First Expert Meeting for Development of Decision

Making tools in Central Asia

CropWatch and DroughtWatch

WU Bingfang, Hongwei Zeng, Nana Yan, Sheng Chang

Institute of Remote Sensing and Digital Earth (RADI), CAS

Outline

◼ Introduction

◼ CropWatch Cloud

◼ DroughtWatch

◼ DroughtWatch for Mongolia

◼ DroughtWatch for GYG

◼ Recommendations

Food security

3

◆ By 2050 the world’s population

will reach 9.8 billion, 29 percent

higher than today.

◆ Nearly all of this population

increase will occur in developing

countries.

More mouth to feed

Nominal wheat price in US $/metric Ton

2010/11 Price hikes

Drought: Russia USA

Landsat 1 Launched

(1972)

1971/2’s price hike

2008 Price hikesDroughts:

Australia & Ukraine

Becker-Reshef et al.

Food Price Volatility

We have to producing 70 percent more

food for an additional 2.2 billion people

by 2050(FAO. How to Feed the World in

2050)

Goal 2: Zero Hunger

◼ Pledges to end hunger, achieve food security, improve nutrition and promote sustainable agriculture

◼ require an integrated approach

➢ Sustainable Food Production and

➢ Resilient Agricultural Practices

➢ Ensure Stable Food Commodity Markets and

➢ Timely Access to Information

Crop Monitoring is essential

◼ Early production forecasts help policy makers to make

evidence-based trade decisions

◼ Early warning information helps early response and actions

on providing food aid to food shortage regions

◼ In season warning (stress due to drought, pest & diseases)

for better farm management

Nominal wheat price in US $/metric Ton

2010/11 Price hikes

Drought: Russia USA

Landsat 1 Launched

(1972)

1971/2’s price hike

2008 Price hikesDroughts:

Australia & Ukraine

Becker-Reshef et al.

Monthly Wheat Prices 1960-2011 ($/Metric Ton)Source: World Bank

Crop Monitoring Systems

Many countries use satellite data to improve food

information availability and transparency

Gaps and challenges

⚫ Ownerships

– Only a few countries or int’l organizations have the capacity to do crop

monitoring

⚫ Lack of transparency

– System are physical or technically difficult to access openly

– Methodology is not well documented and

– Difficult to participate in agriculture monitoring

⚫ Need to enhance

– No automatic processing, manual works mainly

– Crop condition is main output, lack of accurate production

– Lack of forecasting at early stage or even pre-sowing

– Language

Issues for developing countries

◼ The paucity of adequate capacity in obtain and accessing up-

to-date staple crop production information, which is essential

for a country economic governance and securing food supply.

◼ Big financial input and operational cost as well as adequate

technical skills constrain developing countries to set-up,

operate, and maintain such crop monitoring facilities.

◼ Over-dependence on information provided by third parties

and often poses the danger of taking decisions based on

delayed and on not easily verifiable information.

Outline

◼ Introduction

◼ CropWatch Cloud

◼ DroughtWatch

◼ DroughtWatch for Mongolia

◼ DroughtWatch for GYG

◼ Recommendations

CropWatch Cloud at Alibaba

CropWatch-Pro

• An online tool for people to produce crop monitoring products at any time and anywhere.

CropWatch-Explore

• An online interface for people to explore and analysis all the crop information data easily.

CropWatch-Project

• An online platform for people to create and write the crop bulletin.

CropWatch-Bulletin

• An webpage for people to read CropWatch bulletin.

