31 
CHARACTERIZING LAI SPATIAL AND TEMPORAL VARIABILITY USING A 
WAVELET APPROACH 
X. Guo al andB. C. Si b 
department of Geography, University of Saskatchewan, 9 Campus Drive, Saskatoon, SK S7N 5A5 - 
xulin.guo@usask.ca 
department of Soil Science, University of Saskatchewan, 51 Campus Drive, Saskatoon, SK S7N 5 A8 - 
bing.si@usask.ca 
KEY WORDS: Leaf Area Index, Remote Sensing, Variability, Wavelet 
ABSTRACT: 
Vegetation plays an important role in the exchange of carbon dioxide, water, and energy between the land surface and the 
atmosphere. LAI, defined as one-half the total green leaf area per unit of ground surface area, drives the within and the below 
canopy microclimate, determines canopy water interception, radiation extinction, and water and carbon gas exchange. Therefore, 
accurate LAI is a key parameter in all models describing the exchange of fluxes of energy, mass (e.g., water and C0 2 ), and 
momentum between the surface and the planetary boundary layer. Unfortunately, LAI is very difficult to quantify accurately due to 
its spatial heterogeneity and temporal dynamics. The long-term objectives of this study are 1) to improve LAI estimation accuracy 
with considerations of scale, heterogeneity (spatial, vertical, and temporal), and land cover type, 2) to develop a better approach to 
LAI parameterization for models in hydrology, climatology, and ecosystem, and 3) to investigate the effects of land cover and land 
use changes on LAI dynamics, which can cause massive hydrological change. In this paper, we aim to 1) characterize the spatial 
scale of LAI and normalized difference of vegetation index (NDVI) in a Canadian prairie using a wavelet approach based on field 
measured LAI and reflectance data, and 2) to simulate the temporal dynamics of LAI variation intra-annually with both ground 
measured LAI and satellite derived NDVI values. The study area is in St. Denis Wildlife Reserve Area, 40km east of Saskatoon, 
Saskatchewan, Canada. Results indicated that the spatial variation of LAI and NDVI is maximized at 22.5 meters and with several 
small scale variations (4.5m, 12m, and 18m). The temporal LAI dynamics indicated that the native prairie greens up in May and 
senescent in September, and the maximum growing season is in July for the Canadian prairie. 
1. INTRODUCTION 
Vegetation plays an important role in the exchange of carbon 
dioxide, water, and energy between the land surface and the 
atmosphere. LAI, defined as one-half the total green leaf area 
per unit of ground surface area (Chen & Black, 1992), drives 
the within and the below canopy microclimate, determines 
canopy water interception, radiation extinction, and water and 
carbon gas exchange. Therefore, LAI is a key parameter in all 
models describing the exchange of fluxes of energy, mass (e.g., 
water and C0 2 ), and momentum between the surface and the 
planetary boundary layer (Knyazikhin, et al., 1998). 
LAI has been selected in a broad range of models including 
vegetation (Moulin et al., 1998, Cayrol et al, 2000), 
biogeochemical (Running et al., 1999), hydrological (Andersen 
et al.., 2002), and global atmospheric circulation (Avissar & 
Chen, 1993). Some popular examples of these models are BEPS 
(Liu et al., 1997), BGC (Kimball et al., 1997), CENTURY 
(Parton et al., 1988), TEM (McGuire et al., 1997), CLASS 
(Verseghy, 1993), CHRM (Pomeroy et al., 2006), NCAR CCM 
(Chase et al, 1996), AGCMs (Krinner et al., 2005), and MM5 
(Grell, et al., 1994). Currently, these models are initiated by 
either field validation of simulated LAI or remotely sensed 
estimates of LAI (Running et al., 1999). However, many 
climate and ecosystem models are very sensitive to variation in 
LAI (Bonan, 1993) and thus rely on accurate LAI estimates. For 
example, the Global Climate Observation System (GCOS) and 
the Global Terrestrial Observation System (GTOS) requires an 
LAI accuracy of 0.2 to 1.0 for terrestrial climate modeling. 
Unfortunately, LAI is very difficult to quantify accurately, due 
to its spatial (horizontal and vertical) and temporal variability, 
as annual cycles and interannual variability interact with the 
vegetation structure, stratification and heterogeneity. 
Currently, there are three major methods for obtaining LAI 
estimations: ground measurement, remote sensing derivation, 
and hybrid approaches. The major ground-based methodologies 
employ either “direct” measures (involving destructive 
sampling, litterfall collection, or point contact sampling) or 
“indirect” methods (involving optical instruments and models). 
While accurate on a per plant or site basis, direct methods are 
time consuming and tedious (Lang et al., 1985) and destructive 
to plants. A review of the direct LAI measurement techniques is 
given in Norman and Campbell (1989). By contrast, indirect 
optical methods hold great promise because of the potential to 
obtain quick and low-cost measurements over large areas. 
However, several commercial optical instruments, including the 
LAI-2000 plant canopy analyzer (LI-COR, Lincoln, Nebraska) 
and Sunfleck Ceptometer (Decagon Devices, Pullman, 
Washington), are hindered by the complexity of natural canopy 
architecture. Most studies concluded that indirect methods 
underestimated LAI when compared with direct measurements 
(Chason et al., 1991; Comeau et al., 1998). The reported 
underestimation varies from 25% to 50% in different stands 
(Gower and Norman, 1991; Gower et al., 1999). The degree of 
error in the LAI measurement is a result of the canopy’s 
deviation from the assumption of random dispersion, which was 
1 Corresponding author.