Deep Neural Networks with Transfer Learning for Forest Variable Estimation Using Sentinel-2 Imagery in Boreal Forest
Estimation of forest structural variables is essential to provide relevant insights for public and private stakeholders in forestry and environmental sectors. Airborne light detection and ranging (LiDAR) enables accurate forest inventory, but it is expensive for large area analyses. Continuously inc...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2021-06-01
|
Series: | Remote Sensing |
Subjects: | |
Online Access: | https://www.mdpi.com/2072-4292/13/12/2392 |
id |
doaj-b7a0a7d1f39546f2a130d67d8e6766fc |
---|---|
record_format |
Article |
spelling |
doaj-b7a0a7d1f39546f2a130d67d8e6766fc2021-07-01T00:34:47ZengMDPI AGRemote Sensing2072-42922021-06-01132392239210.3390/rs13122392Deep Neural Networks with Transfer Learning for Forest Variable Estimation Using Sentinel-2 Imagery in Boreal ForestHeikki Astola0Lauri Seitsonen1Eelis Halme2Matthieu Molinier3Anne Lönnqvist4VTT Technical Research Centre of Finland Ltd., P.O. Box 1000, FI-02044 Espoo, FinlandVTT Technical Research Centre of Finland Ltd., P.O. Box 1000, FI-02044 Espoo, FinlandVTT Technical Research Centre of Finland Ltd., P.O. Box 1000, FI-02044 Espoo, FinlandVTT Technical Research Centre of Finland Ltd., P.O. Box 1000, FI-02044 Espoo, FinlandVTT Technical Research Centre of Finland Ltd., P.O. Box 1000, FI-02044 Espoo, FinlandEstimation of forest structural variables is essential to provide relevant insights for public and private stakeholders in forestry and environmental sectors. Airborne light detection and ranging (LiDAR) enables accurate forest inventory, but it is expensive for large area analyses. Continuously increasing volume of open Earth Observation (EO) imagery from high-resolution (<30 m) satellites together with modern machine learning algorithms provide new prospects for spaceborne large area forest inventory. In this study, we investigated the capability of Sentinel-2 (S2) image and metadata, topography data, and canopy height model (CHM), as well as their combinations, to predict growing stock volume with deep neural networks (DNN) in four forestry districts in Central Finland. We focused on investigating the relevance of different input features, the effect of DNN depth, the amount of training data, and the size of image data sampling window to model prediction performance. We also studied model transfer between different silvicultural districts in Finland, with the objective to minimize the amount of new field data needed. We used forest inventory data provided by the Finnish Forest Centre for model training and performance evaluation. Leaving out CHM features, the model using RGB and NIR bands, the imaging and sun angles, and topography features as additional predictive variables obtained the best plot level accuracy (RMSE% = 42.6%, <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mo>|</mo><mi>B</mi><mi>I</mi><mi>A</mi><mi>S</mi><mo>%</mo><mo>|</mo></mrow></semantics></math></inline-formula> = 0.8%). We found <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>3</mn><mo>×</mo><mn>3</mn></mrow></semantics></math></inline-formula> pixels to be the optimal size for the sampling window, and two to three hidden layer DNNs to produce the best results with relatively small improvement to single hidden layer networks. Including CHM features with S2 data and additional features led to reduced relative RMSE (RMSE% = 28.6–30.7%) but increased the absolute value of relative bias (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mo>|</mo><mi>B</mi><mi>I</mi><mi>A</mi><mi>S</mi><mo>%</mo><mo>|</mo></mrow></semantics></math></inline-formula> = 0.9–4.0%). Transfer learning was found to be beneficial mainly with training data sets containing less than 250 field plots. The performance differences of DNN and random forest models were marginal. Our results contribute to improved structural variable estimation performance in boreal forests with the proposed image sampling and input feature concept.https://www.mdpi.com/2072-4292/13/12/2392deep learningdeep neural networkstransfer learningsentinel-2boreal forestforest variables |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Heikki Astola Lauri Seitsonen Eelis Halme Matthieu Molinier Anne Lönnqvist |
spellingShingle |
Heikki Astola Lauri Seitsonen Eelis Halme Matthieu Molinier Anne Lönnqvist Deep Neural Networks with Transfer Learning for Forest Variable Estimation Using Sentinel-2 Imagery in Boreal Forest Remote Sensing deep learning deep neural networks transfer learning sentinel-2 boreal forest forest variables |
