AfricaGIS 2013 - GSDI 14 - Global Geospatial Conference 2013 Addis Ababa, Ethiopia, November 4-8, 2013 By Antoine DENIS – PhD student - University of Liège.

AfricaGIS 2013 - GSDI 14 - Global Geospatial Conference 2013Addis Ababa, Ethiopia, November 4-8, 2013

By Antoine DENIS – PhD student - University of Liège - Belgium

Can satellites help organic cotton certification ?

Remote sensing and GIS techniques for supporting organic cotton certification process in West Africa

1. Context & Justification2. Objectives3. The IDEA4. Hypothesis

5. Method6. Results

7. Discussion and conclusion

Can satellites help organic cotton certification ?

1. Context & Justification

Want organic food/products?

Human health Environmentally friendly

1. Context & Justification

Organic crop ? = NO chemical synthetic pesticide & fertilizer= NO GMO= Crop rotation= Organic fertilizer and pesticide= ...

Crop Control· Yearly farm inspection· Documentary accounts· + Unannounced inspection· + Laboratory analysis · Cost? Frequency? Remote areas?

Certification Rules & agencies Labels

Organic food onthe market

To trust or not to trust ?

1. Context & Justification

1. Context & Justification

Why the Burkina Faso ?

Interest from organic certification bodies for developing countries:• Huge amount of organic products ($)

• Remote areas and certification control more difficult

Why the cotton?• Need a crop certified as organic

• That can be studied by RS

• Field big enough

Economic Importance of cotton in Burkina Faso• Cotton accounts for 50 to 60% of the

country’s foreign currency earnings • Cotton is the first export product

contributing largely to the country’s economic development

1. Context & Justification

Development of Organic cotton in Burkina FasoSuccessful since 2004, bright example of sustainable development that contributes to:• Alleviation of poverty•Improved food security by enhancing producers’ income with less risk to run into debt

1. Context & Justification

• Healthy way to crop both for people and the environment resulting in improved human and animal health (absence of chemical pesticides), and improved soil fertility and environment (organic cropping technique).

2. Objectives

Is it possible, in the context of South-West Burkina Faso,

► To help organic cotton certification process with satellites?► To discriminate organic and conventional cotton fields with satellites?

► Need to assess the bio-chemico-physical difference between organic and non organic cotton with diverse field measurements

2. Objectives

3. The IDEA

BIO!

BIO!

BIO!BIO

!BIO

!

BIO!

BIO! BIO

!

BIO!

BIO!

BIO!

BIO!

BIO!

BIO!BIO

!BIO!

BIO!

BIO!

BIO!

BIO!

BIO!

Identification of area to control

Field declared as organic !

Indicator computation

Identification of suspect fields

Too high nitrogen!

Analysis with several indicators

4. Hypothesis

4. Hypothesis

► Management differences between organic and conventional

crops

► Difference in crop bio-chemico-physical

characteristics and general field appearance

► Observable by satellites and transformable into

satellites derived indicators

Cotton management differences

Bio-chemico-physical differences

Indicators

Less fertilizer in organic fields Less biomassLess canopy cover

Field canopy cover Biomass estimation

Lower nitrogen content of the plants

Leaves chlorophyll content

Smaller plants Plant height

Less spatial homogeneous fertilizer application and less efficient pesticide in organic fields

Higher spatial heterogeneity

Standard deviation of other indicators by field

4. HypothesisIn particular :

4. Hypothesis

Cotton Yield in Burkina Faso:

•Organic = 675 kg/ha (std dev = 314 kg/ha)

•Conventional = 1 100 kg/ha (std dev = 391 kg/ha)(Centre for Development and Environment of the University of Berne (CDE), Pineau et al. 2009)

5. Method

5.Method

Study site

5.MethodStudy site

5.Method

Study site

• Several local varieties for organic and conventional

• Several varieties Bt GMO • Low intensive cultivation • Farming operations: manually or workanimals• Rainfed

5.Method

Cotton cropping method in Burkina Faso

Crop cover: hemispherical pictures: 10/field

CAN-EYE software was used to derive 2 indexes from the hemispherical pictures:•A Plant Area Index (PAI)•A Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) index

5.Method

Field measurements

+ GPS

Chlorophyll content: CCM200: 10/field

Height:

meter: 10/field

5.Method

Field measurements

Field spatial heterogeneity = standard deviation of parameters by field

5.Method

Field measurements

• SPOT 5 (via ISIS program / CNES)

• 2.5 m color• 3 BANDS: Green, Red, NIR• Tasking window between

16/08/2011 – 25/10/2011• Nearly permanent cloud cover• 1 image on 15/11/2011

only !!! Very late !• + 1 MODIS image: surface

temperature emissivity (Sept-Oct)

