(1 2004 
DIS 
. On 
1der- 
than 
ence 
the 
land 
y the 
leto 
jon of 
over 
ation 
6 as 
land 
body, 
rtion 
1 the 
ed to 
s and 
jatial 
> Me 
  
  
jah 
nd 
were 
ed on 
he 
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 
  
cloud covers, some pixels certainly has continuing cloud cover 
during the eight days. Fortunately, such cloud cover 
information is supplied with the MODIS LAI product. Figure 6 
showed the cloud coverage at every pixel location and the LAI 
estimation algorithm used. As seen in Figure 6, about 30% 
pixels of the MODIS LAI image were affected by cloud cover. 
The 5% pixels of this MODIS LAI image were produced by the 
backup algorithm based on NDVI. In particular, most of the 
pixels corresponding to the backup algorithm were forest. 
Although the MODIS LAI product at this study area was 
mainly estimated by main RT algorithm and not affected by 
cloud, the effects of the backup algorithm and cloud should be 
considered at the quality assessment of MODIS LAI over other 
sites. 
  
  
ey 
=A, 
oe 
I = 
  
  
Figure 6. MODIS LAI QA images of cloud effect (left: white- 
cloud, black- no cloud) and estimation algorithm 
(right: white- RT model, black-backup model). 
As seen in Table 2 and Figure 7, the MODIS LAI values, which 
were estimated by the backup algorithm and affected by cloud 
cover, showed large difference as compared to the reference 
LAI map. In these pixels, the MODIS LAI is lower than the 
reference LAI value. Since the portion of area of the backup 
algorithm and the cloud cover was relatively small, they might 
have not affected the overall statistics of MODIS LAL 
However, since the more than two third of the study area is 
occupied by forest, the over-estimation of MODIS LAI in 
forest shows great influence on the total image (Figure 7). In 
grass and cropland, although they were small portion, the 
MODIS LAI values were lower than the reference LAI value. 
Under-estimatior II Over-estimation 
E etf mm 
Total image [- 
    
RT model 
Back-up model | 
Cloud area 7 
Nan-cloud area : 
Forest 
Crop&Grass a 
  
0. 10. 20. 30 40.50 60. 70 80 - 90; 100 
Coverage(%) 
Figure 7. Estimation patterns of MODIS LAI for each schemes 
(Under-estimation means MODIS LAI was lower than 
reference LAI Over-estimation means MODIS LAI 
was higher than reference LAI) 
4. CONCLUSIONS 
During the last decades, there was a great amount of time and 
efforts to develop the global ecological variables by the earth 
observing system (EOS) program. The MODIS global LAI 
product is one of such variables that are now being provided. 
To use such valuable information at regional and local scales, it 
is crucial to validate the quality of the product. In this study, we 
made a simple and direct comparison between the MODIS LAI 
data and the reference LAI surface that was derived by an 
empirical approach to relate the field-measured LAI to Landsat 
ETM+ reflectance. 
Although the validation can be further expanded into larger 
areas, the preliminary results obtained from this study indicate 
that the MODIS LAI product is different from the reference 
data. At the temperate forest, the MODIS LAI estimates were 
higher than the reference LAI values during the leaf-on season 
of September. The discrepancy between the MODIS LAI and 
the reference LAI may be caused by the uncertainties in the 
input variables of MODIS LAI algorithm, effects of cloud at 
each pixels, estimation algorithm, and land cover type. In 
general, MODIS LAI values were higher than the reference 
LAI in forest and they were lower than in grass and cropland. 
Additionally, MODIS LAI data affected by backup algorithm 
and cloud showed higher value as compared RT algorithm and 
no-cloud condition. 
The validations of MODIS LAI product are being carried out at 
the several study sites throughout the world. Due to the km 
spatial resolution of the data, the validation is often restricted 
by the size of test area and the collection of accurate ground- 
truth data. Temperate forest is even more complicated than any 
other ecosystems, which include a variety of species and stand 
structure. Further study is planned to assess the quality of the 
. MODIS LAI estimate for local and regional applications. 
5. REFERENCES 
Bonan, G, 1993. Importance of leaf area index and forest type 
when estimating photosynthesis in boreal forests. Remote 
sensing of Environment, 43, pp.303-314. 
Brown, L., J. M. Chen, S.G Leblanc, and J. Cihlar, 2000. A 
Shortwave modification to the simple ratio for LAI 
retrieval in boreal forest: An image and model analysis. 
Remote sensing of Environment, 71, pp. 16-25. 
Carlson, T.N. and D.A. Reley, 1997. On the relation between 
NDVI, Fractional Vegetation Cover, and Leaf Area Index. 
Remote Sensing of Environment, 62, pp. 241-252. 
Chen, J.M. and J. Cihlar, 1996. Retrieving leaf area index of 
Boreal conifer forests using Landsat TM images. Remote 
Sensing of Environment, 55, pp. 153-162. 
Chen, J. M., PM. Rich, S.T. Gower, J.M. Norman, and 
S.Plummer, 1997. Leaf area index of boreal forests: 
Theory, techniques, and measurements. Journal of 
Geophysical Research, 102 (D24), pp.29429-29443. 
Chen, J.M., G Palic, L. Brown, 2002. Derivation and validation 
of Canada-wide coarse-resolution leaf area index maps 
using high-resolution satellite imagery and ground 
measurements. Remote Sensing of Environment, 80, 
pp.165-184. 
Cohen, W.B., T.K. Maierpserger, S.T. Gower, and D. P. Turner, 
2003. An improved strategy for regression of biophysical 
variables and Landsat ETM+ data. Remote Sensing of 
Environment, 84, pp.561-571.