Please login first
Integration of Object-Oriented Remote Sensing and Machine Learning to Create Field Model for Optimized Regional Agricultural Management
1  Middle East Technical University Department of Geodetic and Geographic Information Technologies
Academic Editor: Riccardo Buccolieri

https://doi.org/10.3390/ECRS2023-15834 (registering DOI)
Abstract:

In an era marked by tools like Artificial Intelligence (AI), Machine Learning (ML) and remote sensing (RS), agriculture is a primary beneficiary. These technologies help to optimize agricultural productivity, optimizing resource using and increase yield. They not only optimizing usage but also adapt to climate change and manage risks associated with agricultural practices become inevitable. Vegetation Indices (VI) such as Normalized Difference Vegetation Index (NDVI) are relatively simple and useful algorithms that can be used to implement precision agriculture (PA). Optical satellite images can sense the reflected lights coming from leaves which can provide various crop development information used to implement PA. Agriculture sector is important for regional economy. If managed properly, many problems related with this sector can be eliminated like climate change, environmental problems, and economic development. PA applications can be used to create regional management policies. Remote sensing of agriculture for regional management practices is the main component of this study. This study involves monitoring agricultural production both seasonally and daily using Sentinel-2 multispectral time-series data. Time-series images from 2017 to 2022 are analyzed to detect shifts in phenological dates of crops. To understand these shift, a combination of NDSI ( Normalized Difference Salinity Index), MSAVI (Modified Soil Adjusted Vegetation Index), NDVI and NDMI (Normalized Difference Moisture Index) is used. First, mean MSAVI is calculated by the year, and phenological dates are determined according to the mean MSAVI values. For the bare soil dates, NDSI were calculated to understand the change of soil salinity. For the specific dates, the field is mosaicked and polygonised for each year with the machine learning methods. For the dates of seed germination, MSAVI is used, and the same procedure is developed. For the start of season and the rest, NDVI and NDMI are used. These shifts are then modeled using ML algorithms, and predictions are made for the year 2023. With these significant, planning of agricultural events can be arranged optimally for the next crop season. This process helps for planning the schedule of agricultural production and assists in regional management practices. The second step is controlling mean VI values for the dates found in the previous step. For daily changes, object – oriented and pixel-based methods (land segmentation) for field model are used to trends in the field. The field model includes the characteristics of the field, and MSAVI, NDVI and NDMI are used. In PA, site specific solutions are extremely important to get the optimum results. By characterizing the field, site specific solutions can be applied. Vegetation Indices used to create these characteristics. Both image processing and machine learning algorithms are used. According to the findings, it will be possible to optimize inputs for agricultural production which helps to decide for regional agricultural management. Since agricultural events have a great effect on agricultural applications, using meteorological data is the main milestone to improve this study. Overall, this research aims to contribute to regional agricultural production and management modules by using remote sensing and machine learning technology.

Keywords: remote sensing; precision agriculture; machine learning; image processing; vegetation index; regional agricultural management
Comments on this paper
Currently there are no comments available.



 
 
Top