author_facet |
Heikki Astola Lauri Seitsonen Eelis Halme Matthieu Molinier Anne Lönnqvist |
author_sort |
Heikki Astola |
title |
Deep Neural Networks with Transfer Learning for Forest Variable Estimation Using Sentinel-2 Imagery in Boreal Forest |
title_short |
Deep Neural Networks with Transfer Learning for Forest Variable Estimation Using Sentinel-2 Imagery in Boreal Forest |
title_full |
Deep Neural Networks with Transfer Learning for Forest Variable Estimation Using Sentinel-2 Imagery in Boreal Forest |
title_fullStr |
Deep Neural Networks with Transfer Learning for Forest Variable Estimation Using Sentinel-2 Imagery in Boreal Forest |
title_full_unstemmed |
Deep Neural Networks with Transfer Learning for Forest Variable Estimation Using Sentinel-2 Imagery in Boreal Forest |
title_sort |
deep neural networks with transfer learning for forest variable estimation using sentinel-2 imagery in boreal forest |
publisher |
MDPI AG |
series |
Remote Sensing |
issn |
2072-4292 |
publishDate |
2021-06-01 |
description |
Estimation of forest structural variables is essential to provide relevant insights for public and private stakeholders in forestry and environmental sectors. Airborne light detection and ranging (LiDAR) enables accurate forest inventory, but it is expensive for large area analyses. Continuously increasing volume of open Earth Observation (EO) imagery from high-resolution (<30 m) satellites together with modern machine learning algorithms provide new prospects for spaceborne large area forest inventory. In this study, we investigated the capability of Sentinel-2 (S2) image and metadata, topography data, and canopy height model (CHM), as well as their combinations, to predict growing stock volume with deep neural networks (DNN) in four forestry districts in Central Finland. We focused on investigating the relevance of different input features, the effect of DNN depth, the amount of training data, and the size of image data sampling window to model prediction performance. We also studied model transfer between different silvicultural districts in Finland, with the objective to minimize the amount of new field data needed. We used forest inventory data provided by the Finnish Forest Centre for model training and performance evaluation. Leaving out CHM features, the model using RGB and NIR bands, the imaging and sun angles, and topography features as additional predictive variables obtained the best plot level accuracy (RMSE% = 42.6%, <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mo>|</mo><mi>B</mi><mi>I</mi><mi>A</mi><mi>S</mi><mo>%</mo><mo>|</mo></mrow></semantics></math></inline-formula> = 0.8%). We found <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>3</mn><mo>×</mo><mn>3</mn></mrow></semantics></math></inline-formula> pixels to be the optimal size for the sampling window, and two to three hidden layer DNNs to produce the best results with relatively small improvement to single hidden layer networks. Including CHM features with S2 data and additional features led to reduced relative RMSE (RMSE% = 28.6–30.7%) but increased the absolute value of relative bias (<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mo>|</mo><mi>B</mi><mi>I</mi><mi>A</mi><mi>S</mi><mo>%</mo><mo>|</mo></mrow></semantics></math></inline-formula> = 0.9–4.0%). Transfer learning was found to be beneficial mainly with training data sets containing less than 250 field plots. The performance differences of DNN and random forest models were marginal. Our results contribute to improved structural variable estimation performance in boreal forests with the proposed image sampling and input feature concept. |
topic |
deep learning deep neural networks transfer learning sentinel-2 boreal forest forest variables |
url |
https://www.mdpi.com/2072-4292/13/12/2392 |
work_keys_str_mv |
AT heikkiastola deepneuralnetworkswithtransferlearningforforestvariableestimationusingsentinel2imageryinborealforest AT lauriseitsonen deepneuralnetworkswithtransferlearningforforestvariableestimationusingsentinel2imageryinborealforest AT eelishalme deepneuralnetworkswithtransferlearningforforestvariableestimationusingsentinel2imageryinborealforest AT matthieumolinier deepneuralnetworkswithtransferlearningforforestvariableestimationusingsentinel2imageryinborealforest AT annelonnqvist deepneuralnetworkswithtransferlearningforforestvariableestimationusingsentinel2imageryinborealforest |
_version_ |
1721348207264399360 |