CropWatch Explorer

cloud.cropwatch.com.cn

Data converging and preprocessing

USGS

GSOD

ECWMF

ESA

Other

Auto download

DatabaseProcessing

Resample

Composite

Departure

Batch

http

ftp

python

Data processingStandard algorithm

•Temperature

•Rainfall

•PAR

•Biomass

•Sunshine hour

•Aerosol

•PET

•Other global products

Advance algorithm

•Agro-climatic risk

•CALF

•VCIx

•VHImin

•Cropping intensity

•From other sources

Interactive component

•Clustering

•Crop condition

•Crop maaping

•Crop yield

•……

Models

Tables

Figures

Map servicesIDL

python

arcgis

Automatic

chain

DatabaseData storage

on cloudData from cloud

CustomizeAdvanced

products

Chart

Image

Table

Map

User

Auto converging

Auto preprocessing

Component 1: CropWatch Processing

CropWatch Processing offers an auto-processing chain from pre-processing of raw data to production outlook

Pre-processing

• Data conversion

• Projection and transformation

• Vector to Raster

• Merge

• Clip

• Resample

• Band-combination

• Spectral-merge

• Temporal and spatial merge

Agro-meteorology

• PAR

• Temperature

• Rainfall

• Biomass

Agronomic indicator

• CALF

• CI

• VCIx

• VHI

Crop condition

• NDVI development

• Real Time development

• Cluster

Production Forecast

• Yield

• Area

• Production

Component 2: CropWatch Explore

CropWatch-Explore provide a web service for users to conveniently explore and visualize our data.

CropWatch-Explore

Visual Type

Vec

tor

Ras

ter

Clu

ster

Scale Type

MPZ

MR

U

Co

un

try

Sub

-Co

un

try

Crop Type

Wh

eat

Mai

ze

Ric

e

Soyb

ean

RAIN TEMP PAR BIOMASS

NDVI VCIx VHI CALF CI

Area Yield ProductionEarly

warningPrice

Component 3: CropWatch Analysis

CropWatch Analysis is cloud based participatory tool for the CropWatch teams or invitedpeople from over the world analyzing their CropWatch indicators anywhere. It providescreate document, allocate and manage tasks, monitor schedule and publish thedocument online functions which let people over the world finish their documentstogether on the cloud platform.

CropWatch Team Experts across the world

Component 4: CropWatch Bulletin

Provide global crop report as pdf or html format

Food security early warning

⚫ Cropped arable land fraction (CALF) represents the total

cropping proportion at early growing stage

⚫ Agro-meteorological risk index (AMRI) considering

meteorological suitability for crops at different growing

stage is used for yield alarming

Agro-meteorological

risk index

August 2013July to October 2015

Early outlook

Server drought in South Africa

⚫ Large production drop alert given in November 2015 Bulletin

⚫ January 2016 Bulletin: Maize production was projected at 44.6%

drop: Server drought prevented farms sowing maize, with a

reduction of 34% of maize area; yield was 16% lower than 2015

⚫ April 2016 Bulletin: Maize production is revised to 32% drop, since

Feb 2016, rainfall benefited the maize in fields.

Development of NDVI profiles over maize growing

areas in 2014-15 and 2015-16

Relative distribution of maize in 2014-15 and 2015-16

Features

◼ Analysis-ready products

➢ 26 Indicators ready in CropWatch Cloud considering most indicators

used in existing system

➢ Indicators customizable, easy to include new national or regional

specific indicators

◼ Cloud computing improves efficiency of data processing

◼ Joint work promotes confident and transparency

◼ Major contributor to

➢ GEO/GEOSS Global Agricultural Monitoring (GeoGLAM) flagship

➢ Agricultural Market Information System (AMIS)

Potential

◼ Promote ownership for developing countries

➢ Customized according to the specific demand for each country and

work as a national/regional system

➢ After customization and training, countries will strengthen the

agricultural monitoring capacity on your own

➢ Promote developing countries leap-frag development

◼ Cloud services for crop monitoring

➢ Cloud based system assessible from internet everywhere without

investment on computing infrastructure, storage, etc

CropWatch for Mozambique

◼ Customize CropWatch for Mozambique

◼ Portuguese Interface

◼ Including all provinces and districts

◼ Crop phenology

◼ Portuguese version of GVG tools

Technical training

◼ First round training for selected experts from national and

provincial offices (3 colleagues)