5.Method

Satellite image

Spectral indicators• Simple bands: B1, B2, B3• 2 bands combination (simple ratio:) B1/B2, B1/B3, B2/B3

Spatial heterogeneity indicators• Standard deviation of pixels by field for B1, B2, B3• Coefficient of variation of pixels by field for B1, B2, B3• …

5.Method

Satellite indicators

• Comparative approach• Comparison of indicators values of organic and

conventional (not a threshold indicator value method)

• Selection of most discriminating indicators among those systematically computed

• Univariate and multivariate (Linear Discriminant Analysis)

• Indicator discrimination power = p.value of the Mann–Whitney–Wilcoxon (MWW) test

5.Method

Statistical method

6. Results

6. Results

Field measurements

6. ResultsField measurements

6. ResultsField measurements

Red/NIR

6. ResultsSatellite indicators: univariate - spectral

6. ResultsSatellite indicators: univariate – spatial heterogeneity

Std. dev. Green

6. ResultsSatellite indicators: multivariate, Linear Discriminant Analysis

6. ResultsSatellite indicators: multivariate, Linear Discriminant Analysis

6. ResultsSatellite indicators: multivariate, Linear Discriminant Analysis

6. ResultsSatellite indicators: multivariate, Linear Discriminant Analysis

7. Discussion and conclusion

Differences are observed between cotton types•For both field and satellite indicators•Statistically significant•Not enough pronounced with values ranges that largely overlapThis prevents the use of these indicators alone to be the base of a robust discrimination

But the method enables to target for priority field control, organic fields who present indicator values getting closer to the one of conventional or GM cotton fields

Further research:•Timely satellite acquisition!•Identification of the ideal phenological stage for cotton monitoring

7. Discussion and conclusionGeneral conclusion

Regarding the initial hypothesis

Mixed results regarding the initial hypothesis:

•Most of the indicators: organic fields present significant lower general field development and higher spatial heterogeneity

•CCI indicators don’t show any significant difference between management types and the standard deviation of the canopy cover show a slightly lower spatial heterogeneity for organic fields

7. Discussion and conclusion

Satellite Indicators are questionable:•A single image was acquired very late in the crop cycle •No straight conclusion regarding the general relevance of the use of RS techniques in the study context

Use of satellite images seems to be quite compromised given the unfavourable atmospheric conditions which are most of the time cloudy. Need for daily image acquisition for cloud free image?

Trees in cotton fields can strongly influence the reflectance and the spatial heterogeneity (from no tree to a complete agroforestry system)

7. Discussion and conclusion

Relevance of the use of satellite images in this context

Difference between cotton parcel is also due to other factors, difficult to take into account:•The phenology stages that can strongly vary from one parcel, farmer or region to another due to varying seeding date, itself depending among other on the local climatic condition, with very localized rainfalls. •Varying soil natural fertility•Varying level of development of the farmers (fertilizer availability)

7. Discussion and conclusion

Remaining obstacles

• Given• Lack of experience of organic cotton farmers • Yields already achieved by the organic farmers

“elite” which are close to the conventional ones• If organic farming techniques are encouraged and

tuned (increase of quantity of available organic fertilizers) , the current gap between organic and non organic cotton yields would be considerably reduced

7. Discussion and conclusion

Skills of organic farmers

Thank You !

Université de Liège - Belgique

Acknowledgement

The French “Centre National d’Etudes Spatiales” (CNES) through its « ISIS » program (« Incitation à l'utilisation Scientifique des Images SPOT ») that enabled to acquire a SPOT 5 image at low cost for this study. SPOT "© CNES (2012), distribution Spot Image S.A.", Http://www.isis-cnes.fr/IntroPage.do

The “SOciété Burkinabé des FIbres TEXtiles” (SOFITEX) that allowed the field survey in conventional and GMO cotton fields.

The National Union of Cotton Producers of Burkina Faso (UNPCB – Union Nationale des Producteurs de Coton du Burkina Faso) that enabled the field survey in organic cotton fields and accompanied the entire field survey.

Helvetas Swiss Intercooperation Burkina Faso, for their important documentation on organic cotton production in Burkina Faso and their advices for the field survey preparation.

Contact information

Arlon Campus Environnement (ACE)University of Liège (ULg)185, Avenue de Longwy,6700 ArlonBelgium

Antoine DENIS Bernard TYCHON

TEL 0032 63 230 997 0032 63 230 829

Email [email protected] [email protected]

Website http://www.campusarlon.ulg.ac.be/