Transcript WP5 RSArea - E-Agri

Remote Sensing of Crop Acreage and
Crop Mapping in the E-Agri Project
Chen Zhongxin
Institute of Agricultural Resources and Regional Planning
Chinese Academy of Agricultural Sciences
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Outline
• I. The Objectives for WP5
• II. Main Tasks in WP5
• III. Research Plan and Activities
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
I. The Objectives for WP5
• Adapt and design in-situ segment sampling method
set up crop area extrapolation models for the study
areas (sampling and scaling-up)
• Select the optimal remote sensing classification
options for crop area in spectral and temporal terms
• Generate crop area estimates with in-situ sampling
and remote sensing
• Analyze errors (sampling and non-sampling) and
costs for crop area monitoring with remote sensing
• Demonstrate the selected technology in the study
areas
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
II. Main Tasks in WP5
• To adapt and design segment sampling method
WP51
• To establish the crop area spatial extrapolation model
for the study area
• To execute the segment sampling and track samplingWP53
in the study areas
• To collect the remote sensing data .
• To pre-process and classify the satellite images
• To select the best classification option in both
spectral and temporal terms
• To generate the area estimates using the ground
sampling dataset
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
WP52
WP54
II. Main Tasks in WP5
• To generate the area estimate using best
classification option
• To generate the area estimate combining
regression and remote sensing
• Analysis of sampling and non-sampling errors
• Analysis of mapping costs
• to evaluate what is the impact on the mapping
accuracy when no or very limited ground survey
(for example based on the track sampling) is
conducted.
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
WP52
WP54
WP55
WP56
Participating Institutions
•
•
•
•
•
VITO (WP51,52, 53, 54, 55,56)
CAAS (WP51,52)
AIFER (WP51, 52)
INRA (WP53, 54)
DRSRS (WP56)
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
III. Preliminary Research Plan
•
•
•
•
•
Data Preparation and Collection
In-situ sampling and extrapolation
Remote Sensing Classification of Crop
Error analysis
Generate crop acreage estimates from insitu and remote sensing data
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Data collection and preparation
• Background data
– GIS maps (land use, administrative, road, soil,
vegetation, contour, crop, geology,
geomorphology, hydrology)
– Socio- economic statistical data for 10 yr
– Crop calendar and phenology
– Climate data
• In-situ data: field segments and tracks
• Remote sensing imagery
– Time series of LR images
– HR images
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Data collection and preparation
• Remote sensing imagery
– Time series of LR images: MODIS, AVHRR,
AWiFS, VEGETATION,
– HR images: TM, ALOS, SPOT, IRS, HJ-1
– VHR images: QB, IKONOS, Aerial
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
平原
丘陵
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
山区平原
Crop Mapping for Winter Wheat in Anhui, 2009
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
In-situ data from field segments
• 50 samples @ 1km x 1km
• With 25 km intervals
• Winter wheat and maize
• Existing samples 500m x 500m
• Study region size 40000km2?
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Data preparation (cropland plots and
Agricultural Census)
Process of cropland plots
Construction of survey
unit
Design of spatial
sampling scheme
Two stage sampling
Simple
Randon
sampling
Based on
Agricultura
l Census
and landuse
data
PPS
Regular
grid as PSU
Samples selection
Field survey
Population inference
Technical flow of spatial sampling scheme
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
13
Samples Spatial distribution in Faku county
Samples Spatial distribution in Fengtai county
Samples Spatial distribution in Dehui County
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
14
Fig 4.2 Distribution of sample village
Fig 4.3 Distribution of sample plots in sample village
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
15
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
In-situ Segments
2008
2009
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
In-situ sampling and extrapolation
• Selection of sampling frame
• spatial vs. non-spatial
• Sampling methods:
– Random
– Systematic
– Stratification
• Remote sensing sampling
• Extrapolation (scaling-up)
– Relevant to sampling method
– Regression with remote sensed info
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Remote Sensing Classification of Crop
• Hard classification vs. soft classification
– Hard for HR images
– Soft for LR time-series data with sub-pixel
classification
• Automation vs. visual interpretation
• Supervised vs. unsupervised classification
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
The sub-pixel classification result
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
ALOS：10m,2009-3-20
QuickBird：0.61m,2009-3-25
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Error analysis
• Sampling error
• Non-sampling error
• Cost analysis
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Generate crop acreage estimates
• From in-situ segment and track sampling
– Get crop acreage estimate based on statistics
• HR remote sensing info
– Direct pixel count for full coverage
– Regression if sampled
• LR remote sensing
– Regression with HR or in-situ samples
– Sub-pixel classification
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Activities
•
•
•
•
•
•
•
•
•
Define the research regions (C, M, K)
Background data collection
Remote Sensing data collection/ processing
Field survey (2-3 times)
Sampling and extrapolation model
Remote Sensing classification
Error analysis
Generate crop estimate
WP5.6?
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Define the research regions (C, M, K)
• China – Huaibei, Anhui
• Moroco - ?
• ? Kenya?
• Time: asap (1 month? Before April 30)?
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Background data collection (research
regions)
– Socio- economic statistical data for 2001-10
– Climate data for 2001-10
– GIS maps (land use, administrative, road, soil,
vegetation, contour, crop, geology,
geomorphology, hydrology)
– Crop calendar and phenology
• Time: 6 months (before September 30)
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Remote Sensing data collection
– Time series of LR images: MODIS, AVHRR,
AWiFS, VEGETATION,
– HR images: TM, ALOS, SPOT, IRS, HJ-1
– VHR images: QB, IKONOS, Aerial
• Time:
– 3 months for first datasets
– progressively
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Remote Sensing Image Processing
– Geometric correction
– Radiometric correction
– Time series preparation
– Derived parameters (VIs, Ts, etc.)
– Phenology
• Time:
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Field surveys
• 2-3 times for winter wheat and maize
• 50 samples 1kmx1km (500m x 500m?)
• Track servey
• Time: April, August of 2011, 12 and 13 for
China
– For Moroco?
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
• Sampling and extrapolation model
•
•
•
•
Remote Sensing classification
Error analysis
Generate crop estimate
WP5.6?
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011
Thanks for
Your Attentions!
E-Agri Project Kick-off Meeting, Mol, 24-25, 2011