J.N.R. JEFFERS
Forestry Commission, United Kingdom
SINCE the last Congress of the International Union of Forest Research Organisations (IUFRO),1 one of the notable developments in forest research and management has been the increased use of electronic digital computers. Indeed, it is no exaggeration to say that this development has completely transformed the scientific background against which these congresses are held. For the first time in the history of man it is now possible to make an attack on some of the problems which underlie our fundamental knowledge of forestry, and many of our existing techniques have been outdated by the more powerful mathematical tools that have been made possible by the computer. These developments affect the work of all sections of IUFRO, but it is particularly appropriate that Section 25 should take and lead in promoting the wider use of the electronic computer in forest research, for it is this section whose work has been most affected, and most rendered obsolete, by the new machines.
1 Paper presented at the IUFRO Congress, Vienna, September 1961.
This paper is intended to provide a brief description of electronic digital computers, review the present situation of their application in forest research, and to make recommendations for their more effective use in the period up to the next congress.
Electronic digital computers
The electronic digital computer is a machine which is characterized by the ability to store numerical information, and to carry out arithmetic and logical processes on this information by obeying a sequence of instructions also stored within the machine. The word "digital" indicates that the form in which the computer stores and manipulates the information is closely akin to that by which we have become accustomed to handling numerical information, that is as a series of digits. This is in contrast to certain other types of computers which are able to store and manipulate numerical information in the form of physical quantities, for example, as electrical potentials or resistances.
The sequence of stored instructions by which the computer is controlled is called a program, and these instructions are specially written for each problem that the computer is called upon to undertake. By means of these programs, the computer can be made to perform any desired mathematical operation or logical process quickly and efficiently. The effort of programming can be further reduced by taking care to generalize the procedures in the solution of individual problems so that the resulting programs can be used for a number of separate applications.
It is the speed at which the computer is able to obey the sequence of instructions in order to follow through a particular computation, and the accuracy of the calculations even at these very high speeds, which constitute the main advantages in using electronic computers as opposed to other forms of calculating machines. A further advantage, however, lies in the fact that once the program has been prepared for a particular computation, the computer will obey the sequence of instructions faithfully, without deviation, and any number of similar computations can then be made without expert supervision. In this way, a small number of experts can undertake a vast program of research, without fear that their instructions may be misinterpreted in the handling of the information.
Even in the period between the two congresses of IUFRO, there have been further developments of the computers themselves, and modern computers are more reliable, less expensive, and require less maintenance, thanks largely to the replacement of valves by transistors in the electronic circuitry of the machines. The modern machines are also very much easier to program, that is, the language by which the user communicates with the machine is simpler to understand and to write, and, today, there is no-one reading this paper who would be unable to program a modern computer to carry out computations which he thoroughly understands.
There are of course a very large number of different types of electronic digital computers, differing mainly in the amount of information which they are capable of storing and in the medium used to transfer information into and out of the machine. The differences in the engineering techniques by which the various processes are carried out, though of interest to the electronics engineer and physicist, are less important to the forester. The speeds of individual machines also differ considerably although, here again, the most important difference is the change from the very slow computing speed of electric desk calculators to the entirely different order of speed made possible by the electronic computer and, for most forestry applications, differences in speed within the order obtained by modern machines will not be of great significance. The differences in the amounts of information that can be stored are perhaps slightly more important, although there are few existing machines that are too small for use in forestry research, and most computers have a greater capacity than is required for the majority of forestry purposes.
The differences in the medium used for the transfer of information into and out of the computer are very much more important for the practical user. Many computers were designed to take advantage of the earlier development of conventional punched card machines, and have therefore been designed to take advantage of existing equipment for preparing, sorting and tabulating punched cards as a method of translating information to be read into and out of the computer, and for reading the results produced by the computer. The punched card becomes the input and output medium for these computers. Other machines have been based on input and output techniques which are dependent on punched paper tape, using five-, seven- or eight-hole teleprinter tape, and such machines have the advantage that the machinery for the preparation and reading of tape are considerably cheaper than those for handling punched cards. Both punched cards and punched paper tape, although very fast by conventional standards, have been found to limit the speed of the computer, and most modern electronic computers offer an alternative, and very much faster, medium in the form of magnetic tape or magnetic film.
Use of computers by forest research organizations
In March 1960, the President of IUFRO carried out a survey by correspondence of the use that was being made of electronic digital computers by forest research organizations. The results of this survey indicated that extensive use of electronic computers was taking place in Australia, Canada, Sweden, Switzerland, the United Kingdom, the United States, and that less extensive use was being made of these machines in Japan, the Netherlands and New Zealand. Valuable experience in the application of computers was therefore being gained in all these countries. Rather surprisingly, many organizations stated categorically that they were not interested in the application of electronic computers to problems of forest research, and some were clearly confused between computing methods based on punched cards and electric desk machines and those based on electronic digital computers. While it was expected that forest research organizations in countries in which electronic computers were being actively developed would be the ones most aware of their possibilities, and the ones which would have easiest access to computers, it was not expected that so many organizations would have. been unaware of the possible economies in the use of these machines.
It should perhaps be stressed at this point that the computer is of the greatest importance in those countries which have only a small reserve of trained scientific staff, as it relieves this small body of experts of the tedium of calculation and routine handling of information, and enables them to devote their energies to more worthwhile matters.
