Hello Reza, I recommend you:

Hartemink, A.E., McBratney, A.B. & Mendonca-Santos, M.L. (eds), 2008.

Digital Soil Mapping with Limited Data. Springer, Berlin, Germany.

and 

Boettinger, J.L., Howell, D.W., Moore, A.C., Hartemink, A.E. &

Kienast-Brown, S. 2010. Digital Soil Mapping: Bridging Research,

Production, and Environmental Application. Springer-Verlag, Dordrecht.

The soil classification from satellite remote sensing data it is possible, but the accuracy of estimation depends on satellite's sensor, and in particular it is affected by spatial and spectral characteristics of sensor...but not only.

I deal with the estimation of quantitive soil properties (e.g soc and clay content) from hyperspectral remote sensors (EO1-Hyperion, Chris-Proba). If you are interested you can consult my articles:

 http://www.sciencedirect.com/science/article/pii/S0016706112004338

https://www.soils.org/publications/vzj/abstracts/12/4/vzj2012.0201

http://onlinelibrary.wiley.com/doi/10.1111/ejss.12202/abstract;jsessionid=C31535EB88EE03F74CE5C8A223797608.f04t01

Hello,

Yes, it is possible to discriminate soil types using satelllte imagery. You can search for free Landsat data that are useful for land cover/land use identification in medium spatial resolution using this site:  http://earthexplorer.usgs.gov/ 

Hi,

You can discriminate soil types using satellite imagey. You can search for free of charge datasets from Landsat, useful for medium spatial resolution analyses though this link:

http://earthexplorer.usgs.gov/

Hi

 One way that you can identify Soil Type perfome by using classification image.

but this way need to you have a General Recognition (from your study area) to obtain a good result.

It is possible. We have done this for the Brazilian government at the USGS. Forget remote sensing and image classification. This is how we did it, we used clustered the data, segregate it, and to cleaned it. then we used supervised machine learning algorithm to predict each cluster on the entire set. Because the paper is still in review, I can not be more specific. But this should help you get a start on things.

Depending on what you mean specifically by soil type, and what you mean specifically by identify, and it also depends on the scale of identification.

Visit these links: http://www.nrcs.usda.gov/wps/portal/nrcs/detail/soils/use/?cid=nrcs142p2_054013

http://www.isric.org/

If you want to look deeper into the issue, look at pedotransfer functions that relate environmental variables to soil type or soil properties.

Hello Reza, I recommend you:

Hartemink, A.E., McBratney, A.B. & Mendonca-Santos, M.L. (eds), 2008.

Digital Soil Mapping with Limited Data. Springer, Berlin, Germany.

and 

Boettinger, J.L., Howell, D.W., Moore, A.C., Hartemink, A.E. &

Kienast-Brown, S. 2010. Digital Soil Mapping: Bridging Research,

Production, and Environmental Application. Springer-Verlag, Dordrecht.

The soil classification from satellite remote sensing data it is possible, but the accuracy of estimation depends on satellite's sensor, and in particular it is affected by spatial and spectral characteristics of sensor...but not only.

I deal with the estimation of quantitive soil properties (e.g soc and clay content) from hyperspectral remote sensors (EO1-Hyperion, Chris-Proba). If you are interested you can consult my articles:

 http://www.sciencedirect.com/science/article/pii/S0016706112004338

https://www.soils.org/publications/vzj/abstracts/12/4/vzj2012.0201

http://onlinelibrary.wiley.com/doi/10.1111/ejss.12202/abstract;jsessionid=C31535EB88EE03F74CE5C8A223797608.f04t01

Thank you friends.

Hello,

Soil type identification remains a matter of digging, drilling, describing (eventually taking samples), and taking into account the landscape in order to understand your soil profile. Remote sensing can help to classify and spatialise your ground truth. As has been said in former replies, remote sensing can also be used to characterise some soil properties (which is not the same as soil type identification). So I would say that remote sensing can help, but relying only on remote sensing is dangerous.

Finally, the reply of Stavros Kolios seems to confuse land use and soil type. Careful!

Please clarify your question. Soil type means do you mean to say, black soil, red soil, etc??? When through satellite images, it is possible to identify crop and land with the colour (FCC), soil type (based on colour) will also be possible.

I would like to emphasize and build on some of the earlier comments. The definition of soil type plays an important role because the more general the classification (the wider the ranges of defining soil properties), the easier it is to make a correct prediction. The importance of field sampling cannot be over emphasized. The more field data you have, the better the models can be calibrated; plus the field experience can bring light to issues that are not always obvious in the data.

In regards to available covariates, it can be beneficial to use some type of data mining to identify useful predictors. This can lead to some helpful surprises (e.g., multi-temporal spectral data of vegetation can indicate a lot about subsoil conditions). I've had some good success with Cubist because it does both covariate selection and produces models that are simple enough to be interpreted in the context of field knowledge. In addition to the articles already mentioned, I've attached two more recent ones at the landscape scale that could be helpful. One is a nice example of soil mapping with limited resources in Brazil (relying primarily on the coarse remote sensing of elevation by SRTM). The second is an example of using Cubist and the potential of getting more from available covariates by thinking about multiple scales.

Article A Technique for Low Cost Soil Mapping and Validation Using E...

