Assessing Durum Wheat Yield through Sentinel-2 Imagery: A Machine Learning Approach
Two modeling approaches for the estimation of durum wheat yield based on Sentinel-2 data are presented for 66 fields across three growing periods. In the first approach, a previously developed multiple linear regression model (VI-MLR) based on vegetation indices (EVI, NMDI) was used. In the second a...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-08-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/14/16/3880 |
| _version_ | 1827598564609490944 |
|---|---|
| author | Maria Bebie Chris Cavalaris Aris Kyparissis |
| author_facet | Maria Bebie Chris Cavalaris Aris Kyparissis |
| author_sort | Maria Bebie |
| collection | DOAJ |
| description | Two modeling approaches for the estimation of durum wheat yield based on Sentinel-2 data are presented for 66 fields across three growing periods. In the first approach, a previously developed multiple linear regression model (VI-MLR) based on vegetation indices (EVI, NMDI) was used. In the second approach, the reflectance data of all Sentinel-2 bands for several dates during the growth periods were used as input parameters in three machine learning model algorithms, i.e., random forest (RF), k-nearest neighbors (KNN), and boosting regressions (BR). Modeling results were examined against yield data collected by a combine harvester equipped with a yield mapping system. VI-MLR showed a moderate performance with R<sup>2</sup> = 0.532 and RMSE = 847 kg ha<sup>−1</sup>. All machine learning approaches enhanced model accuracy when all images during the growing periods were used, especially RF and KNN (R<sup>2</sup> > 0.91, RMSE < 360 kg ha<sup>−1</sup>). Additionally, RF and KNN accuracy remained high (R<sup>2</sup> > 0.87, RMSE < 455 kg ha<sup>−1</sup>) when images from the start of the growing period until March, i.e., three months before harvest, were used, indicating the high suitability of machine learning on Sentinel-2 data for early yield prediction of durum wheat, information considered essential for precision agriculture applications. |
| first_indexed | 2024-03-09T03:54:48Z |
| format | Article |
| id | doaj.art-a6377272d4db4a80b550e56a8c131d95 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-09T03:54:48Z |
| publishDate | 2022-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-a6377272d4db4a80b550e56a8c131d952023-12-03T14:23:38ZengMDPI AGRemote Sensing2072-42922022-08-011416388010.3390/rs14163880Assessing Durum Wheat Yield through Sentinel-2 Imagery: A Machine Learning ApproachMaria Bebie0Chris Cavalaris1Aris Kyparissis2Department of Agriculture Crop Production and Rural Environment, University of Thessaly, Fytokou Str., 38446 Volos, GreeceDepartment of Agriculture Crop Production and Rural Environment, University of Thessaly, Fytokou Str., 38446 Volos, GreeceDepartment of Agriculture Crop Production and Rural Environment, University of Thessaly, Fytokou Str., 38446 Volos, GreeceTwo modeling approaches for the estimation of durum wheat yield based on Sentinel-2 data are presented for 66 fields across three growing periods. In the first approach, a previously developed multiple linear regression model (VI-MLR) based on vegetation indices (EVI, NMDI) was used. In the second approach, the reflectance data of all Sentinel-2 bands for several dates during the growth periods were used as input parameters in three machine learning model algorithms, i.e., random forest (RF), k-nearest neighbors (KNN), and boosting regressions (BR). Modeling results were examined against yield data collected by a combine harvester equipped with a yield mapping system. VI-MLR showed a moderate performance with R<sup>2</sup> = 0.532 and RMSE = 847 kg ha<sup>−1</sup>. All machine learning approaches enhanced model accuracy when all images during the growing periods were used, especially RF and KNN (R<sup>2</sup> > 0.91, RMSE < 360 kg ha<sup>−1</sup>). Additionally, RF and KNN accuracy remained high (R<sup>2</sup> > 0.87, RMSE < 455 kg ha<sup>−1</sup>) when images from the start of the growing period until March, i.e., three months before harvest, were used, indicating the high suitability of machine learning on Sentinel-2 data for early yield prediction of durum wheat, information considered essential for precision agriculture applications.https://www.mdpi.com/2072-4292/14/16/3880durum wheatyield modelingSentinel-2machine learningvegetation indices |
| spellingShingle | Maria Bebie Chris Cavalaris Aris Kyparissis Assessing Durum Wheat Yield through Sentinel-2 Imagery: A Machine Learning Approach Remote Sensing durum wheat yield modeling Sentinel-2 machine learning vegetation indices |
| title | Assessing Durum Wheat Yield through Sentinel-2 Imagery: A Machine Learning Approach |
| title_full | Assessing Durum Wheat Yield through Sentinel-2 Imagery: A Machine Learning Approach |
| title_fullStr | Assessing Durum Wheat Yield through Sentinel-2 Imagery: A Machine Learning Approach |
| title_full_unstemmed | Assessing Durum Wheat Yield through Sentinel-2 Imagery: A Machine Learning Approach |
| title_short | Assessing Durum Wheat Yield through Sentinel-2 Imagery: A Machine Learning Approach |
| title_sort | assessing durum wheat yield through sentinel 2 imagery a machine learning approach |
| topic | durum wheat yield modeling Sentinel-2 machine learning vegetation indices |
| url | https://www.mdpi.com/2072-4292/14/16/3880 |
| work_keys_str_mv | AT mariabebie assessingdurumwheatyieldthroughsentinel2imageryamachinelearningapproach AT chriscavalaris assessingdurumwheatyieldthroughsentinel2imageryamachinelearningapproach AT ariskyparissis assessingdurumwheatyieldthroughsentinel2imageryamachinelearningapproach |