SoilNet: An Attention-based Spatio-temporal Deep Learning Framework for Soil Organic Carbon Prediction with Digital Soil Mapping in Europe

Digital soil mapping (DSM) is an advanced approach that integrates statistical modeling and cutting-edge technologies, including machine learning (ML) methods, to accurately depict soil properties and their spatial distribution. Soil organic carbon (SOC) is a crucial soil attribute providing valuable insights into soil health, nutrient cycling, greenhouse gas emissions, and overall ecosystem productivity. This study highlights the significance of spatial-temporal deep learning (DL) techniques within the DSM framework. A novel architecture is proposed, incorporating spatial information using a base convolutional neural network (CNN) model and spatial attention mechanism, along with climate temporal information using a long short-term memory (LSTM) network, for SOC prediction across Europe. The model utilizes a comprehensive set of environmental features, including Landsat-8 images, topography, remote sensing indices, and climate time series, as input features. Results demonstrate that the proposed framework outperforms conventional ML approaches like random forest commonly used in DSM, yielding lower root mean square error (RMSE). This model is a robust tool for predicting SOC and could be applied to other soil properties, thereby contributing to the advancement of DSM techniques and facilitating land management and decision-making processes based on accurate information.

[1]  M. Wiesmeier,et al.  Optimized bare soil compositing for soil organic carbon prediction of topsoil croplands in Bavaria using Landsat , 2023, ISPRS Journal of Photogrammetry and Remote Sensing.

[2]  S. Chakraborty,et al.  Developing regional soil micronutrient management strategies through ensemble learning based digital soil mapping , 2023, Geoderma.

[3]  David P. Hofmeyr,et al.  Multivariate Random Forest for Digital Soil Mapping , 2023, SSRN Electronic Journal.

[4]  P. Lagacherie,et al.  Multiscale Evaluations of Global, National and Regional Digital Soil Mapping Products in France Across Regions and Soil Properties , 2022, SSRN Electronic Journal.

[5]  A. Don,et al.  Spatial prediction of organic carbon in German agricultural topsoil using machine learning algorithms , 2022, SOIL.

[6]  Lei Zhang,et al.  A CNN-LSTM Model for Soil Organic Carbon Content Prediction with Long Time Series of MODIS-Based Phenological Variables , 2022, Remote. Sens..

[7]  J. D. Wegner,et al.  Deep Learning and Earth Observation to Support the Sustainable Development Goals: Current approaches, open challenges, and future opportunities , 2022, IEEE Geoscience and Remote Sensing Magazine.

[8]  T. Scholten,et al.  Improving the spatial prediction of soil organic carbon using environmental covariates selection: A comparison of a group of environmental covariates , 2022, CATENA.

[9]  Deep Learning for the Earth Sciences , 2021 .

[10]  Yannis Avrithis,et al.  All the attention you need: Global-local, spatial-channel attention for image retrieval , 2021, 2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV).

[11]  Francis E. H. Tay,et al.  Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNet , 2021, 2021 IEEE/CVF International Conference on Computer Vision (ICCV).

[12]  M. V. D. van der Heijden,et al.  Agricultural diversification promotes multiple ecosystem services without compromising yield , 2020, Science Advances.

[13]  S. Gelly,et al.  An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale , 2020, ICLR.

[14]  B. Giannetti,et al.  Emergy-based ecosystem services valuation and classification management applied to China’s grasslands , 2020 .

[15]  T. Behrens,et al.  3D mapping of soil organic carbon content and soil moisture with multiple geophysical sensors and machine learning , 2020, Vadose Zone Journal.

[16]  Thorsten Behrens,et al.  The relevant range of scales for multi-scale contextual spatial modelling , 2019, Scientific Reports.

[17]  Alexandre M.J.C. Wadoux,et al.  Using deep learning for multivariate mapping of soil with quantified uncertainty , 2019, Geoderma.

[18]  Zisis I. Petrou,et al.  Prediction of Sea Ice Motion With Convolutional Long Short-Term Memory Networks , 2019, IEEE Transactions on Geoscience and Remote Sensing.

[19]  Jun Huang,et al.  Spectral-Spatial Attention Networks for Hyperspectral Image Classification , 2019, Remote. Sens..

