What is Remote Sensing? Understanding Remote Sensing

 

Remote sensing is the acquisition of information about an object or phenomenon without making physical contact with the object. The advancement of remote sensing technology is key in conducting efficient soil surveys and soil mapping.

Recent technological advances in satellite remote sensing have helped to overcome the limitation of conventional soil survey, thus providing a new outlook for soil survey and mapping. Remote sensing has proved to be an important part of soil survey and mapping.

Soil properties that have been measured using remote sensing approaches include mineralogy, texture, soil iron, soil moisture, soil organic carbon, soil salinity and carbonate content.

In sparsely vegetated areas, successful use of space borne, airborne and in situ measurements using optical, passive and active microwave instruments has been reported. 

In densely vegetated areas however, soil data acquisition typically relied on indirect retrievals using soil indicators, such as plant functional groups, productivity changes, and Ellenberg indicator values.

Optical remote sensing helps in the mapping of properties like land cover, land type, vegetation and soil moisture. Thermal infrared remote sensing is commonly used to estimate moisture and salinity.

Visual image interpretation technique helps in the identification and mapping of soil elements like land type, vegetation, land use, slope and relief.

Microwave remote sensing is a new and effective technique for mapping of soil moisture and salinity which is being commonly used today. Hyperspectral remote sensing is another recent method which is applied in soil salinity mapping as well as identification and mapping of minerals in the soil.

At the end of this article, students will be able to;

·  Understand what Remote Sensing is all about

·  Know different remote sensing methods for soil survey and mapping

 

Read: Kinds of Soil Surveys

Important facts to know

· Remote sensing is the acquisition of information about an object or phenomenon without making physical contact with the object.

· Recent technological advances in satellite remote sensing have helped to overcome the limitation of conventional soil survey, thus providing a new outlook for soil survey and mapping.

· Soil properties that have been measured using remote sensing approaches include mineralogy, texture, soil iron, soil moisture, soil organic carbon, soil salinity and carbonate content.

· Optical remote sensing helps in the mapping of properties like land cover, land type, vegetation and soil moisture.

· Thermal infrared remote sensing is commonly used to estimate moisture and salinity. Visual image interpretation technique helps in the identification and mapping of soil elements like land type, vegetation, land use, slope and relief.

· Microwave remote sensing is a new and effective technique for mapping of soil moisture and salinity which is being commonly used today.

· Hyperspectral remote sensing is another recent method which is applied in soil salinity mapping as well as identification and mapping of minerals in the soil.

Read: Types of Soil: Characteristics and Importance

Different Remote Sensing Methods in Soil Survey and Mapping

1. Optical Remote Sensing

The surface features reflected on satellite image provide enough information to accurately delineate the boundaries which is accomplished effectively through systematic interpretation of satellite imageries.

Optical remote sensing has been used to monitor various properties of soil like land cover, land type, vegetation and even soil moisture.

Optical remote sensing provides a quantitative measure of surface reflectance, that is, the reflected radiation of the sun from the Earth’s surface, which is related to some soil properties.

Organic matter, particle size and moisture content influence soil reflectance primarily through a change in average surface reflectance, and produce only broad spectral expression (Irons et al., 1989).

Optical remote sensing is the most commonly used for soil moisture estimation. Liu et al. (2003) analyzed 18 different soils that represent a large range of permanent soil characteristics and investigated the potential of estimating soil moisture from reflectance measurements in the solar domain.

2. Thermal Infrared Remote Sensing

The thermal infrared remote sensing is commonly used to estimate moisture and salinity. Thermal infrared remote sensing measures the thermal emission of the Earth with an electromagnetic wavelength region between 3.5 and 14μm.

The moisture content is mainly measured by the thermal inertia method and the temperature/vegetation index method.

Thermal infrared remote sensing is also commonly used to detect salt-affected areas from the relationship between crop water stress and temperatures of the crop canopy.

Although thermal infrared remote sensing has many scopes, the potential use of thermal systems for soil monitoring appears to be little investigated.

For the reflected solar radiation, the most important characteristics of a soil that determine its reflectance properties are (Sunita 2016):

i. Moisture: Increasing soil moisture content decreases the reflectance in the water absorption bands but also in the remaining bands due to the internal reflections within the water film covering the soil particles; thus wet soils appear darker (less reflective) than dry soils.

ii. Organic matter: Increasing organic matter content gives darker (less reflective) soils;

iii. Texture: Sandy soils are more reflective than clay soils;

iv. Surface roughness: Decreases in surface roughness slightly increase reflection: an example is the development of soil crust;

v. Iron content: Increasing the content of iron oxide corresponds for many soils to a change incolour towards their characteristic brick-red colour, which implies an increased reflection ionin the red and a decrease in green.

