239
A NOVEL COMBINED OPTICAL
METHOD FOR OBJECTIVELY MAP SOIL IN A NEAR REAL TIME DOMAIN
E. Ben-Dor 3 *, K. Carmina 3 , D. Heller b and S. Chudnovsky c
a The Geography and Human Environment Department
Tel-Aviv University, Israel
b Bar-Kal System Engineering’s ltd. Nethania, Israel
department of Environmental Sciences and Energy Research, The Weizmann Institute, Rehovot, Israel
Commission VI, WG VII/3
KEY WORDS: Imaging Spectroscopy (IS), Penetrating Optical Sensing (POS), Soil mapping, Near Infrared Analysis (NIRS)
ABSTRACT:
The present study demonstrates a new concept for mapping soil, toward developing a semi-automated method to in situ classify soil.
The concept is based on use of optical sensors that operate from both air and ground domains. The airborne sensor is based on
imaging spectroscopy (IS) technology whereby every pixel in the image is characterized by a reflectance spectrum. The ground
sensor is based on the Penetrating Optical Sensor (POS) technology whereby the optical head penetrates to the soil profde. This
concept was examined over a selected field in Ashdod, a southern city in Israel. The IS data used was acquired from air scanner
(AISA-ES) and POS data acquired from an ASD spectrometer over selected drill hole locations in a selected field. Five soil
properties were evaluated at every drill hole depth using the Near Infra Red analysis approach (NIRS). The POS information
enables obtaining an objective description of the soil profile, which was found to be more accurate than a traditional profile
description using open trenches. For every selected point, six soil properties (specific surface area, organic matter, hygroscopic
moisture, field moisture, carbonates, and iron oxide content) were evaluated. This information enables one to generate 5 depth
layers (20-cm intervals) of the field by interpolating all points together. The 3D information provides new insight into the soils and
opens up new frontiers for automatic soil mapping missions. Based on this study, further work is required in order to correlate the
3D information obtained by the IS and POS with any soil classification nomenclature system (e.g., USDA).
1. INTRODUCTION
The conventional soil survey mission is performed by using
extensive field observations (sometimes very subjective) and a
follow-up laboratory analysis, which adds valuable
information about the soil's properties in question (USDA,
1999). The currently used classical soil survey is a rather
complicated, expensive, and time- and money-consuming
process. Consequently, there is a great demand for an
alternative method to map soils rapidly and accurately. We
suggest here to combine two spectral-based approaches to
replace the conventional soil mapping process. They are as
follows: 1) a ground approach based on the recent innovative
method developed by Ben-Dor et al. (2008) that describes the
soil profile by a Penetrating Optical Spectroscopy (POS)
approach using fiber spectral assembly, ASD spectrometer,
and NIRS models, and 2) an airborne approach that describes
the soil surface by Imaging spectroscopy (IS) technology. The
IS approach serves as a convenient method that might replace
the traditional, ancient air photo method, providing cognitive
and quantitative spatial views of the areas in question, and the
POS may replace the wet laboratory measurements and the
subjective field observation of open trenches. The purpose of
this study is thus to demonstrate the integration between these
two spectral-based methods (POS and IS) for mapping soil
digitally, rapidly, and cost effectively in a 3D view.
2. MATERIAL AND METHODS
2.1. The Study Area
An area of about 4000m 2 , characterized by alluvial soils, was
selected to carry out this demonstration. The area is a bare filed
situated in Ashdod, a southern city in Israel. Selected for its
agriculture activity (seasonal wheat and cotton crops), the area
was covered by one strip of the airborne imaging spectrometer,
AISA-ES, and later also by conventional and POS soil survey
approaches.
2.2. Airborne Data
The AISA-ES airborne sensor is a programmed imaging
spectrometer consisting of two sensors mounted on the same
optical bench and aligned to look at (almost) the same focal
plane. The sensors are Eagle for the VIS region and Hawk for
the SWIR region (See Speclm homepage at
http://www.specim.fi/). We used 180 bands (60 in the VNIR and
120 in the SWIR), which acquired information from 7,000 feet,
providing a pixel size of 3 m. After the data acquisition (on
August 2007), the raw data were converted into reflectance
values using a combined ACRON and Empirical Line method.
The reflectance obtained was validated against ground spectra
of several selected targets. The reflectance image was used to
generate a surface map that was based on the surface spectral
information.
2.3 POS - data
The POS measurements were carried out by using a 3S-HED
assembly (Sub Surface Spectral Head Device) to spectrally view
and interpret the borehole’s walls from inside using fiber optic
sensing and an illumination head. This device is hooked to a 25°
bare fiber optic of an ASD field spectrometer that is sensitive to
the VNIR-NIR spectral region (350-2500nm). The spectral
information was modeled against the soil chemistry by using the
NIRS approach (Ben-Dor and Banin, 1995). A detailed
description of the 3S-HED concept can be found in Ben-Dor et
al. (2008). Measurements were taken in intervals of 20cm from