Snow and Cloud Classification in Historical SPOT Images: An Image Emulation Approach for Training a Deep Learning Model Without Reference Data
The lack of revisit in long-term satellite time series, such as Landsat is an issue to assess ecosystems response to snow cover variations in mountains. A recent release of the Satellites Pour l'Observation de la Terre (SPOT) 1-5 satellite images collection by the SPOT World Heritage (SWH...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10422893/ |
| _version_ | 1797269208274305024 |
|---|---|
| author | Zacharie Barrou Dumont Simon Gascoin Jordi Inglada |
| author_facet | Zacharie Barrou Dumont Simon Gascoin Jordi Inglada |
| author_sort | Zacharie Barrou Dumont |
| collection | DOAJ |
| description | The lack of revisit in long-term satellite time series, such as Landsat is an issue to assess ecosystems response to snow cover variations in mountains. A recent release of the Satellites Pour l'Observation de la Terre (SPOT) 1-5 satellite images collection by the SPOT World Heritage (SWH) program offers the opportunity to increase the temporal revisit of Landsat from 1986 to 2015 at 20 m resolution. However, spectral and radiometric limitations of these images hinder the application of well-established pixel-wise methods to extract the snow cover area. As a work-around, deep learning techniques, such as convolutional neural networks can incorporate both spectral and spatial information to classify every pixel as snow, cloud, or snow-free. However, the lack of reference data poses a challenge to the implementation of such data-driven approaches. Here, we develop an emulator of SPOT images, which takes as input Sentinel-2 images. As a result, an emulated SPOT image can be paired with a reference snow map generated from its source Sentinel-2 image to train a deep learning model able to process actual SPOT images. We follow this approach to train a U-Net and evaluate different training strategies. We apply the different models to classify actual SPOT images for which we have reference data for validation. The method yields high precision in detecting snow, with minimal false snow pixel identification. This is at the cost of overestimating cloud pixels around clouds and highly saturated areas. The results confirm the potential of this method to generate time series of snow cover maps using the SWH collection. |
| first_indexed | 2024-03-07T14:05:20Z |
| format | Article |
| id | doaj.art-ce6173fe8d1e442bbc15b1da0f79970c |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-04-25T01:44:43Z |
| publishDate | 2024-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-ce6173fe8d1e442bbc15b1da0f79970c2024-03-08T00:00:09ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352024-01-01175541555210.1109/JSTARS.2024.336183810422893Snow and Cloud Classification in Historical SPOT Images: An Image Emulation Approach for Training a Deep Learning Model Without Reference DataZacharie Barrou Dumont0https://orcid.org/0009-0004-9515-5757Simon Gascoin1https://orcid.org/0000-0002-4996-6768Jordi Inglada2https://orcid.org/0000-0001-6896-0049Centre d'Etudes Spatiales de la Biosphère (CESBIO), Université de Toulouse, CNRS/CNES/IRD/INRAE/UPS, Toulouse, FranceCentre d'Etudes Spatiales de la Biosphère (CESBIO), Université de Toulouse, CNRS/CNES/IRD/INRAE/UPS, Toulouse, FranceCentre d'Etudes Spatiales de la Biosphère (CESBIO), Université de Toulouse, CNRS/CNES/IRD/INRAE/UPS, Toulouse, FranceThe lack of revisit in long-term satellite time series, such as Landsat is an issue to assess ecosystems response to snow cover variations in mountains. A recent release of the Satellites Pour l'Observation de la Terre (SPOT) 1-5 satellite images collection by the SPOT World Heritage (SWH) program offers the opportunity to increase the temporal revisit of Landsat from 1986 to 2015 at 20 m resolution. However, spectral and radiometric limitations of these images hinder the application of well-established pixel-wise methods to extract the snow cover area. As a work-around, deep learning techniques, such as convolutional neural networks can incorporate both spectral and spatial information to classify every pixel as snow, cloud, or snow-free. However, the lack of reference data poses a challenge to the implementation of such data-driven approaches. Here, we develop an emulator of SPOT images, which takes as input Sentinel-2 images. As a result, an emulated SPOT image can be paired with a reference snow map generated from its source Sentinel-2 image to train a deep learning model able to process actual SPOT images. We follow this approach to train a U-Net and evaluate different training strategies. We apply the different models to classify actual SPOT images for which we have reference data for validation. The method yields high precision in detecting snow, with minimal false snow pixel identification. This is at the cost of overestimating cloud pixels around clouds and highly saturated areas. The results confirm the potential of this method to generate time series of snow cover maps using the SWH collection.https://ieeexplore.ieee.org/document/10422893/Deep learningimage classificationSatellites Pour l'Observation de la Terre (SPOT) World Heritage (SWH)sentinel-2snow coveru-net |
| spellingShingle | Zacharie Barrou Dumont Simon Gascoin Jordi Inglada Snow and Cloud Classification in Historical SPOT Images: An Image Emulation Approach for Training a Deep Learning Model Without Reference Data IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Deep learning image classification Satellites Pour l'Observation de la Terre (SPOT) World Heritage (SWH) sentinel-2 snow cover u-net |
| title | Snow and Cloud Classification in Historical SPOT Images: An Image Emulation Approach for Training a Deep Learning Model Without Reference Data |
| title_full | Snow and Cloud Classification in Historical SPOT Images: An Image Emulation Approach for Training a Deep Learning Model Without Reference Data |
| title_fullStr | Snow and Cloud Classification in Historical SPOT Images: An Image Emulation Approach for Training a Deep Learning Model Without Reference Data |
| title_full_unstemmed | Snow and Cloud Classification in Historical SPOT Images: An Image Emulation Approach for Training a Deep Learning Model Without Reference Data |
| title_short | Snow and Cloud Classification in Historical SPOT Images: An Image Emulation Approach for Training a Deep Learning Model Without Reference Data |
| title_sort | snow and cloud classification in historical spot images an image emulation approach for training a deep learning model without reference data |
| topic | Deep learning image classification Satellites Pour l'Observation de la Terre (SPOT) World Heritage (SWH) sentinel-2 snow cover u-net |
| url | https://ieeexplore.ieee.org/document/10422893/ |
| work_keys_str_mv | AT zachariebarroudumont snowandcloudclassificationinhistoricalspotimagesanimageemulationapproachfortrainingadeeplearningmodelwithoutreferencedata AT simongascoin snowandcloudclassificationinhistoricalspotimagesanimageemulationapproachfortrainingadeeplearningmodelwithoutreferencedata AT jordiinglada snowandcloudclassificationinhistoricalspotimagesanimageemulationapproachfortrainingadeeplearningmodelwithoutreferencedata |