It is important that your input data has been converted to reflectance (either TOA or surface) between 0-1.0. Is your input data within this range?

yes. I corrected my images with ENVI FLAASH and reflectance range between 0-1.
If we just substitute numbers into LAI= (3.618*EVI) - 0.118 equation we can see maximum value for LAI can be 3.5 (for max EVI +1). However, LAI can go higher than 3.5. ENVI gives values greater than 3.5. Problem is if we use the same equation in band math, the result is different. It is limited to 3.5.
I seek your guidance further.

Thank you.....!!!!

Thank you for the reply,
yes. I corrected my images with ENVI FLAASH and reflectance range between 0-1.
If we just substitute numbers into LAI= (3.618*EVI) - 0.118 equation we can see maximum value for LAI can be 3.5 (for max EVI +1). However, LAI can go higher than 3.5. ENVI gives values greater than 3.5. Problem is if we use the same equation in band math, the result is different. It is limited to 3.5.
I seek your guidance further.

Thank you.....!!!!

I think it may occur if any pixels are outside of the expected range of 0-1.0 for reflectance. For example, if I process a Landsat OLI image to reflectance using Quac, then divide the image cube by 10000, I get the following statistics:

Basic Stats Min Max Mean StdDev
Band 1 0.000000 0.614000 0.039422 0.054696
Band 2 0.000000 0.658800 0.040490 0.057899
Band 3 0.000000 0.720700 0.042444 0.062057
Band 4 0.000000 0.835200 0.045314 0.071273
Band 5 0.000000 0.836000 0.078498 0.108855
Band 6 0.000000 0.840000 0.062646 0.084091
Band 7 0.000000 0.965900 0.049391 0.068621

However, those pixels in the background that have a value of 0 can affect the results. So I create a mask of the background using the Build Mask tool and create a mask for pixels that are 0 or less. Then I use the Spectral Index tool using the reflectance image and mask, select the EVI and LAI indices, and output this result. Here are the stats for this result:
Basic Stats Min Max Mean StdDev
Band 1 -0.287533 0.836403 0.093981 0.145088 <- EVI
Band 2 0.000000 2.908107 0.288327 0.483652 <- LAI

The values are in the expected range. However, when I did not use the mask of pixels with 0 or less in the Spectral Index tool, I found a few pixels would be outliers when the EVI and LAI were calculated. Can you try creating a mask of pixels that are 0 or less in your reflectance image and use this mask when calculating spectral indices? Alternately, from the Band Math result (to get reflectance between 0-1.0) , you can also set the Data Ignore Value in the header to be 0.0. To do this, go to View Metadata->Edit Metadata and use Add to add the Data Ignore Value to the header.

The method you have given is very useful to eliminate negative values in the lower end of the LAI. However, my image is thickly vegetated. Therefore, EVI value hits +1 most of the time. Then according to the equation, my LAI value mostly limits on LAI=3.5. I can not have LAI value more than 3.5 according to the equation. We can see that if we substitute +1 in LAI= (3.618*EVI) - 0.118 equation. This only happens when I using bandmath. I got LAI more than 3.5 when using envi spectral indices tool. I am pretty sure I correctly used the bandmath. How can envi gives LAI more than 3.5, even it is not possible according to the equation. (my research needs bandmath rather than using inbuilt tools)
Thank you for your continuous help.

Can you show me the basic stats from the input image and then the calculation of the EVI and LAI from this image?

Apology for long silent period,
I research again and again but the problem remains same. I am giving here my image statistics.
Landsat image
Band Min Max
1 0.0161 0.6363
2 0.0081 0.6399
3 0.0141 0.7227
4 0.0077 0.7461
5 0.000 0.77360
6 0.000 1.000
7 0.000 1.000


EVI image statistics (using equation for EVI in band math)
min max
-0.2249 1.053

LAI image statistics (using equation for LAI in band math)
min max
-0.9318 3.6922


LAI image statistics (using ENVI inbuilt spectral indices tool for lai)
0.000 3.6922

Again the problem is,

LAI= (3.618*EVI-0.118)
if EVI= 1 (it's maximum value)

LAI=3.5 (maximum LAI value is 3.5 for any case)

How ENVI giving 3.6922.??

Please clarify me so that I go further in my research.

Thank you

In the above, you show your maximum EVI is 1.053 which is over 1.0, It seems the EVI can exceed 1.0 which would mean the LAI can exceed 3.5, yes? The equation in the documentation is based on Huete. ENVI is not limiting the range of EVI to 1.0 for the LAI equation. If you have a very dense, green canopy with a high NIR and very low R and B value, you will exceed 1.0 for EVI using the equation: Using your maximum NIR value and min Red and Blue values (assuming bands 2,4,5):

ENVI> print, 2.5*(.7736 - 0.0077) / ((.7736 + 6.0 * 0.0077 - 7.5 * 0.0081 +1.0))
1.08851

Thank you very much.

This is really helping. I used a number of indices and the EVI performed quite well.

Another thing you may check is if there is influence by cloud. In the EVI paper they state:

The sudden increase in blue and red reflectances to values slightly higher than NIR reflectance in July and August 2000 is documented in the QA results as residual cloud contamination.
...
These variations resulted in significant ‘noise’ that inhibited the utility of the VIs.