PRESENTED HERE is a list of the main factors for which data are required
in order to forecast crops. The data come from different sources
(ground data, satellites), and many of them have to be collected in real
time. Consolidating this data is usually one of the main difficulties,
as they are also typically provided by various sources (e.g. different
The importance of "reference data" should not be underestimated.
Although most of them do not directly enter calculations, they are used
as a yardstick against which to compare data characterizing the current
Weather data are among the most important factors that condition the
inter-annual variability of crop yields. Depending on the prevailing
climatic conditions, limiting factors can be either rainfall (in semi-arid
areas), sunshine (in many equatorial or temperate countries). In addition,
weather affects crops indirectly, through pests and diseases.
In FAO food security programmes, rainfall is often estimated from cold
cloud duration, a satellite-based parameter - this expresses the number
of hours during which cloud top temperature drops below a certain
threshold, below which clouds are likely to produce rainfall.
Several techniques of weather monitoring are in use, for instance the
comparison of actual rainfall with the normal value. Recently, the Early
Warning System of the Southern African Development Community has begun
using El Nino-Southern Oscillation Index values (ISO) to predict the
likely outcome of the ongoing rainy season. In the graph below, the
upper (green) curve corresponds to the average behaviour of ENSO (ISO)
during wet years, while the red curve usually occurs during dry years.
By plotting the current year (black: 1996/97 season) against the two
previous years, it is possible to evaluate whether the current season
is closer to the "dry" or to the "wet" pattern.
GIF image: 6K / 600x293 pixels
Detailed information about crop stages - also known as the "crop
calendar" - plays an essential role in crop monitoring and forecasting.
This is because the effect of environmental conditions on crops depends
very much on crop growth stages. For instance, water requirements are
normally low at the initial growth stages, while they reach a maximum
just after flowering.
Information about crop stages can be obtained from different sources,
in isolation or in combination. An effective system involves field
observers, usually agricultural extension staff, using a system of
regular reporting, either by radio or by mail.
National crop monitoring systems also rely more and more on satellite
technology. Using a series of Normalized Difference Vegetation Images
(NDVI), it is possible to monitor vegetation development. Assuming that
crops follow a pattern similar to natural vegetation, planting dates -
also known as the "start of the season" - can be estimated.
The image below shows vegetation development in Ethiopia.
GIF image: 12K / 433x335 pixels
Crop reports are essential for proper crop monitoring. They are collected
from a number of different sources, such as agricultural extension staff,
district agricultural officers and meteorological observers. The area covered,
and the level of detail, varies widely. District officers usually try to
summarize the situation in a large area, which necessarily leads to some
loss of detail. On the other hand, meteorological observers might report
only on the area immediate surrounding their station.
Communicating the information is often a problem, particularly in
developing countries. District officers and meteorological stations
may have access to modern communication systems, but remote observers
often have difficulties forwarding their data to the central cop
forecasting or monitoring unit.
Crop reports usually cover two essential aspects. The first is
crop phenology, also known as crop "stage". The second
is crop condition, a parameter that is more difficult to assess.
Reporters can, for instance, rank crop condition on a scale from 1 to 5,
- indicates a very poor crop (almost no harvest expected)
- indicates a below average crop
- stands for an average crop
- above average crop condition
- a bumper harvest is expected
Typically, observers also give indications of the negative factors that
have affected crops, and of the extent of the damage they caused. Some
negative factors can usually be modelled only in extreme cases - e.g.
hail or insect attacks. On the other hand, drought and other factors
that do not physically damage crops can be modelled.
Two main satellite-based variables are used in the FAO crop forecasting
approach. The first is known as Normalized Difference Vegetation Index,
or NDVI; the second is Cold Cloud duration, or CCD.
NDVI is provided by Earth resources satellites (of the NOAA series)
and constitutes a relative and semi-quantitative measure of the living
green plant biomass. It typically presents itself as "images"
with a colour coding of the NDVI intensity, with low values corresponding
to sparse or no vegetation (ochre-brown-green), and high values indicate
dense vegetation (red-pink-purple). An example is given below for Ethiopia
in mid-July 1995.
GIF image: 19K / 512x339 pixels
For different reasons, it is difficult to use NDVI directly to estimate
crop yields. But the index is a crucial variable for monitoring, and it
is used in the FAO methodology to interpolate yields.
The second important indicator is Cold Cloud Duration (CCD). CCD is
an indicator based on meteorological satellites of the METEOSAT type.
It is a measure of the duration, in hours, in which clouds were actually
so cold (below -40 degrees C) that the likelihood that they produce rain
is very high. CCD can be used for monitoring, as it serves as a proxy for
rain. However, its greatest potential is as an aid to estimate and map out
Farm inputs and other factors
The factors which are listed below all play a part in crop yield and
production forecasting. Farm inputs include mainly fertilizers and
pesticides. In developing countries, it is sometimes difficult to
obtain accurate data for inputs, particularly if they are required
by administrative units. Among the "other factors", the
following play an important role:
- technology (e.g. mechanization, new varieties)
which is the main factor in trends affecting crop yields
- management - the decisions which are taken
by the farmer, including the choice of varieties, planting dates,
which crops to grow where, etc.
- prices, which tend to affect the area
cultivated, or the relative crop mix, more than yields
- government policies, which are often
implemented through pricing and taxing mechanisms, subsidies, etc.
Reference data are all those that are not directly used in calculations
but are nevertheless needed for meaningful crop forecasting. This includes,
to start with, a good knowledge of actual farming practices in the areas
being covered - for instance which crops are grown, and where, and how.
It also covers, particularly for agrometeorological crop forecasting,
climate and soil information
Below are three sample images from the IGADD region (Horn of Africa),
taken from FAO Agrometeorological Working Paper No. 10 (by van Velthuizen,
Verelst and Santacroce, 1995).
Length of growing season is important in crop monitoring. While the
values given in the map are averages, during a give year the actual
season may be longer or shorter, which leads to crops being unable
to complete their cycles, and usually to below average yields. The
map below shows average maize yields (only in areas where the crop
is actually grown). This, again, indicates the areas with the highest
potential, and those where unfavourable conditions are likely to cause
most harm. Reference maize yields are also important for forecasting
as one of the variables in yield function.
Finally, while crops constitute the basis of the food consumed by most
people, livestock is also a significant source of calories, particularly
in semi-arid areas. In a food security context, it is thus useful to
be well informed about the herd composition. Relevant information is
given in the map below.
Climatic databases must be assembled in all crop forecasting systems,
since climate is the dominant factor in the variability of crop yields.
(Of course, this may not be so for production, as the areas planted also
depend on economic factors and incentives linked with policies and prices).
In fact, the core of most crop forecasting models is a soil-water balance,
which requires that crop water availability and consumption are known.
Consumption depends on several variables, such as temperatures and wind
speed. This is why agrometeorological data bases usually go beyond simply
Climate conditions sometimes vary significantly even over small distances.
At a continental scale - as shown in the animated image of Africa below -
the number, length and timing of rainy and cropping seasons is extremely
FAO has assembled a major worldwide database of agroclimatic parameters,
published as the FAOCLIM CD-ROM. The database covers about 25,000 stations
and focuses on monthly averages and time series, which are essential tools
for variability analyses and risk studies.
Weather data are recorded at stations, but most analyses also cover the
area between the station. Sophisticated techniques are needed to estimate
the values on a regular grid covering the whole area under study. Gridding
tools are thus needed, such as software (by Bogaert, Mahaut & Deckers)
published in the FAO Agrometeorological Working paper series. Such techniques
can then be applied for mapping the climate, as in the animation above, or
the map below.