Menu
STARS Project
close
STARS Project
close

Refine your search

Knowledge Portal

Vegetation indices

Normalized Difference Vegetation Index (NDVI)

The interaction of electromagnetic radiation with vegetation forms the basis of one of the most widespread applications of remote sensing in natural resource management. Green vegetation reflects different amounts of incident radiation in different portions of the electromagnetic spectrum. The most widely used portions of the electromagnetic spectrum for vegetation analysis are the red and near-infrared (NIR) ranges.

Vegetation response (i.e., the amount of light reflected) in the red band is very low due to high absorption of incident radiation (by chlorophyll — the pigment in plant leaves). On the other hand, response in the NIR is very high due to high reflection of incident radiation. In-between these two bands is the red-edge channel/range, which is equally useful in vegetation analysis. Together, these spectral ranges give a measure of the chlorophyll content in vegetation.

Stress factors such as those enumerated above are known to affect reflectance in the NIR range of the electromagnetic spectrum more than that of the red. Since the NIR portion is not visible to our eyes, RS data provide an early warning with regards to the condition of crops before it becomes evident to the human eye.

Differences in response patterns of vegetation in the red and NIR ranges of the spectrum permit the calculation of an index (NDVI) (Candiago et al., 2015; Jin and Eklundh, 2014; Liu et al., 2012), which is one of the most widely used remote sensing-based indices, mostly in vegetation studies (Equation 4.1). It is defined by the formula

Equation 4.1

where ρNIR is the reflectance in the NIR and ρRed is the reflectance in the red portions of the electromagnetic spectrum. NDVI ranges between -1 (no vegetation) and +1 (green vegetation).

When satellite images record data specifically in the red-edge portion of the spectrum, it is possible to calculate red-edge dependent vegetation indices (Peng and Gitelson, 2012; Shang et al., 2015), by considering the NIR and red-edge or red and red-edge, as alterations to Equation 4.1.

NDVI is a measure of (the level of) greenness and biomass of vegetation and for that matter of plant health. High NDVI values indicate green and healthy vegetation, while low values signify little or stressed vegetation.

Analysis of NDVI values in space (over a farm field) and time (over a time period) can, therefore, provide valuable information on crop status and health. By classifying an NDVI image into different class ranges (using appropriate thresholds), it is possible to identify portions of a field that are likely under stress from any of the above-mentioned stressors (Atzberger, 2013).


Figure 4.3 NDVI image stack for a field in Mexico (Source: STARS team, CIMMYT, Mexico)

Despite its usefulness and widespread application, the NDVI has many known limitations. These include: (1) saturation, (2) soil background, and (3) angular effects. Consequently, a number of other indices have been proposed and have been found useful in natural resource management. The Enhanced Vegetation Index (EVI), Soil Adjusted Vegetation Index (SAVI), Ratio Vegetation Index (RVI) and Perpendicular Vegetation Index (PVI) are example alternatives (Dorigo et al., 2007). When using hyperspectral images, a much wider range of vegetation indices can be calculated by virtue of the high spectral resolution, captured in many different bands. A list of possible indices (from hyperspectral data) can be reviewed here

The core technology of AgriSense-STARS for remotely sensed crop conditions in East Africa is an online portal for the automated processing of MODIS satellite image time series and the production of NDVI time series graphs. The graphs allow the detection of low and high production areas in the country. The processing chain is implemented within the GLAM East Africa portal (UMD Global Agricultural Monitoring System, customized for Tanzania and East Africa). GLAM East Africa is a user-friendly, automated portal for MODIS and precipitation time series analysis to support reporting and crop condition monitoring; GLAM East Africa is being used to track crop condition throughout the growing season.

In addition to NDVI, NDVIg was used for crop type classification and crop area estimations. NDVIg is calculated from RGB imagery and is defined as (Red-Green) / (Red + Green). We found that this index is a useful alternative for NDVI in cases where only RGB imagery is available.

Leaf Area Index (LAI)

LAI is another widely used index for agricultural management (Richter et al., 2012; Thorp et al., 2012). It is defined as the ratio of one sided leaf area per unit of ground area (m2/m2). LAI characterizes plant canopy structure and gives an idea of the amount of biomass available in a field. It is also considered a measure of crop growth and productivity.

LAI can be derived from RS data although, and unlike NDVI, it is not calculated directly from spectral bands of satellite images. Traditional in-the-field LAI measurement approaches are time-consuming, but there exist relatively easy approaches to measure LAI, including the use of a smartphone application (pocketLAI) (Confalonieri et al., 2014). It must be noted, however, that these easy approaches are often less accurate and/or capture a simplified reality (Francone et al., 2014). Within STARS, different measurement approaches were used (see Field data protocol).

In-situ measurements of LAI can be correlated with NDVI (and other vegetation indices) to derive the spatial distribution of LAI in an area of interest (Dorigo et al., 2007). This depends on the establishment of a robust relationship between the two indices (Maki and Homma, 2014; Turner et al., 1999). Since the spatial distribution of NDVI can easily be derived from RS data, the derivation of LAI is easy once a robust relationship has been established.

Similar to NDVI, LAI images of different growth stages can be used to monitor crop growth and performance. Different crop types may have different LAI at different growth stages (i.e., due essentially to different leaf structure/size). Knowledge of LAI dynamics can, thus, be an important input variable to mapping the spatial distribution of different crops.

LAI was one of several parameters measured in Tanzania. Figure 4.4 shows the development of LAI during the growing season in a farm field in Kilosa District, Tanzania.






Figure 4.4: Time series of leaf area index (LAI) measurements in maize field in Njombe District, Ikisa Village. Dates of photographs: 5 March, 1 April, 30 April 2015 (Source: STARS AgriSense team).

Related publications