Comparative Analysis of Wheat Yield Estimation Based on Vegetation Indices Derived From Landsat 8 and Sentinel-2 Imageries Under Rainfed Condition
Abstract
Early crop yield prediction is considered as important factor for planning and in various policy decisions. Lack of reliable estimation of crop yield and production estimates become unrealistic leading to faulty decisions and actions and subsequent uncertainties in agriculture food security and policy making. Its demand increased under changing climate especially in the rainfed regions where agriculture is more vulnerable. Newly emerged remote-sensing technologies provide real time estimation of vegetation and yield both temporally and spatially. Therefore, the current study planned based on remote-sensing data derived from different satellite imageries for real time, cost-effective and reliable solution for wheat yield estimation prior to harvest. The time series cloud free scenes of LANDSAT 8 and Sentinel-2 imageries for the growing season 2015-16 of wheat crop were downloaded for estimation of wheat yield based on vegetation indices such as Enhanced Vegetation Index (EVI2), Green Normalized Difference Vegetation Index (GNDVI), Normalized Difference Vegetation Index (NDVI), Wide Dynamic Range Vegetation Index (WDRVI), Modified Chlorophyll Absorbed Ratio Index (MCARI) and Transformed Chlorophyll Absorbed Ratio Index (TCARI). Yield samples were collected over 43 sites in Chakwal district manually along with GPS positions for validation. The regressional model were performed between vegetation indices derived from the satellite imageries and the ground-truthing wheat yield data on selected sites. The study results showed that all indices showed higher values for the February-2016 months as compared to December-2015 months due to presence of more vegetation. The overall accuracy improved (9-21%) with Sentinel-2 imagery for almost all indices in all the Feburary-2016 month with EVI2, WDRVI and NDVI respectively. Particular EVI2 showed highest R2= 0.81 value with Sentinel-2 imagery with 21% higher accuracy over Landsat 8 imagery while GNDVI performs better with Landsat 8 for the Feburary-2016 month. The assimilation of information derived from Sentinel-2 imagery can enable regional applications with more reliable estimation of wheat yield prior to harvest which can be useful in policy making and maintaining national food security stocks timely.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFMGC31K1383T
- Keywords:
-
- 1622 Earth system modeling;
- GLOBAL CHANGEDE: 1630 Impacts of global change;
- GLOBAL CHANGEDE: 1632 Land cover change;
- GLOBAL CHANGEDE: 1640 Remote sensing;
- GLOBAL CHANGE