Applications of electronic computers in forestry
In countries making extensive use of electronic computers in forest research, the first applications of these machines were very naturally concerned with the speeding up of computations which were already being undertaken by other means. Examples of this class of computation are given by the calculations involved in large-scale forest enumerations, in the construction of yield tables and volume tables, and in the mathematical analysis of designed experiments and surveys.
In all of these applications, little advantage was taken of the basic properties of the computers, except that of the speed of calculations, and the form of the calculations was little different from that which would have been used by more conventional computing aids. Later, however., it became apparent that the most important use of computers was in their extension to types of computation that had never before been attempted, not merely because they would take too long, but also because they were too complex to be handled by the conventional computing machines. Examples of these calculations are multivariate analysis of complex problems, such as the effects of site on choice of species and on growth, of the relative values of physical properties of timber in assessing its. quality, of taxonomy of plants and animals important in forestry, and of crop-weather relationship. Other applications of this kind include the building of mathematical models to simulate practical problems, as in the various techniques known as operational research, e.g., Monte-Carlo methods, linear programming, the queue theory, and the theory of games.
The present phase of the advanced use of the electronic computer in forest research is in the greater exploitation of the basic processes and principles of the computer, in the ability to sort, store, and handle information rapidly and in new ways, as opposed to the substitution of these processes for the more usual and slower methods in previously elaborated theories. These developments will undoubtedly lead to entirely new methods of forest research, exploiting computer techniques to the full, and based on completely automatic data processing systems.
International library of programs
As more and more forest research organizations come to use electronic computers, the stock of programs for computations of interest to foresters will increase rapidly. To avoid waste of time in duplicating work that has already been done, therefore, and to make available the work of any individual organization to the widest possible field of application, it will be desirable to arrange for the interchange of the basic programs that are written. There are however a number of difficulties. First, different machines use different order codes or languages for their programming and, even among the machines made by the same manufacturer, there is frequently no common language between the machines.
On the international scale, the problem is further aggravated by the fact that the persons writing the programs do not speak the same language.
In the United States, there has been some success among a certain range of computers in the use of a common language, capable of being obeyed by a number of machines, and known as "Fortran," and this common language between machines may well help to solve the worst of the problems of the communication of computer programs for that continent. In Europe, an alternative common language between machines is being pioneered, that of "Algol." It is too soon to say whether or not this language will be as successful as "Fortran" has been in the United States, but it may well enable programs to be interchangeable on a wide range of future machines, if not on those already existing.
In the light of this difficulty, perhaps the best that can be done to avoid duplication of effort is to keep a register of existing programs, and to revise this register as new programs are developed. The register should record the existence of the program, stating exactly what it does, and the form in which the data have to be provided. The computer or range of computers for which the program is immediately suitable, i.e., can be run without modification, would also be recorded. To enable the program to be run on other machines, a full "flow-diagram" giving the stages of the computation, and the order in which the individual stages are entered would be given, so that programers for other machines could reproduce the calculation in the shortest possible time on their own computers.
In addition to the library of existing programs, it would also be desirable to maintain an index of those organizations and individuals who are actively using computers, and particularly those which have computing time to spare, and which would be prepared to take on computing work for other organizations on repayment. Although it will probably always be possible to have such computations done by nonforestry organizations, or by commercial computer firms, experience has shown that there are many advantages in having forestry calculations done by organizations that are also concerned with forestry.
In the absence of any permanent secretariat for IUFRO, the library of existing programs of interest to foresters, and the index of available computer time, might be maintained by the Advisory Group of Forest Statisticians, or by persons co-opted by them. Each member of this group might take particular responsibility for the knowledge of the facilities available in his region, and would therefore be the person to be approached in the first place. The distribution of the members of the group is, of course, very suitable for this purpose.
Further development
In the years between the present congress and the next congress, there will be many developments in the field of electronic computers, and in their application to problems of forest research and management. The use by forest research organizations of computer facilities existing in other organizations is essentially a temporary phase and, in a very few years' time, many forest research organizations will have their own electronic computers, and will have developed advanced methods of research based on modern data processing and computer techniques.
There is no substitute for foresters themselves becoming involved in this process, and for foresters themselves learning to program the machines, if the best possible use is to be made of the exciting and almost unlimited possibilities which these machines have disclosed. Actual programming is the only constructive way of becoming familiar with the machines and their possibilities, and it is important that the realization of what these possibilities are should be spread to the largest possible number of foresters. The author of this paper believes that so powerful an apparatus has never before been placed in the hands of the forester to enable him to fulfill his part in the great project for the preservation and rational utilization of the world's forests. The danger is rather that the electronic computer will be ignored in forestry rather than misused.
Recommendations
It is recommended:
1. that the congress should take note of the important development of electronic digital computers, and their application to problems of forest research;2. that the member organizations of IUFRO should be encouraged to carry out research into the application of these machines to their problems;
3. that a register of existing computer programs of interest to foresters, and an index of organizations and individuals actively using computers, should be maintained by the Advisory Group of Forest Statisticians, and that this group should be asked to consider how this might best be done;
4. that a meeting to discuss the experience gained in the use of computers in forestry research and management should be held at the next Congress of the International Union of Forest Research Organizations, under the auspices of Section 25, but open to the members of all sections.
