CLIMATE VARIABILITY, AGRICULTURE AND REMOTE SENSING - Food Security, Food Prices and Climate Variability - page 56

spatial grid and three-hourly resolution (Huffman et al ., 2007), is a significant advance over
previous products and represents the state of the art of satellite rainfall observations. Scheel et
al . (2011) show a decreasing performance of TMPA with increasing precipitation intensity,
resulting in significant errors when calculating cumulative rainfall over a period of a month or
more in areas that receive strong precipitation events. Extreme rainfall events are short-lived
and of high intensity and are therefore very hard to capture using satellites due to lack of
coverage (the satellite is not returning information during the event) and inadequate station
data (needed to relate observations to actual rainfall that arrives at the ground) (Scheel et al .,
2011). This results in poor representation of extreme weather-related events such as flash
floods, landslides and other hydro-meteorological hazards (Scheel, 2012).
Drought monitoring and response
Using satellite-derived rainfall, FEWS NET monitors rainfall in the countries in which they
work. Figure 3.2 shows rainfall in Ethiopia for 2011 and 2012 compared to the short-term
mean. Rainfall remains a very important and reliable way of monitoring growing conditions,
and is easily related across time and space since crops have minimum growing requirements
and lengths of season (Verdin et al ., 2005). If the rainfall received is below these requirements
the likelihood of a failed crop is large even with the diversity of crop types, species and man-
agement approaches present in the developing world. It is challenging, however, to maintain
comparable rainfall data records across decades due to the need for high quality information
from the ground, as was noted above.
Satellite remote sensing of vegetation
Another way of measuring moisture availability and crop productivity is the normalized
difference vegetation index or NDVI. Vegetation indices are usually composed of red and
near-infrared radiances or reflectances (Tucker, 1979), and are one of the most widely used
remote sensing measurements (Cracknell, 2001). They are highly correlated with the
photosynthetically active biomass, chlorophyll abundance and energy absorption by plants
(reviewed in Myneni et al ., 1995) and provide a way to measure directly the impact of
moisture and temperature conditions on plant health. NDVI was first developed using hand-
held radiometers, and its relationship with aboveground plant matter was established by
correlating information from instruments to the weight of dried plant material in a grassland
ecosystem (Tucker, 1977). The use of spectral vegetation indices derived from the Advanced
Very High Resolution Radiometer (AVHRR) satellite data followed the launch of the first
operational polar orbiting satellite in 1979 (Gray and McCrary, 1981; Townshend and Tucker,
1981).
The normalized difference vegetation index (NDVI) is calculated as a ratio of the difference
of the near-infrared (NIR) and the red bands on a sensor: (NIR - Red) / (NIR + Red). Since
vegetation has high NIR reflectance but low red reflectance, vegetated areas will have higher
NDVI values compared to non-vegetated areas ( Figure 3.3 ). The longest continuous record
of global vegetation observations comes from the Advanced Very High Resolution Radiometer
(AVHRR) sensor, which has been flown on operational satellites by the US government for
more than 30 years. Because of careful calibration of the AVHRR NDVI dataset, the record
allows the current image to be subtracted from the long-term mean, enabling an assessment