Employing a Multi-Input Deep Convolutional Neural Network to Derive Soil Clay Content from a Synergy of Multi-Temporal Optical and Radar Imagery Data

Nikolaos Tziolas,John Theocharis,Nikolaos Tsakiridis,Eyal Ben-Dor,George Zalidis

doi:10.3390/rs12091389

Abstract

Earth observation (EO) has an immense potential as being an enabling tool for mapping spatial characteristics of the topsoil layer. Recently, deep learning based algorithms and cloud computing infrastructure have become available with a great potential to revolutionize the processing of EO data. This paper aims to present a novel EO-based soil monitoring approach leveraging open-access Copernicus Sentinel data and Google Earth Engine platform. Building on key results from existing data mining approaches to extract bare soil reflectance values the current study delivers valuable insights on the synergistic use of open access optical and radar images. The proposed framework is driven by the need to eliminate the influence of ambient factors and evaluate the efficiency of a convolutional neural network (CNN) to effectively combine the complimentary information contained in the pool of both optical and radar spectral information and those form auxiliary geographical coordinates mainly for soil. We developed and calibrated our multi-input CNN model based on soil samples (calibration = 80% and validation 20%) of the LUCAS database and then applied this approach to predict soil clay content. A promising prediction performance (R2 = 0.60, ratio of performance to the interquartile range (RPIQ) = 2.02, n = 6136) was achieved by the inclusion of both types (synthetic aperture radar (SAR) and laboratory visible near infrared–short wave infrared (VNIR-SWIR) multispectral) of observations using the CNN model, demonstrating an improvement of more than 5.5% in RMSE using the multi-year median optical composite and current state-of-the-art non linear machine learning methods such as random forest (RF; R2 = 0.55, RPIQ = 1.91, n = 6136) and artificial neural network (ANN; R2 = 0.44, RPIQ = 1.71, n = 6136). Moreover, we examined post-hoc techniques to interpret the CNN model and thus acquire an understanding of the relationships between spectral information and the soil target identified by the model. Looking to the future, the proposed approach can be adopted on the forthcoming hyperspectral orbital sensors to expand the current capabilities of the EO component by estimating more soil attributes with higher predictive performance.

Highlights

In a world subject to constant climate change and increasing pressures by agricultural intensification and other human activities, soil is considered a vital but endangered component of the global life support system [1]
Padarian et al [6] and Wadoux et al [7] expanded the digital soil mapping previously proposed by Behrens et al [8], by including deep learning techniques to optimally search for local contextual information of covariates, while Tsakiridis et al [9] recently introduce an interpretable novel localized multi-channel 1-D convolutional neural network (CNN)
In order to assess the performance of calibration models for soil clay content we calculated the roIont-omrdeearnt-osqausasreesserthroerp(eRrMfoSrmE,aEnqcueaotfiocnal(i1b)r)a, ttihoencmoeoffideclisenfot rosfodiel tcelramy icnoanttioennt(wR2e, cEaqlucualtaiotend(2th))e, arnoodt-tmheearant-isoqoufarpeeerfrorormr (aRnMceSEto, Einqtuearqtiuoanr(t1il)e),rtahnegceo(eRffPicIQien, Et oqfudateitoenrm(3in))a. tTiohne(eRq2,uEaqtiuoantsiounse(2d))w, aenred aths eforlalotiwoso: f performance to interquartile range (RPIQ, Equation (3))

Summary

Introduction

In a world subject to constant climate change and increasing pressures by agricultural intensification and other human activities, soil is considered a vital but endangered component of the global life support system [1]. Digital soil mapping based on spaceborne Earth observation (EO) data is currently undergoing a significant shift This is driven first and foremost by the advent of the big EO data era, mainly spearheaded by open access of the Landsat archive in 2008 [3], and more recently by the operation of the Europe’s Copernicus programme, which provide free and open super spectral imagery data. This is further supported by novel machine learning algorithms for the spatial estimations of soil properties from spaceborne-sourced environmental covariates and harmonized soil observations [4]. Padarian et al [6] and Wadoux et al [7] expanded the digital soil mapping previously proposed by Behrens et al [8], by including deep learning techniques to optimally search for local contextual information of covariates, while Tsakiridis et al [9] recently introduce an interpretable novel localized multi-channel 1-D convolutional neural network (CNN)

Objectives

Methods

Results

Discussion

Conclusion