Here are a few references related to building spectral libraries for crops, classification, etc:

DOI:10.1080/01431160701241779

http://dx.doi.org/10.1016/j.biosystemseng.2011.07.002

http://dx.doi.org/10.1016/j.isprsjprs.2011.05.001

http://dx.doi.org/10.1109/IGARSS.2005.1525234

http://dx.doi.org/10.1590/S0100-204X2004000300004

Although I have not worked specifically with crop signatures, I can recommend that you look for references related to spectral mixture models, if you have not done it yet. Also, I recall some discussions arguing that other than building "pure" spectral signatures (say, obtained under lab conditions) and correcting for obvious factors like sun angle, orientation, etc, for a number of "field" studies it may be meaningful to consider how crops are actually planted.

Hope this can help.

The spectral signature of the same plant species in different part of the world can be little bit different. This may be due to different moisture contents and different content of some soil component, more of all Iron and other metallic minerals like copper...A plant can be used like a proxy of the underground soil chemical component.

Thanks, I agree. Could you suggest any paper comparing the same species in different parts of the workd?

Thanks for the question Isabella. It's an interesting one with some very interesting potential applications.

We can avoid the analisys of many parameters such as humidiy and type of soli, wheather conditions and so on by using averaging of our spectral signatures in time and in space. But actually, after such averaging, we'll lost all useful information. So, I agree with Wout. Try to reformulate your question.

Hi Isabella, I'm afraid I can't recommend any papers, but I can share some anecdotal information that might help. I was recently working on a project focusing on fusion of remote-sensing data for agricultural land classification. We were trying to incorporate some borrowed LiDAR data, but found that specific crop signatures were undetectable in the data because it was flown in the winter, after leaf-off and harvest. However, we did find that a LiDAR-based DEM tended to complement an NDVI layer quite well for differentiating between farming practices used on a land parcel - for example, linear ditching tended to indicate hay land, while the highly heterogeneous areas tended to be pasture. So while the crops themselves might not be differentiable by this method, high-resolution LiDAR data (if available) might provide insight into differences in farming techniques/practices in different parts of the world.

Perhaps a little tangential to your question (not spectral character specifically), but maybe food for thought?

Here are a few references related to building spectral libraries for crops, classification, etc:

DOI:10.1080/01431160701241779

http://dx.doi.org/10.1016/j.biosystemseng.2011.07.002

http://dx.doi.org/10.1016/j.isprsjprs.2011.05.001

http://dx.doi.org/10.1109/IGARSS.2005.1525234

http://dx.doi.org/10.1590/S0100-204X2004000300004

Although I have not worked specifically with crop signatures, I can recommend that you look for references related to spectral mixture models, if you have not done it yet. Also, I recall some discussions arguing that other than building "pure" spectral signatures (say, obtained under lab conditions) and correcting for obvious factors like sun angle, orientation, etc, for a number of "field" studies it may be meaningful to consider how crops are actually planted.

Hope this can help.

The question is a bit too broad. In general the spectral properties of the same crop is similar regardless of location (e.g. same canopy architecture, growth). There are other factors that do impact specific regions in the electromagnetic spectrum (e.g. soil background, stress); however, it has been shown that various biophysical parameters can be estimated for multiple crops across space within an acceptable range of error. If you want a very accurate estimate you will have to 'train' the models for the site; however, it is possible to get fairly accurate 'crude' estimates using a regional or globally calibrated model.

This paper shows how Landsat data can be calibrated in Nebraska to estimate gross primary production (GPP) in Iowa, Illinois, and Minnesota, USA.

http://www.calmit.unl.edu/people/agitelson2/pdf/2012/2012-RSE-GPP-Landsat.pdf

The PROSPECT+SAIL models use radiative transfer modeling to estimate various crop spectral reflectance and have been utilized to identify vegetation indices that are minimally sensitive to problems that arise in developing global models (e.g. soil background).

http://www.sciencedirect.com/science/article/pii/S0034425709000765

While not exactly related to your original question. I have also examined vegetation indices for estimating green leaf area index in my PhD dissertation for four different crops in two different locations. Maize and soybean in Nebraska and wheat and potato in Israel. I am working on getting it published in peer-review; however, it can be accessed now here (Chapter 5):

http://digitalcommons.unl.edu/natresdiss/70/

I have compared the spectral reflectance of wheat leaves from different parts of the world and found no differences in shape. However the effect of leaf biochemicals and structure on the reflectance is undeniable. currently I am working on wheat rust.

I agree with Drs. Mobasheri and Verhoef. There is probably more internal variation within a site spectrally than there are across space due to small variations in micro-topography, nutrient deficiency, etc.

Your question is too general. I would suggest to use the already existing spectral signiture for your model and your area of intrest and emphesize on validation and accuracy assessment of your results.

As good says, Wout Verhoef the spectral signatures of vegetation canopies depend on a large number of properties of the leaves, the stems, other canopy components and the background soil, and I would add the physiological and water status. The illumination and viewing conditions play a role, but they can be normalized. I also think that should make better their restlessness, such a way as being able to help you better.