Some of the following problems can occur while mapping Soil from thermal remotely sensed data:

i. Identifying, categorizing and mapping soils can be a complex procedure which in many cases is based on soil properties that are not even visible to the naked eye and require field or laboratory analyses (e.g. pH).

ii. Soil is a complex three-dimensional body. The majority of remote sensing systems only characterize the surface or, in optimum conditions, shallow depths of soils. In many cases, the surface characteristics may not be representative of the deeper soil body (e.g. soil organic carbon concentration decreases with depth).

iii. Soil properties can vary dramatically both spatially and temporally within a small area.

iv. The upper surface can be subject to frequent alteration by tillage, precipitation, erosion, crusting and other surface processes.

v. Vegetation coverage obscures most soils for most or all the time. Soil subjected to arable cultivation will be exposed after ploughing. Soil undernatural vegetation may never be exposed.

vi. The signal recorded by sensor is the result of a combination of several soil properties (which are frequently interlinked). Such mixtures often mask the signal from a feature under investigation.

vii. The spectral resolution of sensors is not suitable for mapping soil characteristics (i.e. not covering diagnostic regions of the spectrum, focused on observing vegetation).

3. Visual Image Interpretation

Visual interpretation is based on shape, size, tone, shadow, texture, pattern, site and association. This has the advantage of being relatively simple and inexpensive.

Soils are surveyed and mapped, following a three tier approach, comprising interpretation of remote sensing imagery and/or aerial photograph (Mulder, 1987), field survey (including laboratory analysis of soil samples) and cartography.

This technique helps in the identification and mapping of soil elements like land type, vegetation, land use, slope and relief.

Interpretation of aerial photographs have also been used in soil salinity mapping, especially colour-infrared photographs in which barren saline soils (in white) and salt-stressed crops (in reddish brown) can be easily discriminated from other soil surface and vegetation features (Rao and Venkataratnam, 1991).

4. Microwave Remote Sensing

Microwave remote sensing is an effective technique for mapping of soil moisture and salinity, with advantages for all-weather observations and solid physics.

It presents advantages in special soil conditions, such as salt-affected areas (Taylor et al., 1996), sandy coastal and desert zones, waterlogged areas, and places with irregular micro topography such as puffy crusts and cloddy surfaces. 

There are two methods of microwave sensing - active microwave sensing and passive microwave sensing. Great progress has been made in mapping regional soil moisture with active microwave sensors.

In active microwave methods, a microwave pulse is sent and received. The power of the received signal is compared with which was sent to determine the backscattering coefficient of the surface, which has been shown to be sensitive to soil moisture (Wang and Qu, 2009).

The most common imaging active microwave configuration is the synthetic aperture radar (SAR), which transmits a series of pulses as the radar antenna traverses the. 

Active sensors, although having the capability to provide high spatial resolution in the order of tens of meters, have a poor resolution in time with repeat time excess of 1 month.

On the other hand, the space borne passive systems can provide spatial resolutions only in the order of tens of kilometres but with a higher temporal resolution. Passive microwave remote sensors can be used to monitor surface soil moisture over land surfaces (Wigneron et al., 2004).

These sensors measure the intensity of microwave emission from the soil, which is proportional to the brightness temperature, a product of the surface temperature and emissivity (Wang and Qu, 2009). Because of the differential behaviour of the real and imaginary parts of the dielectric constant of soil, microwaves also are efficient in detecting soil salinity.

While the real part is independent of soil salinity and alkalinity, the imaginary part is highly sensitive to variations in soil electrical conductivity, but with no bearing on variations in alkalinity. This allows the separation of saline soils from others.

5. Hyperspectral Remote Sensing

Recent developments in hyperspectral remote sensing offer the potential of significantly improving data input to predictive soil models. The key characteristic of hyperspectral imagery data is the high spectral resolution that is provided over a large and continuous wavelength region.

Each pixel in a hyperspectral image is associated with hundreds of data points that represent the spectral signature of the materials within the spatial area of the pixel. The result is a three-dimensional data set that has two axes of spatial information and one axis of spectral information.

