Predicting Corn Phenological Stages with Multispectral Time Series Remote Sensing Data by Threshold Based and Trend Detection Methods

Date

2023-12-27

Author

Şenkardeşler, Emine

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

80
views

32
downloads

Cite This

Instead of treating a crop field as one homogeneous area, Precision Agriculture (PA) allows site-specific management to optimize the inputs and outputs in agricultural production. The PA concept has gained acceleration with tools like Artificial Intelligence (AI), Machine Learning (ML) and Remote Sensing (RS). Vegetation Indices (VI) which are produced and manipulated by using these tools, are simple yet useful algorithms and facilitate the implementation of PA. Phenology is a study area which helps PA practices and can be estimated with VI data. In this thesis, two remote sensing methods estimating corn phenology are compared to estimate the phenological stage dates for the 2017 to 2022 seasons. With Sentinel-2 NIR and RED bands, Modified Soil Adjusted Vegetation Index (MSAVI) and Normalized Difference Vegetation Index (NDVI) is calculated. While MSAVI is calculated to estimate early stages of corn, NDVI is calculated to estimate late stages. After MSAVI reaches a specific threshold, dates and values are replaced with NDVI values and new merged VI data is created. The merged VI data is pre-processed with Median and Savitzky-Golay (SG) Filter. This merged and filtered VI data is used to compare two phenology estimation methods, the Threshold Based Model (TBM), and the Moving Average Convergence Divergence (MACD) Based Model. Results are compared with Crop Progress Report (CPR) to find the model with least error, which will be used for further field model studies based on phenological dates. Depending on the comparative analysis between the TBM and MACD, the findings clearly indicate that TBM outperforms MACD in terms of estimating phenological stages of corn.

Subject Keywords

Remote Sensing, Machine Learning, Phenology, Vegetation Index

URI

https://hdl.handle.net/11511/108196

Collections

Graduate School of Natural and Applied Sciences, Thesis