Abstract
Agriculture is a cornerstone of global economies and livelihoods, with
precise crop production estimates and yield predictions being critical
for farmers, agriculturalists, and policymakers. Such estimates offer
valuable insights for informed decision-making, especially in the
context of the current climate crisis, which has significantly impacted
key crops like rice and maize. In Dindigul district, official reports
reveal a steady decline in rice production. To analyze the spatial
variation in rice yields, spatial data is crucial, and remote sensing
proves indispensable for managing crops at a site-specific level by
providing both spatial and temporal insights. The integration of remote
sensing and GIS technologies in crop production estimation has
garnered widespread attention for its potential to enhance accuracy
and efficiency in agricultural practices. Crop growth models, which
simulate the interactions between plants and their environment, enable
predictions of crop yields and assessments of climate change impacts
on food security. This paper explores the application of remote sensing
data to estimate rice yields in Dindigul district. Satellite imagery with a
30-meter spatial resolution, acquired from NASA’s Landsat-8 program
during the mature stage of the Kharif season (October), was processed
using GIS platforms. Machine learning techniques were employed to
classify land use and land cover, while a JavaScript algorithm in the
Google Earth Engine platform was utilized to compute the Normalized
Difference Vegetation Index (NDVI), which was correlated with
historical rice yield data. A principal component regression model was
then developed to predict rice yields based on NDVI values from the
study area. The model's predictions were compared with official
agricultural statistics from the International Crops Research Institute
for the Semi-Arid Tropics (ICRISAT), demonstrating an 80% accuracy
rate in estimating rice yields. Furthermore, the study examined long
term changes in rainfall and temperature over the past 70 years,
revealing an increase in both factors, which has notably influenced
local cropping patterns.
Authors
S. Siva Padma Devi1, C. Florance Annal2
Mother Teresa Women’s University, India1, M.V. Mutiah Government Arts College for Women, India2
Keywords
Remote Sensing, NDVI, Machine Learning, Climate Change, and Rice Yield