Optimizing Gaussian Process Regression for Image Time Series Gap-Filling and Crop Monitoring
Image processing entered the era of artificial intelligence, and machine learning algorithms emerged as attractive alternatives for time series data processing. Satellite image time series processing enables crop phenology monitoring, such as the calculation of start and end of season. Among the pro...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-04-01
|
| Series: | Agronomy |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2073-4395/10/5/618 |
| _version_ | 1827717962604216320 |
|---|---|
| author | Santiago Belda Luca Pipia Pablo Morcillo-Pallarés Jochem Verrelst |
| author_facet | Santiago Belda Luca Pipia Pablo Morcillo-Pallarés Jochem Verrelst |
| author_sort | Santiago Belda |
| collection | DOAJ |
| description | Image processing entered the era of artificial intelligence, and machine learning algorithms emerged as attractive alternatives for time series data processing. Satellite image time series processing enables crop phenology monitoring, such as the calculation of start and end of season. Among the promising algorithms, Gaussian process regression (GPR) proved to be a competitive time series gap-filling algorithm with the advantage of, as developed within a Bayesian framework, providing associated uncertainty estimates. Nevertheless, the processing of time series images becomes computationally inefficient in its standard per-pixel usage, mainly for GPR training rather than the fitting step. To mitigate this computational burden, we propose to substitute the per-pixel optimization step with the creation of a cropland-based precalculations for the GPR hyperparameters <inline-formula> <math display="inline"> <semantics> <mi mathvariant="bold-italic">θ</mi> </semantics> </math> </inline-formula>. To demonstrate our approach hardly affects the accuracy in fitting, we used Sentinel-2 LAI time series over an agricultural region in Castile and Leon, North-West Spain. The performance of image reconstructions were compared against the standard per-pixel GPR time series processing. Results showed that accuracies were on the same order (RMSE 0.1767 vs. 0.1564 <inline-formula> <math display="inline"> <semantics> <mrow> <mo>[</mo> <msup> <mi mathvariant="normal">m</mi> <mn>2</mn> </msup> <mo>/</mo> <msup> <mi mathvariant="normal">m</mi> <mn>2</mn> </msup> <mo>]</mo> </mrow> </semantics> </math> </inline-formula>, 12% RMSE degradation) whereas processing time accelerated about 90 times. We further evaluated the alternative option of using the same hyperparameters for all the pixels within the complete scene. It led to similar overall accuracies over crop areas and computational performance. Crop phenology indicators were also calculated for the three different approaches and compared. Results showed analogous crop temporal patterns, with differences in start and end of growing season of no more than five days. To the benefit of crop monitoring applications, all the gap-filling and phenology indicators retrieval techniques have been implemented into the freely downloadable GUI toolbox DATimeS. |
| first_indexed | 2024-03-10T20:12:45Z |
| format | Article |
| id | doaj.art-b681e5e2ea204e6d99dfa5afb604c6d6 |
| institution | Directory Open Access Journal |
| issn | 2073-4395 |
| language | English |
| last_indexed | 2024-03-10T20:12:45Z |
| publishDate | 2020-04-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Agronomy |
| spelling | doaj.art-b681e5e2ea204e6d99dfa5afb604c6d62023-11-19T22:50:40ZengMDPI AGAgronomy2073-43952020-04-0110561810.3390/agronomy10050618Optimizing Gaussian Process Regression for Image Time Series Gap-Filling and Crop MonitoringSantiago Belda0Luca Pipia1Pablo Morcillo-Pallarés2Jochem Verrelst3Image Processing Laboratory (IPL), Parc Científic, University of Valencia, Paterna, 46980 Valencia, SpainImage Processing Laboratory (IPL), Parc Científic, University of Valencia, Paterna, 46980 Valencia, SpainImage Processing Laboratory (IPL), Parc Científic, University of Valencia, Paterna, 46980 Valencia, SpainImage