We have compared a seedlings, young and mature palms spectra (the small ones were induced by disease and the young and mature palms were naturally infected by disease. So the results shows a significant different of the spectra between the seedlings, young and mature palms. Age counts in determining the spectra.

Using spectral signature of a crop is the best way to characterize it.

Maybe this attachment could help to you.

In general, the spectral curve of the crop/ plant cnopy in general is same. By using contact probe and leaf clip for measuring the signature then you may see the differences. The absorption depth & position in different parts of the curve ie. pigments, cell structure and water content enhance the differences of the same plant in different growth conditions. Currently I am using the same method for measuring the growth rate and phnelogy of wheat, sugercane etc.

Hope this could help you. Waiting for reply.

There might be some similarities but not 100% accurate.

the spectral response of a target is the response of this target in different wavelengths and in the case of cultures, the spectral response is characterized by a strong reflectance in the near infrared, which is due to the chlorophyll contained in the different cultures. However, small differences appear in different parts of the world and this is why some softwares provide models of spectral signatures always specifying the corresponding location. the spectral response of a target is the response of this target in different wavelengths and in the case of cultures, the spectral response is characterized by a strong reflectance in the near infrared, which is due to the chlorophyll contained in the different cultures. However, small differences appear in different parts of the world and this is why some softwares provide models of spectral signatures always specifying the corresponding location.

Reflectance in the NIR is NOT due to chlorophyll but rather the scattering of light by the canopy structure. The biggest driver of differences in canopy spectral response across regions is due to differences in background (e.g. soil) and possibly to differences in solar illumination and view angles due to the satellite sensors used. That said, there is more internal variation in a field (for the same crop in the same growth stage) due to other extraneous factors (e.g. disease, water stress, etc.). These factors impact the chlorophyll, other pigment concentrations (carotenoids), and canopy structure (e.g. leaf curling).

I believe the reflectance in the NIR is not due to the canopy structure and is totally due to the leaf structure and probably due to the soil underneath. Of course all other factors mentioned by Anthony is correct.

Reza,

You are right, the leaf is the major driver of scattering and when I made the statement I was primarily thinking of the canopy as a 'big leaf'. The combination of multiple layers of leaves (i.e. the canopy) does increase the scattering more so than just a single leaf. The 'big leaf' model does have its own problems. Here is a paper describing the modeling of the canopy as a 'big leaf' and describes some problems with it.

http://www.cnr.berkeley.edu/biometlab/espm228/Friend%202001%20big%20leaf%20models.pdf

Here is a lecture from Berkeley providing some insight into the 'big leaf' model.

http://nature.berkeley.edu/biometlab/espm228/Lecture%208%20Notes%20ESPM%20228%20Integrating%20and%20scaling%20information%20from%20leaf%20to%20canopy%20scales,%20part%202%20big%20leaf%20models.pdf

Dear Isabella,

Although Price (1994) reported that there is no specific spectral signatures for certain crops, Rao (2011) established the existence of unique spectral reflectance pattern for certain vegetation species.

He used temporal and spatial transfer of crop signatures to prove his methodology.

Mr. Diogenes Alves has given reference of the paper that i mention.

I suggest you that uniqueness of spectral signaturesis more dependent on the phenological status of the crop and it may very season to season.

In lab condition, as it is a controlled environment the spectral signatures of same crop from different may look similar.

I field condition, the controlling factors are many which determines the signature.

In agricultural crops, the canopy structure may be similar in all fields. So similarity of reflectance can be expected if phenological status of two areas are similar.

But in other types of vegetation, say wetlands or forests, the canopy structure will be different. It obviously affects the output signature. So location specific signature will be crucial in this case.

Thank you.

The key point is spectral signatures are not absolute for any crop/vegetation, because of the spectral properties varies from place to place and time to time, also they are scale dependent. One might find the relative response for different bands may be similar, and that is why we see vegetation are in the gradient of red in FCC image. Believe, we all here understand that, for example if there is leaf, it doesn't mean that there will be high response in IR, rather the amount of green chlorophyll is the issue compare to the others (Example, Water).

Regards to all,

We have some experience comparing spectral signatures of different crop types from a temporal point of view. Perhaps it can be useful for you the following references:

http://www.sciencedirect.com/science/article/pii/S0140196313001316

http://www.tandfonline.com/doi/abs/10.1080/01431160701408444

Most of the intial replies in this thread were inclined to the thinking that there is a unique signature for every crop. Our experience in India is that it is variant across latitudes,, plant phenology, plant pathology for the same crop. In other words, a crop would be having not one unique signature, but several unique aignatures depending upon other variables.

One factor which is mentioned here which we have not considered yet, is the interference of soil reflectance on leaf reflectance. We are planning to expand our research to cover this and other variables too. for better and more more accurate measurement,

Atmospheric corrections too deserve more efforts at better calibration - For example, in Rabi season, as the winter crop in india is called,, atmosphere in northern India is vitiated with higher degree of pollution compared to the atmosphere in eastern UP and North Bengal, . Pollution particles (physical as well as chemical particle composition of air) will differently defract the incoming solar radiation and hence potentially different reflectance and crop signatures even along the same latitude. This too is in the agenda of our research over the next 6 months.