Agriculture 
One aspect of monitoring in agriculture involves the observation of trends and 
Changes in land use. Most critical are the problems at the interface between 
forestry and agriculture expressed by, for instance, abandoned farm lands or the 
encroachments on the tropical forest through shifting cultivation or colonization. 
There is a requirement to monitor the rapid rate at which prime agricultural land is 
being swallowed by urban development, and the degradation of lands through 
erosion, overgrazing, increased salinity or even climatic change. 
However, the most ambitious monitoring task is the development of a global 
food information and warning system which integrates historical records on climate 
and yield with up-to-date information on soil,weather and crop development. One 
approach to this is described by Park (1975) and is applied in the Large Area Crop 
Inventory Experiment (LACIE) (MacDonald, Hall and Erb, 1975). LACIE is a test of 
the feasibility of developing an agricultural crop production inventory on a global 
scale. It is an enlargement and expansion of earlier studies, including an 
international experiment in which scientists in the United States and Canada 
explored the feasibility of a spring wheat inventory obtained with the help of 
Landsat, aerial photography and ground work (Mack, Peet and Crosson, 1975). 
The two main components of a crop inventory are the estimation of crop 
areas and the estimation of yield per unit area. Remote sensing plays a role in 
both. 
The feasibility of identifying agricultural crops, other land use classes and 
some crop diseases on aerial photographs and satellite photographs and imagery has 
been well established. Indeed, the first partial successes in resource inventories by 
satellite have been in agriculture; as examples one can recall the work on the 
Phoenix and Salton Sea test sites executed in the United States with Apollo 9 
photography. 
Work in crop identification has proved the tremendous advantage of se- 
quential aerial photography or satellite imagery. Identification success rates of the 
order of 80 to 95% are not unusual with Landsat; such rates are higher than most 
that have been reliably documented in forestry projects. One reason is that most 
fields in the temperate zones of the world are large, have regular geometric 
shapes, clear boundaries and usually contain a single crop. The difficulties in 
documenting ground truth are much smaller than those encountered even in a 
relatively simple, temperate forest, where heterogeneous mixtures of species and 
gradual transitions from one type to the next can lead to uncertain classification, 
even on the ground. The high accuracies documented for agricultural crop 
identification will not be maintained in tropical or montane areas with a pattern of 
mixed cultivation, small fields and shifting cultivation. 
Once areas have been classified by crop type, estimates of yield must be 
produced and updated throughout the year. Remote sensing — aerial photography 
and satellite imagery — contribute by assisting in the mapping of soil and ecological 
units which must be recognized because they express potential differences in yield. 
The prediction of crop yields and any possibility of a disaster warning system, 
however, go far beyond the identification of crops and soils. Such a monitoring 
system must rely upon a long-term systematic record of past climatic and 
phenological data and upon immediate utilization of meteorological information 
which is compared to past norms. Meteorological conditions are the most 
important influences governing year-to-year changes in yield. Remote sensing 
again enters the scene through the vital role of satellites in assessing and 
predicting weather.