NORMALIZED DIFFERENCE OF THE VEGETATION INDEX [NDVI] |
Definition: The normalized difference of the vegetation index [NDVI] is a non-linear transformation of the visible (red) and near-infrared bands of satellite information. NDVI is defined as the difference between the visible (red) and near-infrared (nir) bands, over their sum. The NDVI is an alternative measure of vegetation amount and condition. It is associated with vegetation canopy characteristics such as biomass, leaf area index and percentage of vegetation cover.
NDVI = nir - red / nir + red
For vegetation monitoring, the NDVI obtained by the combination of Channels 1 (0.54-0.68 mm) and 2 (0.73-1.10 mm) [(Ch2-Ch1)/Ch2+Ch1)], visible and near infrared respectively of AVHRR data, is commonly used. The NDVI is representative of plant assimilation condition and of its photosynthetic apparatus capacity and biomass concentration (Groten, 1993; Loveland et al., 1991). In particular vegetation index dynamics in time are correlated with the Canopy Leaf Index (LAI) and other functional variables (Cihlar et al. 1991). These variables are strongly conditioned by the behavior of precipitation, temperature and daily radiation of the observed area (Davenport et al., 1993). Vegetation index therefore is representative of plants' photosynthetic efficiency, and it is time varying due to changes in meteorological and environmental parameters. The NDVI values range from -1 to +1 (pixel values 0-255).
The AVHRR- data is particularly suited to
monitoring seasonal and inter-annual changes in land cover/land use because of its low
cost and temporal and spatial characteristics. There have been a number of studies which
have directly linked AVHRR-NDVI to plant phenology (DeFries 1995; Reed et
al., 1994). For instance, the number of periods when the NDVI exceeded a
threshold might indicate the number of growing seasons, the time integrated NDVI
might indicate gross primary production and the length of the period when NDVI exceeded a
threshold might indicate the length of the growing season. Seasonal and inter-annual
variations can be derived form multi-temporal series of NDVI that can be
associated with other ecological variables (Mora and Iverson 1995).
The NDVI is also calculated from LANDSAT-TM information by using the combinations of bands 3 (0.63-0.69 mm) and 4 (0.76-0.90 mm) [(B4-B3)/(B4+B3)]. Healthy vegetation will have a high NDVI value. Bare soil and rock reflect similar levels of near-infrared and red and so will have NDVI values near zero. Clouds, water, and snow are the opposite of vegetation in that they reflect more visible energy than infrared energy, and so they yield negative NDVI values.
REFERENCES
Definition: Albedo is defined as the ratio of the reflected to the incident solar radiation on a surface. Surface albedo is often calculated for clear-sky conditions and it is expressed as percentage. Measurements of reflected radiation in several portions of the electromagnetic radiation made by several polar-orbiting and geostationary satellites can be used to estimate surface albedo. Satellites with a large areal coverage such as the AVHRR can provide the necessary information to analyze regional patterns of spatial distribution [Gutman et al. 1989a; Gutman et al. 1989b]. {See a experimental map of surface albedo for the study area}
The Surface albedo that is calculated from information of the AVHRR sensor is often referred as "broadband surface albedo". Broadband surface albedo is calculated from spectral albedos by a narrow-to-broadband conversion using a linear combination of the individual isotropic albedos of the visible (channel 1) and near-infrared (channel 2) bands [Wydick et al 1987] as:
d = l + b1 * a1 + b2 * a2
Where: a1 and a2 denote the visible and near infrared observed albedos, respectively; and l, b1, and b2 are empirically derived coefficients [1]. Observed albedos are calculated assuming that the radiation field is isotropic and the intensity is equal to the filtered radiance detected by the satellite sensor [Gutman 1988], as:
ai = ri / m
where ai is the isotropic albedo for each AVHRR channel; ri is the reflectance factor for each AVHRR band, and m is the cosine of the solar zenith angle.
