Using deep learning to predict soil properties from regional spectral data

Abstract

Diffuse reflectance infrared spectroscopy allows the rapid acquisition of soil information in the field or thelaboratory. The vis-NIR spectroscopy research enthusiasm around the world has created regional to globalsoil spectral libraries. While machine learning methods have been utilised in processing spectral data, suchlarge regional datasets are better dealt with big data analytics. Deep learning is an exciting discipline thathas already transformed the way data are analysed in many fields and could also change the way we modelsoil spectral data. This study developed and evaluated convolutional neural networks (CNNs), a type ofdeep learning algorithm, as a new way to predict soil properties from raw soil spectra. We demonstratedthe effectiveness of CNNs on the LUCAS soil database, which consists of around 20,000 topsoil observationswith physicochemical and biological properties from Europe. To fully utilise the capacity of the CNN model,we represented the soil spectral data as a 2-dimensional spectrogram, showing the reflectance as a functionof wavelength and frequency. We showed the capacity of a CNN to be trained in a multi-task setting tosimultaneously predict six soil properties in one model (OC, CEC, clay, sand, pH, total N). Compared withconventional methods such as PLS regression and Cubist regression tree, the CNN performed significantlybetter, especially the multi-tasking model. In the case of soil organic carbon prediction, the multi-task CNNdecreased the error by 87% compared to PLS and 62% compared with Cubist. This approach proved to beeffective when trained on a relatively large dataset. The high accuracy of CNN makes it an ideal tool formodelling soil spectral data.Keywords: Convolutional Neural Networks, spectrograms, multi-task learning, simultaneous prediction

Tasks

Reproductions