Vegetation monitoring with remote sensing€¦ · 2. 2. Intermediate relationships implicit to...
Transcript of Vegetation monitoring with remote sensing€¦ · 2. 2. Intermediate relationships implicit to...
Vegetation monitoring with
remote sensing
Dr Toby Waine, Lecturer in Applied Remote Sensing
Course Tutor MSc Geographical Information Management
Vegetation monitoring with
remote sensing, multi-scale
Scale
Regional crop inventory - Food security - Asset management - Governance (tax) - Policy
Precision Farming - Food security - Sustainable agriculture - Precision Irrigation - Environmental compliance
Magellium, DMCii Ltd.
Commodity traders UNODC, LUCAS 2012/15
Growers, ESA
Field level inventory - Food security - Asset management - Enforcement (Levy) - Environmental modelling
Contents
• What’s possible with remote sensing imagery?
• Imagery at different scales – spatial resolution and NDVI
• Case studies:
– Yield assessment
– Disease monitoring
– Crop area estimates
– Vegetation productivity indicator
What’s possible with Remote
Sensing imagery?
• Exploit an underlying physical relationship between a crop biophysical
parameter and spectral response, e.g. %cover, LAI, GAI, biomass
• Vegetation indices such as NDVI are often used to calibrate crop yield
indicators, to assess crop development, or disease
• Apply at different spatial scales (within-field to farm scale and above)
Different scales 250 to 1 m,
(satellite, aerial and ground)
Different scales <1 m
Near ground 1 m quadrat
0.1 m
0.25 m
0.6 m
0.1 m pixel
• Biomass • Pasture
• Forage
• Grain/fruit • Harvest index
• Harvest losses
• Roots/tubers • Harvest index
• Harvest losses
• Extract • Sugar
• Opium gum
Vegetation Index
Yiel
d/P
rod
uct
ion
Desired practical result
Yield/Production estimation
using vegetation indices
Vegetation Index
Cro
p C
ove
r
Photosysnthetically Active Vegetation PAV
1.
Biomass
Cro
p Y
ield
4.
3.
Crop Cover
Bio
mas
s 3.
2.
2.
Intermediate relationships implicit to
correlations with yield
NDVI: R-NIR images captured
simultaneously
(0.5 m pixels)
Near Infrared (840 nm) Visible-Red (640 nm)
Normalised difference
vegetation index (NDVI)
R
NIR
RIR
RIR
RIR
RIR
DNDN
DNDN
RIR
RIRNDVI
Reflectance values
NDVI is dimensionless with a range of -1 to 1
NDVI sensor
Ground data collection
for NDVI calibration
Lettuce monitoring with UAV
using NDVI
y = 0.0005x + 0.1303 R² = 0.8351
0.000
0.100
0.200
0.300
0.400
0.500
0.600
0.700
0.00 200.00 400.00 600.00 800.00 1000.00
ND
VI
Fresh Weight (g/m2)
Whole heads vs NDVI
F.W (g)
• NDVI values correlates with plant canopy • Plant population (> 95% accuracy)
• Mean foliar diameter (> 96% accuracy)
• Yield potential in lettuce
Derived crop information –
cauliflower (30 cm NDVI)
Predicting Harvest resource and planning marketing
Max diff of 286g
lettuce yield variation with soil
DD
DDDD
DD
DD
DD
DD
DD
DD
DD
DD
DD
DD
DDD
D
Etude cru
Antartica cru
Etude
Glassica cru
Elsol cru
Challenge
Etude cru
Antertica cru
Yucaipa cru
Excalibur cru
Yucaipa cru
Elsol
Location: Cambs Field: Spooners 3 Date: 16/07/15 Crop: Iceberg
´ 0 140 28070
MetersCopyright of G's Growers Ltd
VEGETATION INDEX MAP (NDVI)
NDVI
High
0.689 - 0.715
0.662 - 0.688
0.635 - 0.661
0.608 - 0.634
0.581 - 0.607
0.553 - 0.58
0.526 - 0.552
0.499 - 0.525
0.472 - 0.498
0.445 - 0.471
0.418 - 0.444
0.39 - 0.417
0.363 - 0.389
0.336 - 0.362
0.309 - 0.335
0.282 - 0.308
0.255 - 0.281
Low
1
10
rodham
Disease detection and progression - Celery
1 w
eek
Crop survey using a UAV with multispectral camera enable to detect the location of areas affected by Septoria Apiicola (late blight)
Affected area can be measured from imagery and tracked back to variety, planting batch, seed lot…
False colour composite images form UAV. Evolution of an organic celery patch affected by ‘late bright’ in a week difference (Oct-2014)
Integration of UK-DMC and IKONOS
for opium and wheat cultivation
estimates in Afghanistan
Impact: Poppy cultivation estimates were used to inform UK and international counter narcotics policy
Probability of poppy:
Distribution of poppy
UK-DMC strata 2009
Statistical analysis exploiting spectral clusters in UK-DMC with VHR interpretations for poppy area estimates
MODIS NDVI profiles (250 m pixel)
Figure 1. NDVI profiles from MODIS imagery at agricultural locations in - Helmand (Altitude 741 m, Latitude 31.43° N), - Balkh (Altitude 1463 m, Latitude 35.80° N) and - Badakhshan (Altitude 2502 m, Latitude 36.31° N) Ref: Taylor, et al., 2010
Vegetation productivity indicator (VPI)
crop information across Afghanistan
4 year min
4 year max
4 year ave
2009
4 year min
4 year max
4 year ave
2009
Agro-ecosystems underpinned by
Agri-informatics – Agri EPI-Centre
©Alastair Parvin for Building Design Magazine
What is the appropriate spatial and temporal scale?