, environment, crop management) •As long as the information is available it can be applied in any other region