Global Crop Estimates Via Satellite

The Reagan Administration's plans to sell the Landsat and weather satellites could affect an extremely wide range of activities—among them, information supplied to the World Meteorological Organization, research on abnormal temperatures in the Pacific Ocean (such as the El Niño phenomenon that destroys fisheries along the coast of South America), data on wind patterns that affect climate changes, measures of major vegetative changes, and assessments of global crops. This article looks briefly at the last of these—the use of remote sensing in estimating global food supplies.

The United States has five civilian satellites, all operated by the National Oceanic and Atmospheric Administration.

Two are in a geostationary orbit above the equator longitude 75° and 135° west and collect visible and infrared images every thirty minutes for the portion of the earth within their orbit. Their main function is to aid weather forecasters in tracking meteorological systems, but they also provide data on wind, sea surface temperatures, and snow cover. Their observations are sent over high-speed links to central facilities and stored in a computer for immediate processing and distribution. Two other satellites are in orbit over the North and South Poles and cover the globe four times a day. They provide atmospheric temperature and moisture profiles and weather photographs, and measure sea surface temperature.

The fifth satellite is the most current of the Landsat series, which produces digital images of the earth's surface—forests, deserts, and mountains. Landsat data are relayed to ground stations throughout the world and, pending installation of a satellite relay system, are then sent back to the United States.

The possibilities for using data collected by all these satellites to supplement or replace ground-based observations on agricultural conditions are being explored under a cooperative interagency effort called AgRISTARS. The following sections describe some of the techniques used or being developed under the program to assess global crop production.

Assessment tools

The Foreign Agricultural Service (FAS) of the U.S. Department of Agriculture, which monitors and forecasts the production of major crops in all areas of the world, needs accurate estimates of acreage and correct information on production inputs (seed, fertilizer, pesticides, and the like) and the condition of the crop during the growing season. In 1978 the FAS formed a new division to employ satellite technology as another source of information. While it has responsibility for all countries of the world, interest for this new unit primarily is in the USSR, Brazil, Argentina, Australia, Mexico, China, and India. It is in evaluating such large land areas as these that remote sensing is particularly helpful. The crops monitored are wheat, corn, soybeans, and rice.

Many sources of information are used to monitor world crop conditions—embassy attacks, press reports, foreign government publications, and trade and weather reports, among others. In the effort to integrate satellite data into existing information systems, meteorologists, plant researchers, mathematical modelers, data analysts, computer programmers, and engineers gather and analyze information from data banks. Daily weather data from all over the world are fed into computers and checked for conditions that might affect crops—sharp changes in temperature, precipitation, or winds. Then a mathematical model using meteorological information, historic agricultural statistics, and data on soils is run for the area and crop in question to see if, indeed, a problem exists. If the model indicates trouble, satellite data are analyzed for further information. In addition, other questions must be considered, such as: When was the crop planted? Did it mature earlier than usual? Have cultural practices in the area changed? All of these are pertinent in evaluating crop condition.

This oversimplifies a complex process, but it conveys some idea of the scope of the operation. In the past year, a number of tools using remotely sensed data have been refined to the point of being useful in making these assessments, while others are in an advanced experimental stage. They include alarm models, empirical measures of soil moisture, crop water and vegetative indexes, and automated extraction of crop information.

Alarm models

It now is possible to anticipate by several months potentially damaging conditions for wheat, corn, sorghum, sugar beets, and soybeans. The alarm models provide information on stored soil water, crop growth stage at risk, and hazardous and optimum moisture and temperature conditions.

Winter kill of wheat can be predicted in this way. First, degree of hardiness (frost resistance) is calculated through several stages. Next, snow depth is calculated from data supplied by the World Meteorological Organization, and then the percentage of potential winter kill of the plant population in a specific area is calculated. If the model reports more than one day of potential killing conditions, it is a signal to monitor the wheat areas via satellite imagery when the winter wheat resumes its growth in the spring.

Soil moisture models

Models that predict plant stress, crop yield, and watershed runoff all need information on soil moisture. Current methods of measuring soil moisture range from simply weighing a soil sample, drying it in an oven, and then calculating the water content from the weight and density difference, to sophisticated systems using nuclear, electromagnetic, or tensiometric techniques (the latter measure the energy with which water is held by the soil).