Processing Laboratory (IPL), Parc Científic, University of Valencia, Paterna, 46980 Valencia, SpainImage processing entered the era of artificial intelligence, and machine learning algorithms emerged as attractive alternatives for time series data processing. Satellite image time series processing enables crop phenology monitoring, such as the calculation of start and end of season. Among the promising algorithms, Gaussian process regression (GPR) proved to be a competitive time series gap-filling algorithm with the advantage of, as developed within a Bayesian framework, providing associated uncertainty estimates. Nevertheless, the processing of time series images becomes computationally inefficient in its standard per-pixel usage, mainly for GPR training rather than the fitting step. To mitigate this computational burden, we propose to substitute the per-pixel optimization step with the creation of a cropland-based precalculations for the GPR hyperparameters <inline-formula> <math display="inline"> <semantics> <mi mathvariant="bold-italic">θ</mi> </semantics> </math> </inline-formula>. To demonstrate our approach hardly affects the accuracy in fitting, we used Sentinel-2 LAI time series over an agricultural region in Castile and Leon, North-West Spain. The performance of image reconstructions were compared against the standard per-pixel GPR time series processing. Results showed that accuracies were on the same order (RMSE 0.1767 vs. 0.1564 <inline-formula> <math display="inline"> <semantics> <mrow> <mo>[</mo> <msup> <mi mathvariant="normal">m</mi> <mn>2</mn> </msup> <mo>/</mo> <msup> <mi mathvariant="normal">m</mi> <mn>2</mn> </msup> <mo>]</mo> </mrow> </semantics> </math> </inline-formula>, 12% RMSE degradation) whereas processing time accelerated about 90 times. We further evaluated the alternative option of using the same hyperparameters for all the pixels within the complete scene. It led to similar overall accuracies over crop areas and computational performance. Crop phenology indicators were also calculated for the three different approaches and compared. Results showed analogous crop temporal patterns, with differences in start and end of growing season of no more than five days. To the benefit of crop monitoring applications, all the gap-filling and phenology indicators retrieval techniques have been implemented into the freely downloadable GUI toolbox DATimeS.https://www.mdpi.com/2073-4395/10/5/618Gaussian processes regressiontime seriescrop monitoringSentinel-2phenology indicatorsoptimization |
| spellingShingle | Santiago Belda Luca Pipia Pablo Morcillo-Pallarés Jochem Verrelst Optimizing Gaussian Process Regression for Image Time Series Gap-Filling and Crop Monitoring Agronomy Gaussian processes regression time series crop monitoring Sentinel-2 phenology indicators optimization |
| title | Optimizing Gaussian Process Regression for Image Time Series Gap-Filling and Crop Monitoring |
| title_full | Optimizing Gaussian Process Regression for Image Time Series Gap-Filling and Crop Monitoring |
| title_fullStr | Optimizing Gaussian Process Regression for Image Time Series Gap-Filling and Crop Monitoring |
| title_full_unstemmed | Optimizing Gaussian Process Regression for Image Time Series Gap-Filling and Crop Monitoring |
| title_short | Optimizing Gaussian Process Regression for Image Time Series Gap-Filling and Crop Monitoring |
| title_sort | optimizing gaussian process regression for image time series gap filling and crop monitoring |
| topic | Gaussian processes regression time series crop monitoring Sentinel-2 phenology indicators optimization |
| url | https://www.mdpi.com/2073-4395/10/5/618 |
| work_keys_str_mv | AT santiagobelda optimizinggaussianprocessregressionforimagetimeseriesgapfillingandcropmonitoring AT lucapipia optimizinggaussianprocessregressionforimagetimeseriesgapfillingandcropmonitoring AT pablomorcillopallares optimizinggaussianprocessregressionforimagetimeseriesgapfillingandcropmonitoring AT jochemverrelst optimizinggaussianprocessregressionforimagetimeseriesgapfillingandcropmonitoring |