1. Results from Lao People’s Democratic Republic, Philippines, Thailand, Viet Nam Earth Observation Technologies for Crop Monitoring: A Workshop to Promote Collabrations among JECAM/Asia-RiCE 2018
Lakshman Nagraj Rao
Statistician
Economic Research and Regional Cooperation Department
Asian Development Bank

2. INTRODUCTION
Rice is a staple food for the majority of population in Asia and the Pacific.
Importance of timely and reliable crop production statistics
Relevance to mitigating food security issues (SDGs)
Planning government interventions in the agriculture sector
Surveys, censuses are expensive to conduct and take longer to process. Administrative data are prone to measurement error.
Alternative or supplemental data support system would be through remote sensing or the use of satellite data.

3. INTRODUCTION (2) With satellite data, But….
Rice area may be estimated several times a year instead of only during the survey period. Also open doors for forecasting.
Land use can be studied better by comparing area maps taken in different times.
But….
New technology may be difficult to adapt
Statisticians have to be trained on the software
Costs may be higher
Small and varied rice plots (& cropping patterns)
Mind set of policy makers and government planners has to be influenced to accept the new methodology
This is where ADB comes in!

4. Innovative Data Collection Methods for Agricultural and Rural Statistics
Source of Funds: Japan Fund for Poverty Reduction
Technical Advisor: Japan Aerospace Exploration Agency
Pilot Countries: Savannakhet (Lao PDR), Nueva Ecija (Philippines), Ang Thong (Thailand), and Thai Binh (Viet Nam)
Implementation Period: June 2013 to October 2017
Objectives:
Development of customized software applications and methodology to estimate paddy rice cultivation area and crop production using satellite data
Training of counterpart staff in the four pilot countries, and
Development of an online training program on the use of satellite data for agricultural and rural statistics.

5. Innovative Data Collection Methods for Agricultural and Rural Statistics
Implementing Partners
Center for Agricultural Statistics, Department of Planning and Cooperation, Ministry of Agriculture and Forestry, Lao People’s Democratic Republic
Philippine Statistics Authority, Philippines
Office of Agricultural Economics, Ministry of Agriculture and Cooperatives, Thailand
Geo-Informatics and Space Technology Development Agency, Thailand
Center for Informatics and Statistics, Ministry of Agriculture and Rural Development, Viet Nam

6. Innovative Data Collection Methods for Agricultural and Rural Statistics
JAXA software application (INAHOR) used to transform satellite images into rice area maps, but modified and customized according to the needs of the country. Transformation undertaken by Remote Sensing Technology Center of Japan (RESTEC).
INternational Asian Harvest mOnitoring system for Rice
(and “INAHO” also means “rice year” in Japanese)
The main functions :
Providing a rice planted area map (including the growing stages classification)
Providing a rice planted area and production
Input satellite data : Time-series SAR data (ALOS-2 PALSAR-2 – 100m resolution, 350 km swath)

7. Stratified three-stage sample design
Stratified random sample of 120 meshes (200m x 200m) selected for the pilot survey in each province
Random sample of reserve meshes for possible replacement for each province
Sample allocation concentrates sample in strata with more rice based on data collected during growing stage.
First stage
Selection of sample meshes within each stratum
Second stage
Random selection of 4 sample plots with rice within each sample mesh
Third stage
Random selection of one sub-plot within each sample rice plot for crop cutting
Rainy season of 2015

8. FIELD SURVEYS (CROP CUTTING AND FARMER RECALL)
Selection of 120 meshes
Random selection of "plots" in the 200m x 200m meshes
Tracking of plot boundaries
Random selection of sub-plot
Harvesting
Moisture reading and weighing of samples
Cleaning of samples
Threshing
Drying
Weighing and moisture reading of dried samples

9. 30% Rice 100% Rice About 120 points were surveyed 200m 200m
200m x 200m x 30% = 1200m2
200m x 200m x 100% = 40000m2
Sum of all surveyed mesh rice area = 1,842,640 m2 or hectares
Calculated result for each mesh with INAHOR
= 1,777,800 m2 or hectares
The estimated % = INAHOR / Survey = 96%
Meaning the estimated result is 4% under estimation

10. Comparison of Paddy Area Estimates (in ‘000 ha.)
Stratum
Country/Province
Savannakhet
LAO PDR
Nueva Ecija
PHILIPPINES
Ang Thong
THAILAND
Thai Binh
VIET NAM
Official statistics
181.3
172.8
47.7
81.3
INAHOR-AD
258.0
221.9
40.9
78.6
Unmodified Tracks
270.0
207.5
29.2
47.4
Digitized Tracks
249.7
205.5
29.9
48.5
Re-digitized Rice Areas Inside the Mesh (ALIS Method)
260.7
202.8
39.8
73.3

11. R-CDTA 8369
RESEARCH WORK IMPLEMENTED (1)
Land Measurement Bias: Comparisons from Global Positioning System, Self-Reports, and Satellite Data
ASSOCIATED BLOG
Land measurement bias revisited: Using Google Earth in agricultural land measurements as an alternative technique produces land area estimates close to Global Positioning System (GPS) estimates at reduced fieldwork costs.
In addition to the three (3) main outputs, there will be four (4) economic working papers and one (1) Handbook on the Use of Remote Sensing for Paddy Area and Production Estimation. Two EWPs are now finalized and will be published within the year. One EWP has just been presented to ADB staff and will be finalized soon. While one EWP (e.g. country report???) and the handbook have been drafted and will be available by first quarter 2018.

