2022
DOI: 10.3390/rs14143409
|View full text |Cite
|
Sign up to set email alerts
|

High-Resolution Snow-Covered Area Mapping in Forested Mountain Ecosystems Using PlanetScope Imagery

Abstract: Improving high-resolution (meter-scale) mapping of snow-covered areas in complex and forested terrains is critical to understanding the responses of species and water systems to climate change. Commercial high-resolution imagery from Planet Labs, Inc. (Planet, San Francisco, CA, USA) can be used in environmental science, as it has both high spatial (0.7–3.0 m) and temporal (1–2 day) resolution. Deriving snow-covered areas from Planet imagery using traditional radiometric techniques have limitations due to the … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
9
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
6
1
1

Relationship

1
7

Authors

Journals

citations
Cited by 12 publications
(9 citation statements)
references
References 57 publications
0
9
0
Order By: Relevance
“…The accuracy of snow cover estimation is subject to errors in the transformation and illumination conditions (Fedorov et al, 2016; Hinkler et al, 2002). Combining imagery collected by drone (Belmonte et al, 2021) or CubeSat (Cannistra et al, 2021; John et al, 2022) with machine learning methods may offer new approaches to deriving high spatial resolution snow coverage products. The accuracy of these methods is sensitive to forest canopy structure and landscape topography (Belmonte et al, 2021; Cannistra et al, 2021).…”
Section: Discussionmentioning
confidence: 99%
“…The accuracy of snow cover estimation is subject to errors in the transformation and illumination conditions (Fedorov et al, 2016; Hinkler et al, 2002). Combining imagery collected by drone (Belmonte et al, 2021) or CubeSat (Cannistra et al, 2021; John et al, 2022) with machine learning methods may offer new approaches to deriving high spatial resolution snow coverage products. The accuracy of these methods is sensitive to forest canopy structure and landscape topography (Belmonte et al, 2021; Cannistra et al, 2021).…”
Section: Discussionmentioning
confidence: 99%
“…PlanetScope satellites have been equipped, depending on the generation of the sensor system, since 2016 with four bands (Blue, Green, Red, Near Infrared-NIR; PS2 and PS2.SD generation), while the new generation-since 2019 (SuperDove; PSB.SD) has enabled eight-bands products (Coastal Blue, Green I, Yellow, and Red-Edge spectral bands are added) [24]. Based on the newly added spectral bands, the Red-Edge band played an important role in improving the accuracy of crop classification, due to the region where the spectral reflectance of green vegetation rises rapidly [35].…”
Section: Planetscope Satellite Imagery and Preprocessingmentioning
confidence: 99%
“…To overcome these issues, PlanetScope (PS) satellite imagery [22] offers significant benefits over previous RS data due to its daily temporal resolution and 3 m spatial resolution, providing an effective solution to overcome existing issues. Hence, PS data has been successfully used, for example, in mapping rubber plantations [23], snow-covered areas in forested mountain ecosystems [24], mapping of lava flows in barren regions [25], and land-cover classification in urban areas [20]. Furthermore, in 2019, PlanetScope announced the general commercial availability of the next generation of PS Monitoring product, which includes eight spectral bands, instead of previously offering four bands [26].…”
Section: Introductionmentioning
confidence: 99%
“…Potential solutions to the "training data bottleneck" include crowd-sourcing methods [103], attempts to adapt pre-existing labeled datasets [104], and weak labeling approaches [105]. The deep learning approaches developed by Cannistra et al [106] and John et al [107] produce binary snow cover maps from 3 m PlanetScope imagery using convolutional neural networks trained on thresholded aerial lidar snow-depth products. As mentioned earlier, the training requirements for neural networks greatly exceed those for random forest models-Cannistra et al [106] trained their models on 370 million pixels compared to our ~65,000 samples per land cover class (~400,000 training pixels per model).…”
Section: Limitations and Considerationsmentioning
confidence: 99%