Assessing Soil Organic Matter (SOM) on a Landscape Scale by Combining Non-Invasive (Spectroscopic) and Invasive Methods
Zusammenfassung der Projektergebnisse
The overall goal of this study, which combined two cooperative projects, was to explore how much information about the quantity and quality of SOM could be obtained with VIS-NIR spectroscopy combined with lab FTIR spectroscopy, Py-FIMS and laboratory (physicochemical) fractionation schemes. To this end a database was built with field and lab data from 172 sites that were sampled in a multiform landscape with widely differing conditions of soil and SOM formation. New cooperations enabled to use further analytical techniques (SEM image analysis, 13C CPMAS solid state NMR, hyperspectral VIS-NIR imaging, portable FTIR) beyond the initial project plan. For molecular and functional characterization of SOM, a subset of 24 soil samples was density fractionated. The molecular composition of SOM was analyzed using Py-FIMS and SEM (scanning electron microscopy) analysis to image the contribution of microbial biomass to SOM. Most distinguishing were the high contents of carbohydrates for the MOM fraction and of lipids for the lighter POM fractions. Microbial derived carbohydrates and fatty acids were both enriched in MOM. POM fractions of all four soils were analyzed for structural properties by 13C CPMAS solid state NMR. Recent work investigates in how far FTIR measurements allow the prediction of NMR characteristics. To use the complete database with VIS-NIR spectra of all 172 samples we defined a number of calibration/validation sets from these data, characterized by different sizes and different degree of heterogeneity. PLSR plus bagging and CARS (for spectral variable selection) can be recommended as standard approach for more parsimonious models and more accurate predictions (obtained in our study for OC, N, Cmic, Cinert). CWT with an appropriate mother wavelet facilitated the physical interpretation of the underlying spectral predictive mechanisms as it allows a separate representation of high and low frequency features. Additionally, CWT-CARS-PLSR has shown to be a promising approach especially for regional studies which have to deal with a certain level of heterogeneity in soil conditions and soil spectral behaviour. In a multi-regional approach, we found spiking to be very appropriate for regionally-adapted soil spectroscopic inventories with the need of only a very limited number of samples being necessary for wet-chemical analysis. Hyperspectral VIS-NIR snapshot imaging can support in-field inventories, as image segmentation was helpful to provide improved estimates for soil properties. Also roughness classification directly from the images could allow for stratified calibration approaches. To investigate spatio-temporal changes of SOM and their impact on VIS-NIR spectra of arable soil surfaces a monthly soil sampling was done for one year at 21 arable sites and in three depths, i.e. 0- 1, 1-5 and 5-10 cm. Contents of chlorophyll α, OC, Nt, Cmic, Chwe and of PLFA were analyzed. It was found that soil surfaces are highly dynamic and do not represent deeper layers of the topsoil horizon. Soil algae are a driving factor for SOM properties of arable topsoils. The strong spectral properties of algal chlorophyll enabled determination by VIS-NIRS and predicting the quantity of algal biomass by PLSR models from chlorophyll α contents. Several common wavelengths for chlorophyll and Cmic were identified by CARS selection and both parameters were significantly correlated. Hence, chlorophyll was supposed to trigger the modelling of Cmic, while it is also assumed that chlorophyll masked OC in the visible part of the spectrum. The contribution of soil algae to SOM in arable soils is further investigated in ongoing research and planned for follow-up research. MIR (FTIR) spectroscopy was superior to VIS-NIRS for all studied soil properties. 2D-correlations between both spectral regions did not reflect the existing physical couplings and were blurred probably due to the heterogeneity of the studied sample set. We found spectra of a currently available portable FTIR instrument to match well with spectra measured with a classical FTIR lab device. Thus, we consider the portable FTIR technology to provide a promising alternative to VIS-NIR field spectroscopy with higher capabilities for quantitative approaches and a very fast data acquisition; follow-up research should explore the full potential of this technique for in-situ soil monitoring.
Projektbezogene Publikationen (Auswahl)
-
(2014): Determination of soil properties with visible to near- and mid-infrared spectroscopy: Effects of spectral variable selection. Geoderma 223-225, 88-96
Vohland M., Ludwig M., Thiele-Bruhn S., Ludwig B.
-
(2014): Snapshot Hyperspectral Imaging for Soil Diagnostics – Results of a Case Study in the Spectral Laboratory. PFG (Photogrammetrie – Fernerkundung – Geoinformation) 6, 511-522
] Jung A., Vohland M.
-
(2014): Spectral Mobile Mapping for Rapid Soil Diagnostics – Results of a Laboratory Based Feasibility Test. Joint Conference 2014 of DGfK, DGPF, GfGI and GiN, Hamburg, 26.-28.03.2014, DGPF conference proceedings 23 / 2014
Jung A., Vohland M.
-
(2014): The use of full range spectroradiometer data to assess properties of a heterogeneous soil set in a regional scale survey. Proc. SPIE 9239, Remote Sensing for Agriculture, Ecosystems, and Hydrology XVI, 92391E (29 October 2014)
Harbich M., Udelhoven T., Jung A., Vohland M., Ludwig M., Thiele-Bruhn S.
-
(2015): Microbial contribution to SOM quantity and quality in density fractions of temperate arable soils. Soil Biology and Biochemistry 81, 311–322
Ludwig M., Achtenhagen J., Miltner A., Eckhardt K.-U., Leinweber P., Emmerling C., Thiele-Bruhn S.
-
(2015): Use of a Portable Camera for Proximal Soil Sensing with Hyperspectral Image Data. Remote Sensing 2015, 7(9), 11434-11448
Jung A., Vohland M., Thiele-Bruhn S.
-
(2015): Usefulness of middle infrared spectroscopy for an estimation of chemical and biological soil properties – Underlying principles and comparison of different software packages. Soil Biology and Biochemistry 86, 116- 125
Ludwig B., Sawallisch A., Heinze S., Joergensen R.G., Vohland M.
-
Quantification of Soil Variables in a Heterogeneous Soil Region With VIS–NIR–SWIR Data Using Different Statistical Sampling and Modeling Strategies. IEEE Journal of Selected Topics in Applied Earth Observation and Remote Sensing, Vol 9 Issue 9, Sept. 2016, pp 4011-4021
Vohland M., Harbich M., Ludwig M., Emmerling C., Thiele-Bruhn S.
-
Using variable selection and wavelets to exploit the full potential of visible–near infrared spectra for predicting soil properties. Journal of Near Infrared Spectroscopy, 24 (3), 255-269, 2016
Vohland M., Ludwig M., Harbich M., Emmerling C., Thiele-Bruhn S.