16. LAND SIZE – PRODUCTIVITY RELATIONSHIP

17. R-CDTA 8369
RESEARCH WORK IMPLEMENTED (2)
Measuring Rice Yield from Space: The Case of Thai Binh Province, Viet Nam
ASSOCIATED BLOG
Measuring rice yield from the sky: An innovative data fusion technology, which combines Landsat and MODIS data, thereby increasing spatial and temporal resolution is used to map paddy area and estimate yield in Thai Binh Province, Viet Nam.
In addition to the three (3) main outputs, there will be four (4) economic working papers and one (1) Handbook on the Use of Remote Sensing for Paddy Area and Production Estimation. Two EWPs are now finalized and will be published within the year. One EWP has just been presented to ADB staff and will be finalized soon. While one EWP (e.g. country report???) and the handbook have been drafted and will be available by first quarter 2018.

19. Crop Yield Map for Thai Binh
Check analysis

21. Innovative Data Collection Methods for Agricultural and Rural Statistics
RESEARCH WORK IMPLEMENTED (3)
Technological Innovation for Agricultural Statistics: Special Supplement to Key Indicators for Asia and the Pacific 2018
ASSOCIATED BLOG
Technological Innovation is a game-changer for Agricultural Statistics: Both emerging and existing technologies should complement each other to bolster agricultural data quality.
In addition to the three (3) main outputs, there will be four (4) economic working papers and one (1) Handbook on the Use of Remote Sensing for Paddy Area and Production Estimation. Two EWPs are now finalized and will be published within the year. One EWP has just been presented to ADB staff and will be finalized soon. While one EWP (e.g. country report???) and the handbook have been drafted and will be available by first quarter 2018.

23. Innovative Data Collection Methods for Agricultural and Rural Statistics
Developed an online training con Estimating Rice Paddy Extent and Production with ALOS-2/PALSAR-2 and INAHOR-AD
Promotional video for the course:
Link for the course:
Webpage for ADB’s Agricultural Statistics activities and training materials:
INAHOR-AD has three major functions: (1) mapping of rice planted area, (2) calculation of rice planted area, and (3) calculation of rice production. The software application uses time series Synthetic Aperture Radar (SAR ) data in planting stage and well-growing stage and yield information.
Between 14 to 25 ministry staff were trained and participated in the different training/workshop conducted under the TA.
Important points from LNR:
Can we be specific on SNAP QGIS and other software used during the final training? We must ask Chrys/Tin because I was only intermittently in and out.
We may also want to break them down since ii has a lot of information. Suggestion:
(II) CAPI using Survey Solutions
(iii) Use of QGIS and SNAP using Sentinel 1 and 2 data
(Iv) Crop Cutting and Farmer Recall Survey
I think we can drop basic remote sensing since it is covered in point 2.

24. R-CDTA 8369
UPCOMING RESEARCH
Area Sampling Frame for Paddy Rice Statistics (November 2018)
Handbook on Estimating Rice Paddy Extent and Production with ALOS-2/PALSAR-2 and INAHOR-AD (November 2018)
In addition to the three (3) main outputs, there will be four (4) economic working papers and one (1) Handbook on the Use of Remote Sensing for Paddy Area and Production Estimation. Two EWPs are now finalized and will be published within the year. One EWP has just been presented to ADB staff and will be finalized soon. While one EWP (e.g. country report???) and the handbook have been drafted and will be available by first quarter 2018.

25. LESSONS LEARNED
Provision of satellite data with minimal fee (if not for free) is critical. JAXA has very kindly entered into partnership agreements to facilitate this work.
Possibility of using cloud based platform for data access and analysis.
Consider models for yield monitoring and not just area.
Mixed/intercropping and/or crops beyond rice.
Integrating surveying techniques into field mapping activities through a common map.

26. LESSONS LEARNED
Policymakers are open to new techniques, but pushing this agenda still requires collaborations with IOs such as ADB to facilitate strategic partnerships.
More systematic studies comparing administrative, survey, and remote sensing based estimates critical to the mainstreaming of such methods. This involves statisticians and remote sensing experts to work more closely together.
How to get down to the plot/parcel level for heterogeneous smallholder plots?

27. FUTURE PLANS Project: Data For Development (being processed)
Preparation of digitized field maps

28. THANK YOU!

29. PRESENTATION OUTLINE Introduction
TA 8369: Innovative Data Collection Methods for Agricultural and Rural Statistics
Sampling Strategy and Field Surveys
Comparison of Paddy Rice Area Estimates from field methods, administrative data, and satellite data
Land Measurement Bias
Rice Yields from Space
Technological Innovations for Agricultural Statistics
Lessons Learned and Future Directions