◼ Extended training for 29 participants including local

experts from 8 provincial offices attend the training

◼ In-situ data collection training at different major

agricultural provinces

CropWatch for Mozambique

Monitoring units: every districts and provinces

2019 March Bulletin after Cyclone IDAI

CropWatch for Thailand

Based on the extensibility of cloud-based system, CropWatch and Agr-Map

jointly develop API for exchange information.

monitoring

ValidationAgronomic

Agrometeorological

⚫ Handover Zambia cropland products in 20 June 2019

Cropland product for Zambia

Outline

◼ Introduction

◼ CropWatch Cloud

◼ DroughtWatch

◼ DroughtWatch for Mongolia

◼ DroughtWatch for GYG

◼ Recommendations

Disasters to food production

◼ Flooding, storms and droughts are top three disasters

resulting in yield loss and unstable production

Numbers of disasters per type,

1998-2017

Drought characteristics

◼ It builds over a period of time (may be even a year or two)

with increased scarcity of water.

◼ It does not have a well-defined start. It is a creeping

phenomenon.

◼ Drought may be localized covering a district or a group of

districts, and even widespread covering a few provinces

or several countries.

◼ Drought intensity, duration and frequency may be different

in a district or a piece of land.

Drought Definition

More than 150 published definitions of drought in the academic

literature were found.

Meteorological drought: a prolonged period of

below average precipitation

Agricultural drought: there is not enough

moisture to support average crop growing

Hydrological drought: water reserves in

aquifers, lakes and reservoirs fall below an

established statistical average

DroughtWatch◼ Established for

Ministry of water resources and

Ministry of civil affair: Center of disaster mitigation, Ministry of Emergency Managt.

◼ Meteorological Drought (5 indices) (1996-)

– Rainfall Anomaly Index (RAI)

– Annual Rainfall Anomaly Index (ARAI)

– Deciles (DECILE)

– Standardized Precipitation Index (SPI)

– Palmer Drought Severity Index (PDSI)

◼ Agriculture Drought (DroughtWatch）(4 indices) (1998-)

– Vegetation Condition Index (VCI)

– Temperature Condition Index(TCI)

– Vegetation Health Index(VHI)

– Normalized Difference Water Index( NDWI)

◼ Hydorlogical Drought (2 indices) (2005-)

– Soil Moisture Index(SMI) SMI=SM/FC

– Soil Moisture Anomaly Percentage Index (SMAPI) SMAPI=SM/SMavg

Evolution of DroughtWatch

Version Major Revision Improvement Time

V1.1 Several calculation modules built by program Try to finish the part of drought

monitoring by the computer1998

V1.2The drought monitoring system based on

AVHRR, named as DroughtWatch for ChinaDrought monitoring can be calculated

automatically 2006

V1.3 Replacement of AVHRR with MODISMODIS is beyond AVHRR specially in image quality and geometry location

accuracy2008

V2.1 automatic operation system was emerged The system can be automatically run 2009-2010

V2.2Update the basedata(cropland, maxmin

data)Improving the accuracy and stability 2012

V3.1 Extend to the other countries Developing the system applicability 2013-2014

V4.1Interactive drought monitoring system for

globeInteraction functions and information

demonstration were involved2015-2017

V4.2 Drought forecasting in short termsImproving drought forecasting

functions2018-2019

Pre-processing Drought IndexDrought

monitoring Statistics

Upgrading: AVHRR to MODIS; MODIS to MERSI

DroughtWatch1.1

DroughtWatch4.1

Extension

Global Country/region Field

Data:

TRMM/GPM;

AMSR-E/

MWRI/SMAP;

MODIS/VIIRS

Methodology:

SPI/VHI

SM anomaly

NDVI anomaly

Temporal and

spatial resolution:

Daily-Monthly

1km-25km

Data:

MODIS

FY-3/MERSI

VIIRS

Methodology:

VCI/TCI/NDWI

ESI/TANDVI

SM anomaly

Temporal and

spatial resolution:

Daily-Monthly

250m-1km

Data:

HJ-1A/B CCD

GF-1/2

Sentinel-1/2

Methodology:

NDWI

MPDI

ESI

Temporal and

spatial resolution:

variable

30m/10m

Outline

◼ Introduction

◼ DroughtWatch

◼ DroughtWatch for Mongolia

◼ CropWatch Cloud

◼ DroughtWatch for GYG

◼ Recommendations

❑ Developing drought monitoring methods for Mongolia

❑ Indicator selection

❑ Validation

❑ Localization

❑ Building up the spatial information database

❑ Enhancing capacity for Drought Monitoring in Mongolia

❑ On the job training and joint academic research

❑ Customizing and deploying the drought monitoring system

❑ Field campaign support and validation work

❑ Academic workshops

❑ Information services and technical support

Objectives and Contents

Data processing, building database, indices selection were

achieved jointl by China and Mongolia experts in RADI (2014-

2018)

Joint work on data processing

Joint field works

Parameters: Soil moisture, vegetation biomass, height, coverage, biodiversity, livestock loss number by drought and spectrum.

Participants: IRIMHE and RADI. 2014 to 2017 (July to August)

Joint Validation

❑ Drought products validation with field data from 2014-2017:

❑ Soil moisture

❑ Biomass

❑ Regional drought affected data from field observation❑ Annual validation report

Localization for local ecosystem

Forest steppe & steppe & desert

steppe

Localization for seasonal variation

⚫

ba

Weights May June July August September

Wtci (VHI a) 0.41 0.31 0.27 0.31 0.42

Wvci (VHI b) 0.59 0.69 0.73 0.69 0.58

Regional workshop on understanding

the operational aspects of the

drought observation system in

4 July 2018, 14:00

to 15:00

43

System Customization

17 Sept 2018

Monitoring Results

Summer condition assessed by observers at

Meteorological stations

Summer condition /2nd decade, June 2018/Remote sensing drought map /2nd decade, June 2018/

/3rd decade, June 2018//3rd decade, June 2018/

Products dissemination to users

http://irimhe.namem.gov.mn

www.icc.mn

www.eic.mn

Drought product

dissemination to local

meteorological departments

by internal networkServicing to organizations

Ownership

DroughtWatch system have been deployed in NRSC of Mongolia in 2014, and fully operated by NRSC staff on monitoring, field work, and analysis.

Now, DroughtWatch products and results are useful for

planning, decision making at crop farming, forest andpastoral animal husbandry sector in Mongolia.

Technical advisory and support

Technical Training

On the job training

Joint work from 2014 to 2017.

Customization

Localization

Ph.D fellowships

Full Technical Transfer

Experiences and Lessons

ESCAP coordination, Mechanism of ownership and

full technical transfer are essential to the success

ESCAP and CAS support are guarantee to the

commitment

A good partnership between RADI and IRIMHE

Stakeholder engagement

– Need to give more training or advertisement to other

users about the drought products

– Make stakeholder use of products

Extending to other fire, dzud, and crop

Satellite based Dzud risk model

Outline

◼ Introduction

◼ DroughtWatch

◼ DroughtWatch for Mongolia

◼ DroughtWatch for GYG

◼ Recommendations

DroughtWatch for Kyrgyzstan

Case of DroughtWatch-KGZ

Outline

◼ Introduction

◼ CropWatch Cloud

◼ DroughtWatch

◼ DroughtWatch for Mongolia

◼ DroughtWatch for GYG

◼ Recommendations

Recommendation

Stakeholders needs to engage at the earlier stage

Incorporating existing data, big data

Full technical transfer and capacity building

➢ On job training, Ministry of emergency and Ministry of Agriculture

➢ Localization of models

Validation to build up confidence

Drought forecast and impact assessment,

➢ Loss of crop

➢ Water use