Optimizing agriculture for sustainability and productivity by ICT - - PowerPoint PPT Presentation

1

Optimizing agriculture for sustainability and productivity by ICT

Seishi Ninomiya Institute for Sustainable Agro-ecosystem Services, The University of Tokyo

Agriculture and world population

6 5 4 3 2 1

10 10 10 10 10 10

7

10

4

10

10

10 15000 5million 0.5billion 6.5billion Engineering Chemistry Agriculture

Tools (implements and fire)

Population

Years ago

Revised from Robert W.Kate(1994)

Grain productivity in last forty years

1961 2003

• Wheat

1.1 t/ha 2.9 t/ha (2.7 times)

• Rice

1.9 t/ha 4.0 t/ha (2.1 times)

• Corn

1.9 t/ha 4.7 t/ha (2.4 times)

• Population

3 billion 6.3 billion (2.1 times)

• Labor (hrs/ha)*

1,750 hrs 250hrs (1/7th)

FAO statistics * Case of Japan 1 ha = 2.5 acre

Agriculture based on chemistry and engineering along with high input = Maximization

Technologies to have increased crop productivity in 20th century

• Chemical Fertilizers

– Haber Process (1908)

• Agro-chemicals

– DDT (1938) Parathion (1944), Organic mercury, 2-4D (1944)

• Machineries

– Steam Locomotive Tractor (1902), Tractor with crawler

• Irrigation

– Pumping, dams, channels

• Plant Breeding

– Mendelian Low (1865)

Drawbacks of agriculture in 20th century

• Serious impacts on environment

– Agricultural chemicals – Water pollution, damage on ecosystem – Exhausted and unhealthy soil

• Agriculture based on high energy consumption

– Machinery, chemicals

• Food safety and reliability

Non-sustainable agriculture based on chemistry and engineering

Agriculture in 21st century need to fulfill

• High productivity

– To fulfill demand increase – Limited arable land, desertification, limit to deforestation

• Stable production under unstable and varying climate

– Global warming, floods, drought, unusual emergence of pests,..

• Sustainability

– Lower impacts on environment, energy consumption, CO2 output

• High quality and high functionality

– High nutrition, good taste

• Safety and reliability
• Welfare of farmers

Paradigm shift from maximization to optimization is needed

Optimization? e.g. Reduction of pesticide application

• Results in

– Cost reduction

• Material cost, labor cost

– Lower impact on environment – Lower CO2 output – Food safety and reliability

• To reduce pesticide

– Timely and pinpoint protection (application)

• For timely and pinpoint protection

– Prediction of pest occurrence – Optimal crop management

ICT can help in many aspects

ICTs for reduction of pesticide application

• Pesticide prediction model (early warning system)

– Weather data (observed and forecasted) – To monitor field and crop condition (e.g. trap data to know trend)

• Navigation to right use of pesticide

– To follow complicated regulation in order not to violate it

• Farm recording of pesticide application

– To know cost (materials and labor) – To certify the correct use (GAP) and traceability information

• Estimation of contribution for CO2 reduction

– Data for farm level LCA

9

ICT helps optimization in many aspects

• Cost reduction and competitive agriculture

– Optimal farm planning, efficient management of large number of fields – Efficient distribution

• Robust and stable farm production under extreme weather

and global warming

– Optimal crop / variety recommendation, optimal cropping timing – Early warning system of extreme weather

• Sustainable agriculture

– Optimal agro-chemical application

• Food safety and reliability

– Tractability, right use of pesticide – GAP risk management

• High quality products

– Visualization of quality

Approaches to reach the goal

• Data collection

– To know what is happening in each field quantitatively

• Efficient Knowledge transfer

– Quantify invisible empirical knowledge – To transfer Tacit Knowledge to Explicit Knowledge – Case base reasoning

• Optimization and risk management

– To support decision making based on acquired data and knowledge

• Framework to support decision making

11

Data collection and recording

• To know present status of fields and crops

– Site-specific optimization is needed based on site-pacific data because of site-specificity of agriculture (no generalization) – Long term data collection is necessary

• To know present status of farm management

– Many farmers do not know income and expenditure balance of each parcel basis

• Basis for risk management

– GAP

• Visualization of technology of each farmer

– To show the level of skill a farmers has by quantitatively comparing the present level with a target level – e.g. nutrition content level, soil organic content, energy consumption

Key points:

• long term and continuous collection, low cost
• minimization of manual handling, easy-to-use interface

Multi-sensor data collection

Fieldserver

• Air temp., humidity, solar radiation,
• soil moisture, CO2, etc.
• Camera (0.3 to 10 M pixels)
• WIFI hot spots