Article Impact of multi-scale predictor selection for modeling soil properties

It is pretty much possible using Microwave data. Use RISAT1 or RISAT2 data.

Lagacherie, P., McBratney, A.B., Voltz, M., 2007. Digital Soil Mapping: An Introductory Perspective. Developments in Soil Science, 31. Elsevier, Amsterdam.

Boettinger, Janis L., David Howell, Amanda Moore, Alfred E. Hartemink, and Suzann Kienast-Brown. Digital Soil Mapping. Springer, 2010.

As matter of fact, vis–NIR spectra holds important information about soil. Once relationships between soil properties and soil reflectance have been established, the spectral library become potentially useful and effective analytical tool for surveying and digital soil mapping.

This relevance of vis-NIR spectra can be taken advantage for modeling and mapping soil at the landscape scale. Dematte et al. (2004) elaborate approach using a spectral reflectance-based strategy to assist soil surveys. Organic matter, total iron, texture, and mineralogy were the soil attributes identified as most influential on reflectance features and intensity. Soil map and number of classes detected from soil spectra evaluation closely resembled those detected from conventional analytical soil survey methods.

The detection and description of soil variation directly from vis–NIR spectra is an attractive strategy, as neither spectral nor reference libraries have to be created for calibration purposes. Nevertheless, when it comes to soil classification or surveys, spectral clusters have to be related to conventionally estimated soil units and reliability has to be validated. However, this may be accomplished with much less reference sampling than true calibration for specific soil parameters.

In my opinion, at present it's not possible, but Earth Observation (EO) satellite data can contribute with a complementary information compared to the traditional sources of data & information. I agree with Mr. van Dijk's comments.

1. I am referring now to a population of polypedons that has to be classified. All soil classifications I am aware -  WRB of FAO,  Soil Taxonomy in US, other national classifications - are principally based on  properties that cannot be retrieved from optical or microwave data (i.e., the types of data acquired by the EO satellites).

Advanced statistical methods can be applied, as it was mentioned earlier, e.g., Mr. Miller wrote about the significant results he had obtained this way. Nevertheless, Mr. Sadeghikhoo, you must be aware of important difficulties of this approach when addressing your specific question.

Reason: soil humidity and roughness (the latter, by small shadows), as well as solar illumination and large shadows (as effect of hills, trees, bushes, even shadows of clouds, which were not properly - 100% - masked during the preprocessing of the satellite images) are some of the factors having significant impact on the reflectance values of bare soils. In my experience, the performance of operational preprocessing - geometry & radiometry, including the removal of atmospheric effects - is not high enough for this type of application, making rather difficult the interpretation of the results obtained by statistical methods.

***NB: When soils are covered, obviously the satellites do not measure them, but the vegetation / artificial features reflectance (in the optical domain) or backscaterring coefficients (in the microwave domain).

Even when soils are only partially covered - at the spatial resolution of the satellite data - there are serious problems to unmix the signals (i.e., signals of the bare soils and signals of the covered soils within the same pixel).***

2. On the other hand, various soil regimes, especially topsoil moisture evolution and temperature dynamics, can be successfully monitored by using satellite data in microwave and far infrared, respectively. This fact can help in extrapolating the point data gathered on field, so to better delineate the soil map units.

Of course, the extrapolation should be based on many other sources (traditionally used), but the EO satellites can bring unique, complementary information as regard the actual functioning of the soil cover. This is indeed a very important contribution of the EO satellites. However, at this moment, it is not included in the criteria of soil classification.

3. Besides ex-post utilization, if soil functioning multitemporal maps (moisture / temperature/... ) generated by satellite remote sensing are available before the soil survey, they can help to design the field sampling more effectively.

4. As it was mentioned, patterns recognized by data mining are significant in soil mapping (I would rather say "in soilscapes' delineation"). Satellte remotely-generated DEMs are mostly used. However, as far as I may imagine, the spatial resolution of these satellite DEMs is not sufficient high for your purpose (e.g., the free SRTM version has 90m x 90m resolution).

Anyway, I am suggesting  -in addition to using one static thematic map, such as elevation or slope- to mine in time series of satellite images, so to gather relevant information on the soil functioning. Why ? Because satellite RS can specifically exploit the time dimension, in the most possible efficient and effective way as compared to any other method.

To sum up my reply, at present satellite remote sensing technology can only be a complementary source of information for soil map units delineation, but indeed valuable. In fact, this technology is rather used - and developed - in academic frameworks, because for the time being it is not mature enough for an operational use. The good news is that the time of free high resolution EO satellite data, which has started recently, makes possible an accelerated progress, if these data are properly exploited.

This is a loaded question and many answers provided above address this. However, as hinted here in some of the answers soil types are a result of some classification schema based on some aggregation of natural features and/or properties for utilitarian purposes. RS sensed data (elevation and/or spectral) would reveal patterns that can in turn be used to identify sometimes soil types but not always. And YES, temporal data on soil moisture from RS could be used to identify soil types in broad sense, sometimes. The other issue to consider is that soil types are also identified based on features observed with depth and most of the RS sensors (at least the ones from satellites) fall short. So, there is no straight answer YES or NO. The last thing to consider is scale. I always try to ask the questions about scale and what property or thing am I trying to classify. I wish I could give you a more specific answer. Hope that helps.   

Dear you can do that regarding to the spectral signature of your soil types