Abstract:
The deployment of high-revisit satellite-based radar sensors raises the question of
whether the data collected can provide quantitative information to improve agricultural
productivity. This thesis aims to develop and test mathematical algorithms to describe
the dynamic backscatter of high-resolution Synthetic Aperture Radar (Sentinel-1) in
order to describe the development and productivity of wheat at field-scale. A time series
of the backscatter ratio (VH/VV), collected over a cropping season, could be
characterised by a growth and a senescence logistic curve and related to critical phases
of crop development. The curve parameters, referred to as Crop Productivity Indicators
(CPIs), compared well with the crop production for three years at the Rothamsted
experimental farm. The combination of different parameters (e.g. midpoints of the two
curves) helped to define CPIs, such as duration, that significantly (r = 0.61, p = 0.05)
correlated with measured yields. Field observations were used to understand the wheat
evolution by sampling canopy characteristics across the seasons. The correlation
between the samples and the CPIs showed that structural changes, like biomass
increase, influence the CPIs during the growth phase, and that declining plant water
content was correlated with VH/VV values during maturation. The methodology was
upscaled to other farms in Hertfordshire and Norfolk. The ANOVA identified significant
effects (p<0.001) of farm management, year (weather conditions) and the interaction
between soil type and year on the selected CPIs. Multilinear regression models between
yields and selected CPIs displayed promising predictive power (R²= 0.5) across different
farms in the same year. However, these models could not explain yield differences within
high-yielding farms across seasons because of the dominant effect of weather patterns
on the CPIs in each year. The potential impact of the research includes estimation of
yield across the landscape, phenology monitoring and indication biophysical parameters.
Future work on SAR-derived CPIs should focus on improving the correlations with
biophysical properties, applying of the methodology in other crops, with different soils
and climates.
Description:
Richter, G. M. Industrial supervisor ( Rothamsted Research)
Burgess, Paul J. and Meersmans, Jeroen Associate supervisors