1Summary of empirical coefficients (b1, b2, and l) used to estimate surface albedo (d) from AVHRR data.| Source | b 1 |
b 2 |
l |
Brest and Goward |
0.526 |
0.418 (veg) 0.474 (soils) |
0 |
Wydick et al (1987) |
0.360 |
0.730 |
-0.7 |
He et al. (1987) |
0.332 |
0.678 |
0 |
Saunders (1990) |
0.5 |
0.5 |
0 |
Potdar and Narayana (1993) |
0.798 |
0.188 |
0.051 |
Hucek and Jakobowitz (1995) |
0.347 |
0.650 |
0.746 |
Valiente et al (1995) |
0.545 |
0.320 |
0.035 |
Russell et al (1997) |
0.441 |
0.670 |
0.044 |
The reflectance factor for AVHRR data can be calculated as albedo units [Gutman 1989a] as:
ri = p Li / Si
where, Li is the filtered radiance detected by the satellite sensor; and Si is the filtered solar irradiance at normal incidence at the mean earth-sun distance, and "i" denotes the AVHRR channels.
The narrow-to-broadband conversion can be affected by the amount of green vegetation present in the surface, and it can have a stronger influence in the reflectances within the two AVHRR bands than on the reflectances within other wave bands of the solar spectrum. More accurate estimates of surface albedo can be then obtained by including the effect of vegetation amount in the empirical coefficients to calculate broadband albedo.
Vegetation amount can be estimated indirectly by using the normalized difference of the vegetation index [NDVI] and then, the empirical coefficients can be calculated as (Song and Gao 1999); giving an albedo estimate calibrated (corrected) by vegetation amount:
b1 = 0.494 * NDVI2 – 0.329 * NDVI + 0.372 b2 = -1.439 * NDVI2 + 1.209 * NDVI + 0.587REFERENCES:
Definition: Leaf Area Index [LAI] is the leaf area per unit ground area. LAI is a factor that indicates how many leaf (or photosynthetically active) surfaces are in a column extended from, the ground area under the canopy diameter, up through the canopy. {See a experimental map of LAI for the study area}
LAI Modeling in GIS: LAI can be estimated from the normalized difference of the vegetation index [NDVI], because NDVI represent the relative seasonal changes in vegetation rather than vegetation amount. There is a significant relationship between NDVI and LAI. Assuming that NDVI/LAI relationship is linear (Wiegand C.L. 1979, Tucker 1980, Wardley and Curran 1984); and the maximum NDVI value in a season correspond to the maximum LAI of vegetation cover (Justice 1986). LAI can be inferred from NDVI as (Zhangshi and Williams 1997)
LAIi = LAImax * (NDVIi – NDVImin) / (NDVImax – NDVImin)
Where max, min and 'i' are the maximum, minimum and period values observed, respectively.
Maximum and Minimum NDVI values can be determined by multi-temporal NDVI observations from the AVHRR sensor. The formulation of LAI as a fraction of the maximum NDVI observed in a season facilitates the integration of data from different sensors. High and coarse resolution satellite observations can be combined to get more reliable estimates of LAI patterns in a landscape.
LAImax can be determined empirically by assigning different values to a land cover categories [1], and LAIi can be then obtained by combining NDVI information from different dates.
Even when a linear relationship between NDVI/LAI is often assumed, the relationship is not always linear since the vegetation indices approach a saturation level asymptotically for LAI ranging from 2 to 6, depending on the type of vegetation cover, and environmental conditions (Clevers 1989; Carlson and Ripley 1998). However, by assuming a non-linear relationship, the LAI estimates from NDVI are then highly dependent upon certain factors such as canopy geometry, leaf and soil optical properties, sun position and cloud covergae. The variation of NDVI as a function of LAI can be expressed by a modified Beer's law (Baret and Guyot 1991):
NDVI = NDVI¤ + (NDVIbs – NDVI¤) * exp (-Kndvi * LAI)
Where NDVIbs = vegetation index corresponding to that of the bare soil; NDVI¤ is the asymptotic value of NDVI when LAI tends towards infinity; and Kndvi is the coefficient that controls the slope of the relationship (extinction coefficient).