Cell phone with GPS and camera

Automatic detection of farm action by image analysis and IC tags

IC Tag Subject material Automatic record

• f farm action

In-laboratory analysis Data analysis and archive

Residual pesticide test Micro array micro-

• rganism analysis

Spectrum analysis Thermograph Heavy metal analysis Simplified elementary analysis Infra red sensor Laser induced florescent analysis Florescent X ray Leaf color Color distribution Digital pen record

On-site evaluation and analysis

Collected data Analysis results Analysis results Collected data

Evaluation Comparison Technical support

Fixed point field monitoring

Air temp., soil temp., solar radiation.,. soil moisture, humidity, image etc. Fieldserver

Patrol wagon

Periodical screening and diagnosis of field and crops Quantification of farmer’s skill by achievement level to target goal Farmers can know the gap between their level and ideal level Guidance for improvement

Field Doctor: Integrated field monitoring and diagnosis service

Efficient knowledge transfer

• Knowledge of skillful farmers is disappearing

along with aging of them

• Empirical knowledge takes an important role in

agriculture

– Quantify invisible empirical knowledge – To convert Tacit Knowledge to Explicit Knowledge

• Technologies

– Case base reasoning (CBR) to utilize cases – Text-mining to extract knowledge from text – Automatic detection of farmers’ actions

cyfars@yahoo. co.jp 南大成 選別 収穫 終了 トヨシロ 09120001.jpg 馬鈴薯 42

Cyfar’s (Cyber farmer) diary

• Mobile phone based blog system with photos
• To share farm information among neighboring farmers
• 10 years of data collection is now working as a

valuable case database to make decisions

17

Optimization and risk management

• Risk management and optimization by maximally

utilizing collected data, knowledge and models

• Simple data mining is the first step
• Risk management for human mistakes and

farming optimization

– GAP – Farm management system

• Optimal management against environmental risks

– Extreme weather – Pesticide

• Fundamental databases are extremely important

– Weather DB, soil DB, farming system DB, market price DB, map DB, etc.

Fieldserver Fieldserver

Images Temperature Humidity etc… Yield Farm work records Growth rate etc…

Farmer Farmer

Simple data mining to find out rules

e.g. High relationship between yield and air temperatures of 4 to 7 days before harvest

Heuristic findings by comparison using data viewer

19

Identification of best timing of harvest

ログイン メニュー 事前判定 予定を入力 判定結果 履歴登録へ 計画からの入力 計画を参照 予定を入力 事前判定

GPS

２９人の農家の方で、５０歳未満の方は 全員今後も携帯を利用したいという回答

• Adjudication of proper use of pesticide by mobile phone.
• Result of adjudication is automatically recorded as farm record

携帯電話による事前判定と履歴記帳

Pesticide navigation system: To support proper use of pesticide

Farm management system for GAP

Farming system database Farming record Field data collection Pesticide DB Pesticide navigation GAP Rule DB Fertilizer DB Market Price DB

• To navigate farmers to most optimal farming based
• n GAP standard linking several databases

Immigration Route

4 mm 3 mg Rice Hopper

Airborne pest immigration prediction

• Weather forecast + diffusion model + insect behavior model

+ crop growth model + satellite image analysis

• Optimization of pesticide application

Utilization of satellite images / remote sensing

• To identify the best timing of wheat harvest

– Water content estimation of wheat grain to keep the grain quality best

• Rice grain quality estimation

– Estimation of nitrogen contents per field – For quality classification and guidance for next cropping

• Rice paddy damage estimation for agricultural

insurance

– Substitution of complete enumeration sampling by humans Examples practically used in Japan

24

Framework to support decision making

• Data integration is necessary in many of

agricultural decision making

• To provide efficient data and program usage, a

framework to seamlessly integrate and exchange data is necessary

MetBroker is now covers over 22,000 stations

• It covers 22,000 weather stations of 25 DBs

Time series integration of weather data

Observed Short term prediction Normal year value Real time prediction Yield prediction Harvest plan Pest prediction Protection plan Fertilizer application plan Labor plan Shipping plan Normal year prediction Optimal crop Prediction of potential Growth period Future Prediction Impact assessment Cropping map under global warming Long term prediction Normal year Today

Comparison of rice growth under several conditions: a glocal (global + local) approach

• Comparisons among different cultivars and locations
• To be used by farmers as well as policy makers

Cultivar Planting date Temperature assumption Heading and maturing date Prediction of potential yield

Conclusions

• ICT helps the shift from maximization to optimization in

agriculture

• ICT has to help continuous data collection which is

absolutely inevitable in agriculture

• Utilization and transfer of empirical knowledge by ICT
• Decision support systems are only useful with fully

collected data collection

• Package of technologies as a service should be

provided for farmers

• A framework to integrate data and application to create

a total service is needed

• To hide ICT from farmers

29

http://www.agmodel.net/DataModel/ http://model.job.affrc.go.jp/FieldServer/default.htm 二宮正士 snino@isas.a.u-tokyo.ac.jp

Thank you very much