Abstract:
Rice (Oryza sativa L.) is a staple crop in Asian countries. Plant parameters,
such as above-ground biomass, leaf moisture and plant height are important indicators
in rice crop monitoring. The objective of this study was to test the correlation between
ground truth data and remotely sensed satellite data. The experiment was done at rice
farmer-field at Gampaha, Sri Lanka cultivated with Bg 374 variety. Crop measure-
ments were collected from five quadrant samples per single satellite pixel across the
field. This was repeated in two growth stages, panicle initiation and booting stage.
The satellite images with 10 m x 10 m resolution were downloaded using Google
Earth Engine platform. The sample quadrant locations were labelled by an RTK-
enabled drone flown over the field before data collection. The zonal statistics tool of
QGIS software was used to extract waveband data from corresponding satellite pixels
and used to compute vegetative indices (VIs). Regression analysis results showed best
relationship with AGB Soil Adjusted Vegetation Index (SAVI: R 2 =0.48) in panicle
initiation stage and Green Vegetation Index (GVI: R 2
=0.46) in booting stage. The
Greenness Index (GI: R 2 =0.20) exhibited the best relationship with leaf moisture in
panicle initiation stage and Normalized Difference Vegetation Index (NDVI:
R2 =0.93) in booting stage. For plant height (Float disc method), Ratio Vegetation In-
dex (RVI) exhibited the best relationship both in panicle initiation (R 2 =0.19) and in
booting stage (R 2 =0.26). The findings could be of future use for advanced remote
sensing techniques in monitoring rice crops for smart agriculture.