The high resolution of hyperspectral imagery makes it possible to uniquely identify different materials at the earth's surface. The large number of spectral bands permits direct identification of minerals in surface soils.

Clark and Swayze (1996) mapped over 30 minerals using hyperspectral sensor, Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) at Cuprites, Nevada.

AVIRIS measures a contiguous spectrum in the visible and near-infrared, and thereby better characterize atmospheric and surface properties (Rimjhim et al., 2013).

6. Airborne topographic Lidar

Light Detection and Ranging (lidar) is an emerging geospatial technology that is improving our characterization of terrestrial landscapes.

Advantages over other forms of remotely sensed data include spatial data collected in 3D, geo-referenced during acquisition, and ability to classify 3D elements within point clouds into user-defined surface features and above-surface features (Renslow, 2012).

Improved representations of the Earth’s surface, surface feature structure, and reflectance intensity allow broad use of lidar technology for mapping terrain derivatives and landscape conditions critical for soil investigations. 

High horizontal and vertical accuracy allow mapping of terrain features that contribute to our knowledge of soil properties and dynamic processes across multiple scales.

At a suitable resolution, lidar helps to identify subtle topographic controls on soil variability traditionally missed at coarser scales.

Topography controls water redistribution on the landscape, which in turn controls pedogenesis over geologic time and subsequent soil distribution across a landscape. These scientific concepts are not new to soil resource inventories.

However, data such as lidar and the other aforementioned tools provide spatially explicit representations of soils and soil processes in a quantifiable format.

Digital soil mapping processes quantify and capture soil patterns determined by topography, parent materials and other soil forming factors (McBratney et al., 2003) and this information in a digital format for computer-based applications (Sunita, 2016).

 

Read: Soil Mapping: Meaning, Types and methods

Components of Remote Sensing

The major critical component for any remote sensing program is reliable ground truth information. Without ground data to identify land cover categories, to train the classifier and validate the output image products, it is impossible to run a defensible program that provides reliable results.

Ground truth is mentioned first, because it must be seriously considered before initiating plans for any remote sensing application.

Secondly, a source of satellite imagery is required. There are many sources of satellite imagery which vary considerably in cost, as well as, spatial, temporal, spectral and radiometric resolution.

Finding an imagery source that also provides a guarantee of future continuity is an important consideration, since once a program has been researched and implemented, it becomes more difficult to transition to another satellite.

Thirdly, using remotely sensed data requires a sizable investment in Information Technology (IT) resources. However, with the speed of computers continuing to increase and the price of disc storage on the decline this has become much less of a hindrance.

1. Ground Truth

NASS has two sources of field level crop information for ground truth, its own June Area Survey (JAS) and the USDA Farm Service Agency (FSA). NASS collects the June Area Survey (JAS) segment data and the FSA collects CLU polygon data. The scope of the FSA CLU program is comprehensive including all states and extensive coverage of major crops.

The program is run at the county level in over 2,300 FSA county offices.

There are two important differences between JAS and FSA data, as ground truth, in the CDL program.

First, the JAS data requires manual digitization of individual segments by NASS staff or cooperators while the FSA data does not.

The individual polygon boundaries of the JAS segments are regularly digitized to support the survey but the individual fields within each segment require additional digitization. 

The FSA CLU polygon data are digitized and crop specific attribute data collected in the FSA county offices as part of a standardized GIS layer that collects information on all fields in FSA programs on a near real time basis for compliance and administration purposes (Mueller et al., 2009).

A second difference is that the coverage of major crops provided by the FSA are more comprehensive than the 150 – 400 one square mile area segments included in the JAS data, approximating full coverage in major speculative states.

However, there are several shortcomings to using the FSA data. First, approximately fifty percent of CLU polygons include more than one crop type per CLU while JAS segments are digitized to the field (Craig, 2005).

In order to use the FSA data, CLUs with mixed crop types, except certain double crops such as winter wheat followed by soybeans, are excluded from the ground truth.

Second, specialty crops are not well represented in the FSA data leading to a bias toward “program crops”, for which farmers received subsidies.

Third, not all CLU polygons are attributed each year (Mueller et al., 2009). Fortunately, these shortcomings are greatly overshadowed by the sheer volume of crop data available from the FSA CLU program. Being a comprehensive agricultural data set that requires minimal preparation and can be updated multiple times during the growing season greatly outweighs the disadvantages.

