Three dimensional soil organic carbon monitoring using VNIR reflectance spectroscopic techniques


CALL: 2010

DOMAIN: SR - Environmental and Earth Sciences


LAST NAME: Schlerf





START: 2010-12-01

END: 2012-12-31


Submitted Abstract

Current monitoring techniques cannot produce a full inventory of the soil organic carbon (SOC) taking into account both its spatial variability and its vertical distribution in the soil profile. Traditional sampling and analysis techniques are often not capable of dealing with the large spatial variability requiring very large numbers of data as they are too laborious and costly. Recently, imaging spectroscopy has proved to be a powerful tool to map SOC in the plough layer of croplands at a high resolution, although the accuracy still has to be improved. Hence, a large number of SOC data can be extracted from these datasets allowing the analysis of the spatial patterns and a very good estimation of fields SOC average values. However, full SOC inventories, to be used for greenhouse gas accounting, would also require taking into account the subsoil that can be an important reservoir of SOC, and the soil bulk density. Furthermore, there is an urgent demand for continuousmapping of the vertical SOC distribution to improve process-based SOC dynamics modeling taking into account the effect of variability in agricultural management and lateral flows of sediment withassociated carbon and water. Proximal and remote sensing techniques, such as those based on reflectance spectroscopy, can provide an efficient way to collect the amount of data required in SOC monitoring or modeling. Since both the surface and the first meter of the soil are important for such investigations, a combination of imaging spectroscopy and in situ spectral measurements of soil profiles is needed to create a complete assessment of the soil. These data sources will be used as input for spatial models producing a 3D SOC inventory. The main advantage of such models is that they give an insight in the factors determining the spatial patterns and that they can be used topredict the vertical SOC distribution at un-sampled points. The aim of this proposal is to develop an innovative methodology that can produce high resolution SOC inventories for the entire soil profile using a combination of imaging spectroscopy and spectroscopic sounding techniques. Three techniques will be combined to produce threedimensional maps of SOC stocks in cropland: imaging spectroscopy for the plough layer, spectroscopic profilers for the vertical SOC gradient and soil cores for the bulk density. These techniques will be applied in Luxembourg where a variety of SOC contents can be found within a short distance thus enhancing the predictive capacity of the empirical models. Furthermore, a previous project has given us the experience and infrastructure to ascertain the field campaigns.The continuous surface SOC data (imaging spectroscopy) with the point data on SOC at different depth (spectral profiler) will be integrated in a spatial model. Previous studies have shown that pedotransfer functions could be derived to estimate all variables and parameters of the SOC depth distribution model at the regional scale. Here we will scale down and adapt this approach to include variability and intensity of landscape scale processes such as tillage regime, erosion and sedimentation. For surface SOC content, an area of ~350 km2 will be over flown by an aircraft carrying the Airborne Prism Experiment (APEX) sensor. For SOC content in the first meter of the soil, an adapted head attached to an ASD spectrometer will be lowered in the soil in c. 10 recently ploughed fields at the time of the flight campaign. Bulk density will determined from sections of the cores needed to lower the profiler in the soil using the traditional volume/weight method. A relationship will be established between the SOC content of calibration samples collected in the testfields and the spectral information (corrected spectral data from the APEX in the case of surface SOC and the ASD recordings for the profiling) using partial l

This site uses cookies. By continuing to use this site, you agree to the use of cookies for analytics purposes. Find out more in our Privacy Statement