Two recently developed experimental methods that hold promise for wide-scale future application are nuclear-magnetic resonance sensors and remote sensing techniques, especially microwave measurements.

In the nuclear-magnetic resonance (NMR) method, a large magnetic field is created in the soil, causing the hydrogen atoms to line up on an axis. A radio pulse turns the atoms 180 degrees on the axis. By measuring the spin echo between pulses, it is possible to calculate the amount of hydrogen in the soil and thus the amount of moisture. The NMR instrument itself can be put on a tractor and pulled across a field.

While NMR is still basically a research tool—scientists are just beginning to understand how microwaves can be used to measure soil moisture—the data obtained with NMR can be correlated with satellite-mounted radar to develop an index of soil moisture for large areas.

Electromagnetic energy reflected and emitted from the soil surface is slowed down by moisture and thus can be measured by remote sensing. Despite its operational failure after only three months, the Seasat satellite was remarkably sensitive in detecting small increases in soil moisture. The focus of current research is to find out what an earth-orbiting satellite to detect moisture should look like. By 1986, the information required to design such a system should be available.

Although the number of people working on soil moisture is fairly small—a handful of researchers at the Agricultural Research Service, the Soil Conservation Service, the National Aeronautics and Space Administration (NASA), and co-operating universities—the potential value of their work is large. Applied on a regional and global scale, accurate information on soil moisture can be used to predict stream flows, to indicate how much water is available for irrigation, to anticipate gross soil erosion, and to forecast crop condition and drought.

Crop water and vegetation indexes

The crop water index, developed last year, is based on the difference between plant canopy and ambient air temperature and vapor pressure deficit. It is still at the research stage but appears to be successful in predicting drought stress for cotton, wheat, and alfalfa.

Vegetative index numbers (VINS) are being developed from satellite-collected data for selected areas of the world, which have been divided into a series of grids. Data for each grid include political entities, land resource areas, major soil types, and major crops with associated crop calendars. In addition, there are meteorological data for each grid—maximum and minimum temperature, daily precipitation, evapotranspiration, solar radiation, and snow cover. These vegetative indexes indicate changes in greenness throughout the growing season and are used to develop curves that indicate changes in normal growth patterns.

If accurate vegetative indexes can be obtained for rangeland, it may be possible to use these lands as an indicator of soil moisture and stress for adjacent crops, since the years of high rangeland VINS are also the years of maximum yields of wheat and corn in nearby areas.

Automated extraction of crop information

An automated method of assessing crops has been successfully tested in the United States over the past year. A profile of crop growth, or greenness (a crop temporal profile), has been developed that makes it possible to interpret satellite data in terms of such growth indicators as date of emergence, peak greenness, length of growing season, and stage of maturity.

These can be related to crop types, all profiles have been developed and successfully tested for corn and soybeans in key growing regions in the Corn Belt and the Mississippi Delta over a three-year period. Once the computer is initially trained on crop data from sample segments, no further human action is needed. The next step for the technology is testing on foreign corn and soybean crops.

Condition assessment

These tools, as well as many others, are—or may be—used to assess the condition of a crop. Once this is known, satellite data can be compared with an earlier or base year, and the change in yield or production potential estimated.

Of course, some uncertainty is inherent in the operation of all models and a certain amount of judgment is involved in all analyses, no matter how objective the data. Nevertheless, over the past four years war, FAS has achieved an accuracy of ± 4 percent in estimating crop production in selected areas of the world.

These advances in forecasting crops are encouraging and make an important contribution to a critical activity. The question is how much further such research can go. Certainly if satellite information becomes private property, access will become a question of price, and presumably even the supply of some information will depend on the demand for it. Also, it is not clear what would happen to the data banks and computer archives—an integral part of the climate information system—that have been assembled with some effort and cost.

Then there is the question of funds. In past years, NASA has supplied nearly half the budget for AgRISTARS, but this may not continue. Although some funding will be extended into fiscal year 1984, there will be no appropriated funds for that year and it is not yet apparent what that will mean for research programs.