Using the FSA CLU and 578 attribute data for training has dramatically increased the volume and timeliness of available ground truth and thereby increased the scope, efficiency and accuracy of the operational CDL program.

2. Imagery

In the late 1990s, NASS used both Landsat TM and ETM+ data with a 30 meter spatial resolution inCDL production.

The Landsat sensors have a 185 km swath; seven spectral bands including a visible blue, visible green, visible red, near infrared red (NIR), two mid infrared (MIR) bands and a thermal band; a 16 day repeat and 8 bit quantization. 

The synchronization of the two sensors to achieve an 8 day repeat cycle was appropriate for acquiring crop information during the growing season.

Landsat data were purchased and made available to NASS via the USDA’s Foreign Agricultural Service (FAS), which established the satellite image archive (SIA) for the purpose of coordinated purchases of satellite imagery for the entire Department of Agriculture (Craig, 2009).

On May 31, 2003, the Landsat 7 ETM+ sensor experienced an anomaly in its scan line corrector. At the time, the imagery was considered unusable by NASS and the CDL program experienced a 50% reduction in the inventory of available satellite imagery.

In 2004 the USDApurchased imagery, for evaluation purposes, from the Indian Remote Sensing Satellite (IRS) RESOURCESAT-1 launched in October of 2003.

The moderate spatial resolution (56 meter) Advanced Wide Field Sensor (AWiFS) data were selected for evaluation as a substitute for Landsat imagery inCDL production. 

NASS conducted investigations to assess the effectiveness of AWiFS data for crop acreage estimation including: Nebraska, 2004 (Boryan and Craig, 2005); Arkansas (Delta Region);Nebraska, 2005 and Coincident Studies, (Arkansas, Illinois, Iowa) 2005 (NASS, 2006; Seffrin, 2007;Johnson, 2008) after which time NASS decided along with its partner, FAS, to purchase AWiFS dataexclusively for the USDA’s SIA, International Productions Assessment Unit.

In 2006, NASS began using AWiFS data as the primary source of imagery. The AWiFS sensor offers a moderate spatial resolution (56 meter); a large swath width (720 km), appropriate spectral characteristics for agriculture monitoring and a rapid revisit (5-day repeat) capability.

The 56 meterspatial resolution, though coarser than Landsat’s 30-meters is sufficient for the accurate identification of large homogenous crop fields (NASS, 2006).

Additionally, the full swath width of 720 kilometers, when using both camera A and B acquisitions, provides an excellent opportunity for large area coverage with single day acquisitions.

AWiFS offers four spectral bands that closely resemble the most useful of Landsat 5 TM and Landsat 7 ETM+.

The sensor acquires data in the visible green, visible red, near infrared (NIR) and short wave infrared (SWIR) bands. The 5-day temporal resolution ofAWiFS is a significant improvement from the 16-day revisit of Landsat 5 TM providing the opportunity for abundant nearly cloud free imagery collected throughout the growing season. From the 2006 - 2008 growing seasons, AWiFS was collected from April 1 through the month of October. Acquisitions were excluded based on a 50% cloud cover criteria.

Fortunately, with new software a large volume of satellite imagery and ancillary data could be used in the classificationprocess.

In 2006, Moderate Resolution Imaging Spectroradiometer Data (MODIS) 16-day Normalized Difference Vegetative Index (NDVI) composites began to be used in the classification process. 

With its250 meter spatial resolution, MODIS could not replace AWiFS but was useful when collected during the late fall over specific states where the winter wheat crop was beginning to emerge.

In 2009, NASS regularly supplemented AWiFS data with Level 1T (terrain corrected) Landsat 5TM and Landsat 7 ETM+ data for CDL production, as the entire USGS Landsat Data Archive became available for public consumption, at no charge (USGS, 2005).

The Landsat data were downloaded from Glovis.

Post processing steps included converting the data from GeoTIFF toERDAS Imagine image (.img) format, reprojecting from Universal Transverse Mercator (UTM) toAlbers, resampling from 30 to 56 meters using cubic convolution (CC) resampling method, andmosaicing same day acquisitions.

During the 2009 CDL season, AWiFS experienced technical problems including an on-board data recorder failure and degraded solar panel capacity. Further, increased competition from international customers reduced the availability of AWiFS data for purchase over the U.S. by the FAS archive.

Fortunately, the freely available Landsat data were available for use as a source of supplemental imagery.