[20]  Joachim Denzler,et al.  Deep learning and process understanding for data-driven Earth system science , 2019, Nature.

[21]  Jun Fu,et al.  Dual Attention Network for Scene Segmentation , 2018, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[22]  Budiman Minasny,et al.  Using deep learning for digital soil mapping , 2018, SOIL.

[23]  In-So Kweon,et al.  CBAM: Convolutional Block Attention Module , 2018, ECCV.

[24]  Radu Horaud,et al.  A Comprehensive Analysis of Deep Regression , 2018, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[25]  J. Abatzoglou,et al.  TerraClimate, a high-resolution global dataset of monthly climate and climatic water balance from 1958–2015 , 2018, Scientific Data.

[26]  Abhinav Gupta,et al.  Non-local Neural Networks , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[27]  Stephen E. Fick,et al.  WorldClim 2: new 1‐km spatial resolution climate surfaces for global land areas , 2017 .

[28]  Gang Sun,et al.  Squeeze-and-Excitation Networks , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[29]  Lukasz Kaiser,et al.  Attention is All you Need , 2017, NIPS.

[30]  Marvin N. Wright,et al.  SoilGrids250m: Global gridded soil information based on machine learning , 2017, PloS one.

[31]  S. Jade,et al.  Uncertainties in the Shuttle Radar Topography Mission (SRTM) Heights: Insights from the Indian Himalaya and Peninsula , 2017, Scientific Reports.

[32]  C. Wirth,et al.  On the combined effect of soil fertility and topography on tree growth in subtropical forest ecosystems—a study from SE China , 2017 .

[33]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[34]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[35]  Dominique Arrouays,et al.  GlobalSoilMap : Basis of the global spatial soil information system , 2014 .

[36]  Hui-Ming Cheng,et al.  Direct reduction of graphene oxide films into highly conductive and flexible graphene films by hydrohalic acids , 2010 .

[37]  S. K. Alavipanah,et al.  Estimating soil organic carbon from soil reflectance: a review , 2010, Precision Agriculture.

[38]  Thomas Scholten,et al.  Pedogenesis, permafrost, and soil moisture as controlling factors for soil nitrogen and carbon contents across the Tibetan Plateau , 2009 .

[39]  H. Elsenbeer,et al.  Soil organic carbon concentrations and stocks on Barro Colorado Island — Digital soil mapping using Random Forests analysis , 2008 .

[40]  Thomas Scholten,et al.  Spatial and vertical variation of soil carbon at two grassland sites — Implications for measuring soil carbon stocks , 2007 .

[41]  J. L. Parra,et al.  Very high resolution interpolated climate surfaces for global land areas , 2005 .

[42]  S. Hochreiter,et al.  Long Short-Term Memory , 1997, Neural Computation.

[43]  B. J. Carter,et al.  Slope gradient and aspect effects on soils developed from sandstone in Pennsylvania , 1991 .

[44]  B. Donald Theoretical basis for differentiation of ferric-iron bearing minerals using Landsat MSS data , 1982 .

[45]  Wilfred M. Post,et al.  Soil carbon pools and world life zones , 1982, Nature.

[46]  Huanjun Liu,et al.  Digital Soil Mapping Based on Fine Temporal Resolution Landsat Data Produced by Spatiotemporal Fusion , 2023, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[47]  Cuixian Lu,et al.  Global Ocean Wind Speed Retrieval From GNSS Reflectometry Using CNN-LSTM Network , 2023, IEEE Transactions on Geoscience and Remote Sensing.

[48]  Changming Zhu,et al.  The effectiveness of digital soil mapping with temporal variables in modeling soil organic carbon changes , 2022 .

[49]  K. Millard,et al.  SatViT: Pretraining Transformers for Earth Observation , 2022, IEEE Geoscience and Remote Sensing Letters.

[50]  Weitao Chen,et al.  GCSANet: A Global Context Spatial Attention Deep Learning Network for Remote Sensing Scene Classification , 2022, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[51]  Maryam Hassanzadeh,et al.  Enhancing the accuracy of machine learning models using the super learner technique in digital soil mapping , 2021 .

[52]  B. Minasny,et al.  On digital soil mapping , 2003 .

[53]  Amro F. Alasta,et al.  Using Remote Sensing data to identify iron deposits in central western Libya , 2022 .