REFERENCES:
Definition: Thermal emission from the landscape "surface", including the top of the canopy for vegetated surfaces as well as other surfaces (such as bare soils). {See a experimental map of surface temperatures for the study area}
Surface temperature response is mostly determined by incoming solar radiation but also it is determined by variables associated with the substrate and atmospheric conditions, such as soil moisture, thermal inertia and albedo (Figure). Over vegetated surfaces, surface (canopy) temperature is also indirectly controlled by available water in the root zone and more directly by evapotranspiration (Carlson 1986). Thermal infrared measurements made by satellites reveal temperature patterns of surface temperatures over large spatial and temporal scales. Land surface temperatures can be estimated from the split window algorithms that use the information conveyed in the thermal infrared channels of several satellites1, but mostly AVHRR information is used (Pozo Vázquez et al 1997).
Split window coefficients will depend upon the atmospheric state and surface emissivity. Land surface temperature can be retrieved directly from sensors such as the AVHRR by using the split window channels 4 (10.8 mm) and 5 (12 mm) (Price 1984). The formulation proposed by Price (1984) for calculating LST from channels 4 and 5 is:
LSTs = T10.8 + 3.33 (T10.8 - T11.9)
An algorithm that uses a "surrogate" for surface emissivity based on the normalized difference of the vegetation index [NDVI] was presented by Kerr el al (1992) which basically results in similar estimates than for other split window algorithms, especially when compared to the PRICE algorithm. The KERR algorithm is:
LST = CTv + (1 +C) Tbs
Where: Tv = -2.4 + 3.6T4 - 2.6T5; Tbs = 3.1 + 3.1 T4 - 2.1T5; C = NDVI-NDVIbs / NDVIv - NDVIbs ; NVDIbs = NDVI for bare soil; NDVIv = NDVI max for vegetation canopies; NDVI = NDVI value for period of observation.
REFERENCES
Definition: Not all incoming solar radiation is available for biomass production and photosynthesis. The portion of the electromagnetic radiation that is used for photosynthesis (400-700 nm) is called photosynthetically active radiation [PAR], which corresponds mostly to the blue and red visible portions of the electromagnetic spectrum (green light is mostly reflected by plants).
Photosynthetic active radiation and LAI are often used to model evapotranspiration and plant productivity. Light absorption follows the seasonal changes of some combinations of plant reflectance (Hipps et al. 1983) and PAR can be linearly related to the normalized difference of the vegetation index (Asrar et al. 1984).
Due to photosynthetic activity is highly correlated with vegetation indices, both PAR absorption [aPAR]and the fraction of the surface that contains PAR [fPAR]can be indirectly estimated from NDVI. The fPAR can be estimated by using the expression (Chang and Wetzel 1991):
fPAR = 1.5 (NDVI - 0.1), NDVI = 0.547 3.2 (NDVI) - 1.08, NDVI = 0.547
In addition, the green vegetation fraction, Fg, can be derived from NDVI using a simple linear relationship with an assumption of dense vegetation (high leaf area index) (Gutman and Ignatov 1997):
Fg = (NDVIi - NDVImin) / (NDVImax - NDVImin)
where NDVImin= 0.04 and NDVImax= 0.52 can be prescribed as global constants (Gutman and Ignatov 1997) as a first approximation. The values of Fg should be restricted to be between 0 and 1.
PAR absorption can be estimated from LAI using a non-linear function (Asrar et al 1984):
aPAR = 93.5 (1.0 - exp (-0.90 * LAI)
And from NDVI (normalized difference) using the inverted relation:
NDVI = 0.087 + 0.798 * PAR
REFERENCES