The CDL program would not have been able to meet all program deadlines, as well as, expand its scope to include the forty eight conterminous states without the use of Landsat data.

3. Software and IT infrastructure

a. Remote Sensing Classification Software: In 2004, transitioning the CDL program from research to operational status appeared to be in the realm of possibility. Changes including new imagery, ground truth, and image processing and estimation software were required.

Already in place was the FSA CLU data which provided an expansive source of agricultural ground truth and required no in-house digitization, a significant advance.

Additionally, the JAS segment boundaries could still be used as an independent data source for regression modeling. Also available were the AWiFS data which showed promise for large area coverage at a 5-day repeat cycle.

The next step was the identification of commercial remote sensing software that could perform the functions of Peditor, NASS’ original in-house remote sensing maximum classifier and estimation software. NASS evaluated ERDAS Imagine, Definiens’ eCognition and Rule quest Research’s See5decision tree software.

The remote sensing software selected needed to be affordable, efficient and accurate. See5 came highly recommended by EROS Data Center researchers and was used to produce the National Land Cover Database (NLCD) for 2001 and was found to be the most appropriate as a replacement for the Peditor maximum likelihood classifier (Homer., et al., 2004; 2007). See5 was the remote sensing classification software used by NASS since 2006 and was the primary driver of the expansion of the CDL program. The most important factor was the time requiredto produce a state wide CDL.

Once the See5 method was fully developed, an experienced analyst could produce a state wide CDL, after all preprocessing of ground truth, imagery, and ancillary data was complete, within several days. It required one to several months for the same analyst, using the Peditormethod, to produce a state wide CDL product.

The difference is in large part because See5 is able to generate a state wide CDL in one process incorporating all input data.Although Peditor was an excellent classifier, there were a number of limitations that made theclassification process more time consuming.

Peditor operated by creating multiple smaller classifications. The intersection of Landsat scenes defined “analysis districts” (AD).

A separate classification would be generated for each analysis district. Using the Peditor method, some states required as many as twelve separate analysis districts which in turn required running twelve separate classifications to produce a state wide CDL.

The individual classifications were merged to create the state wide CDL. With See5, even though by definition it classifies the intersection of inputs, there is a technique to get around this obstacle so that the entire state or region can be classified in one process.

All input data including imagery and ancillary data must be set to a specific map extent when created. Consequently, even though all of the imagery does not cover, for example the entire state of Nebraskaif all of the inputs are set to this specific map extent then See5 categorizes all land cover within this region.

This is a tremendous time saver. It takes additional time preparing the input data, but the time saved in the classification phase is significant.

Additionally, See 5 provides options which improve the quality and accuracy of the CDL products. These options include allowing for the ingestion of an abundance of satellite imagery and other non-parametric data sources; incorporating a boosting algorithm in which the classifier reviews the results multiple times to refine or “prune” the decision tree; tolerating image noise, such as clouds haze or even gaps in the imagery and generating confidence layers which corresponded to the resulting classifications.

Lastly the NLCD 2001 can be used with See5 for training on non-agricultural categories and can be combined with the agricultural training to create a complete training set for the state or region.

In 2009, NASS used AWiFS, MODIS, Landsat TM and ETM+ data to produce the CDL products. Imagery was acquired from the fall of 2008 until late September 2009. Using imagery collected over the entire growing season facilitated the separation of crop phonologies and the accurate identification of cropland.

In some instances over a particular area six or more satellite scenes acquired throughout the growing season were used to classify the land cover. This was extremely useful when attempting to identify double crops such as winter wheat followed by soybeans or crops with similar phonologies.

Peditor could only ingest a maximum of two scenes of a study area. This was a significant limitation. Non parametric data sets such as the USGS Digital Elevation Model (DEM), USGS percent canopy layer, and USGS percent impervious layer were used from 2007 – 2009 to help identify non-agricultural categories and separate them from crops.

The DEM is most useful in regions with significant topographic variation. Further, crops are most often grown in areas of low topographic relief.

For example in Mississippi, Louisiana and Missouri a significant percentage of the agriculture grown in the region is located in the low lying portion of the delta.

The percent canopy layers help identify the forested areas and the percent impervious layers helps identify urban infrastructure. These raster layers could not be used with Peditor.

Boosting or bagging, in which the classifier reviews the results multiple times to refine or “prune” the decision tree, was available with See5.

