Data necessary is entirely based on the kinds of questions you want to answer. What geographical regions are you looking at? Areas of high biodiversity will be more difficult to discriminate between species.

Sensors make a difference too. MODIS has great temporal resolution, but the spectral resolution is poor. Most hyperspectral data lack the spatial and temporal coverage necessary for large GIS projects.

Ground truth data is also very important. Without it, your models are at best limited to past research, at worst pure 'guess-work'.

We need more details from you about your goals for your research to help make better suggestions.

There are two important things here: 1. vegetation indices and 2. plant diversity. In physical or physiological terms vegetation indices should be associated to vegetation density or biomass, but that association depends on both the vegetation type and the sensor characteristics. Frequently the relationship between vegetation indices and vegetation biophysical parameters is not strong and any direct association can be frustrated. Vegetation indices should be seen as good variable in time rather than associated to vegetation typology identification. Plant diversity is an universe of possibilities and constrains. You have taking into account the complete definition of "diversity" here. Species? Physiognomic differences? In both cases the size of the area, the size of plant communities, the sensor characteristics, temporal aspects are quite important. There is no a unique alternative and neither a trivial solution.

Flávio Ponzoni "Frequently the relationship between vegetation indices and vegetation biophysical parameters is not strong and any direct association can be frustrated"

I am sorry I have to disagree here. There are many, many papers that show that vegetation indices (VIs) can be used to for estimating biophysical characteristics (BPC). The problem is generally in the data collection (either reflectance or ground-truth). Once a suitable protocol is developed, very strong relationships (R2 >0.9) have been found for many BPCs. You are correct in that it is very difficult to separate vegetation species. Approaches usually require multi-temporal data which examines differences in phenology between species or hyperspectral data which can discriminate between species through different spectral profiles. Another tool is to use active sensors such as LIDAR and SAR. As biodiversity increases, species identification becomes less likely. Typically in a richly diversified area, the best one can do is identifying plant types.

I am assuming you are primarily interested in terrestrial vegetation indices? If you do have interest in submerged aquatic vegetation architectural diversity, I can advise

Thanks allot. The Tirunelveli coastal area in my study area. I am very much interested to do the vegetation analysis (seasonal variation for herb, shrub and trees )of our study area. I have previous experience with the software Arc view and handling. What are the basics to run Arc view software and how to operate it? How will i get the protocol? My institute doesn't have the facility to run the Arc software, If you have more knowledge please guide me as i have to submit this work in a couple of months.

Once again thanks for such nice reply

I am fortunate to have ArcGIS provided through my university. You may try contacting ERSI to see if they have a cheaper option for your country/region.

While I do use a lot of open-source software, I don't use open-source GIS software; however, here is a link of various open source products available. Unfortunately, there isn't anything remotely close to ArcGIS in quality.

http://opensourcegis.org/

Here is a tutorial for ArcView.

http://www.globe.unh.edu/FinalGIS_Tutorial_ArcView.pdf

I suggest examining papers that are in similar topics for their methodology.

A list of vegetation indices can be found here: http://geol.hu/data/online_help/Vegetation_Indices.html

Object based image analysis is now being widely used for plant species discrimination using several vegetation indices extracted from high resolution satellite images.

I think, you have to change detection for to periods. you can use from vegetation indexes, image classifications and Visual interpretation methods. ERDAS and ArcGIS or ILWIS are suitable for this means.

Just vegetation indices alone dont work to get the plant diversity. A thorough understanding of the sensor characteristics, temporal variations of plants, seasonal changes in the study area, along with the proper choice of indices with extensive ground truth verification is needed. There is definitely not one solution for question, whichever is feasible and acceptable to the study area and the data products available should be used.

Thanks allot. i have one doubt, what is the best satellite image for vegetation analysis? for ex. PAN, LISS- III etc..

Thanks once again

Pan data will not be of much use for plant diversity mapping with a single band information, but does have a much higher resolution. LISS IV or LISS III will be definitely useful since 4 spectral bands can be played around with. A merge data product of LISS III or LISS IV with PAN will give the resolution of PAN along with the Spectral characteristics of LISS III sensor. But apart from these RISAT SAR data with dual polarisation capabilities can also be used to explore plant diversity mapping.

Vegetation indices (VIs) alone cannot discriminate plant species as they give you the amount of greenness; then, two different species can give the same VI. The best approach is to use hyperspectral narrowbands and run, e.g., discriminant analyses that can result in the best bands capable to discriminate plant species. Classification techniques, such as object-oriented classification can also help but you need to have high spatial resolution images available.

About which is the best satellite image for vegetation analysis, all depends on what your specific question is. What is the scale of your study? Small or large? Where is your study area located? Do you have already any ground information available on the vegetation? Is the area very dense vegetated, or sparsely vegetated, etc.? What is your budget? All of this is necessary in order to make a decision on a satellite product.

Thank you Sir,Maharajan Natarajan and Isabella Mariotto

Our study area covering 30 Km radius of Kudankulam Nuclear Power Project Site (approximately 1350Km2) lies between Latitudes 805’ and 8028’ of North and Longitudes 77028’ and 77057’ of East.

My study area major occupied human population, agriculture, plains and coastal dune. then this study area unprotected ecosystem.

Now i am working as unstippend research scholar. My major research work vegetation analysis and checklist of flora. but i have intrested in gis analysis.

Toposheet and erdas imagine software and ground gps data were available.

sir,

kindly request to your idea and suggestion for my part of research work.

I am very thankful for your kind attention.

Thanking you,

with regards

Again, the questions you wish to answer are important. Currently MODIS and LandSat are free to download and use. MODIS has great temporal resolution, but poor spatial resolution. LandSat has good spatial resolution, but poor temporal resolution. Both lack the red edge band. The ESA plan on launching Sentinel-2 and -3 here in the next year or two. If they make these products free, you will have fairly good spatial, temporal, and spectral resolution. There isn't an instrument that excels in all three categories. Some products do cost money. That is one reason people use UAVs for site specific applications. Once you identify your end-goals and budget, you can identify the optimal product.

It is clear that thr study area is not large, so you can use high spatial resolution images and narrow spectral bands mainly the NiR and R. You also need a detailed field data collection program inorder to conduct a supervised classification. It is important to choose a suitable time and date of imaging to optimize species spectral characteristics contrast.

i am in a similar situation here, i wish to identify extent of encroachment by sicklebush in a small nature reserve in Swaziland just next to the border to mozambique.

please advise on the methods to use, i know NDVI alone can not help as i want to differentiate between plat species.

Im assisting a university student on this so any kind of help would go a long way.

Hello. Try to use high resolution imagery e.g spot. Try to correlate several indices to diversity data measured directly in the field and decide on the index that gives you the highest coefficient of determination. The standard deviation of NDVI and NDVI itself have been shown by other scholars (e.g Skidmore, Mutowo and Murwira, Chapungu and Nhamo, etc) to show a positive relationship with species diversity in various geographical locations. Perform ground truthing please. You can also differentiate species by assessing their spectral reflectance using a spectrometer.

You can use LandSat data...