This was shown to improve accuracy in the literature (Quinlan, J., 1996). In 2009, ten boosts were generally run to refine the CDL classification. Boosting was not available with Peditor. The NLCD 2001 is currently used for training for non-agricultural categories.

The NLCD 2001was released in 2006 at which time; NASS began using it for non-agricultural sampling. When using editor, an analyst would have to manually create non-agricultural ground truth. “Extra signatures “were created for clouds, water, grass, trees, wetlands and many other non-agricultural categories, a very time consuming process. Additionally, these “extra signatures’ were created for each individual classification or analysis district with Peditor.

A tremendous advantage of See5 and improvement in operational efficiency was its tolerance of image noise such as clouds, haze and the scan gaps in the Landsat 7 ETM+ data. As long as there was an abundance of clear imagery overlaying the same location as the image noise, the software seemingly ignored the bad data.

When using Peditor, “extra signatures” would have to be created for all analysis districts ADs in which clouds were evident.

b. RSP to ESRI ArcGIS: Starting in 2006 when the FSA CLU data became the primary source of ground truth for the CDL program, the switch was made from Remote Sensing Program (RSP) to ESRI’s ArcGIS software. ArcGIS was the clear choice as USDA has an enterprise software license and many staff member strained in its use.

The preparation of the FSA CLU data was dramatically more efficient when using ArcGIS than it was with RSP. Models were written in ArcGIS to merge the original county FSA CLU shape files into statewide shape files.

The shape files were then “cleaned, projected to Albers Conical Area (Albers) and buffered inward 30.0 – 56.0 meters.

All of these steps, which were relatively time consuming for 48states, were completed on the CLU polygon data in 2009, prior to the crop season. All of these processes could not be performed with RSP which was primarily used for digitizing and editing crop attribute information.

The JAS segment data required approximately one month, during the crop season, for digitizing in the FOs and two weeks for editing by a CDL analyst.

Once the FSA CLU polygons were linked to the FSA 578 attribute data, ArcGIS models were used to exclude non matching CLUs, separate CLUs into training and validation data sets, and rasterize the shape files for use in See5. The ArcGIS models dramatically improved the efficiency of the process whereby the most current ground truth could be used prior to in season deadlines.

ESRI’s ArcGIS was an important contributor improving the efficiency and quantity of the ground truth available for use in producing CDLs.

3. Peditor to SAS: From 1997 to 2005, Peditor performed all of the functions of both a remote sensing classification and estimation software. Once the decision was made to transition from Peditor to See5,new estimation software was required.

SAS was selected as it was widely used within NASS and had the statistical analysis capabilities that NASS required.

In 2006, the regression estimator in Peditor was well developed and documented (Day, 2002). Consequently, the identical programs written and run in Peditor were transitioned to SAS. By 2007, SAS was able to increase the efficiency of estimation modeling and help transition the CDL program from research to operational status.

One of the most important advantages of SAS was the ability to interactively review the regression analysis results in IML Workshop (called Stat Studio inSAS 9.2).

This made the process of removing outliers and rerunning the regression modeling less labor intensive for statisticians.

Second, the original format of JAS segment data was formatted in SAS and made the data easier to use.

Third, results tables in SAS were output in .pdf files and Excel files that were easier for NASS headquarters statisticians to import and analyze than the .ascii tables generated in Peditor. Another important advantage that occurred during the transition from Peditor to See5 and SAS was the ability to run estimates for the entire state at one time.

 

Read: Mapping Unit: Meaning, Types and Facts

Conclusion on what is Remote Sensing? Understanding Remote Sensing

The advancement of remote sensing technology is key in conducting efficient soil surveys and soil mapping. Recent technological advances in satellite remote sensing have helped to overcome the limitation of conventional soil survey, thereby providing a new outlook for soil survey and mapping. Remote sensing has proved to be an important part of soil survey and mapping.

Soil properties that have been measured using remote sensing approaches include mineralogy, texture, soil iron, soil moisture, soil organic carbon, soil salinity and carbonate content.

Optical remote sensing helps in the mapping of properties like land cover, land type, vegetation and soil moisture.

Thermal infrared remote sensing is commonly used to estimate moisture and salinity. Visual image interpretation technique helps in the identification and mapping of soil elements like land type, vegetation, land use, slope and relief.

Microwave remote sensing is a new and effective technique for mapping of soil moisture and salinity which is being commonly used today.

Hyperspectral remote sensing is another recent method which is applied in soil salinity mapping as well as identification and mapping of minerals in the soil.