Agricultural Meteorology: evolution and application
J.L. Monteith
∗Institute of Terrestrial Ecology, Bush Estate, Penicuik EH26 OQB, Midlothian, UK
Received 1 October 1998; received in revised form 20 May 1999; accepted 2 June 1999
Abstract
Attempts to relate agricultural production to weather go back at least 2000 years and are still evolving. Mainly qualitative studies in the 19th Century were followed by statistical analyses, then by microclimatic measurements and most recently by modelling. From 1968 onwards, publications in Agricultural and Forest Meteorology chart the progress of research and major trends. The main body of published work has dealt with the response of crops to climate and microclimate. Much less attention has been paid to livestock environments and the impact of weather on pests and diseases. Models purporting to describe the impact on production of climate and climate change have performed erratically when tested against field measurements. © 2000 Elsevier Science B.V. All rights reserved.
Keywords: Agricultural (and forest) meteorology; History of agricultural meteorology; Trends in agricultural meteorology; Yield models in agricultural meteorology
1. Introduction
1.1. Early history
Agriculture is an industry even more dependent on weather than other systems for harnessing biological resources such as forestry and fishing. Since crops were first cultivated and livestock reared, farmers have acknowledged the overriding importance of weather in setting both potential levels of production related to sunshine and rainfall and achievable levels that depend on the severity of droughts, gales and the prevalence of pests and diseases.
To survive, the earliest farmers must have learned from experience how to maximize harvests in good seasons and to minimize losses in bad ones; and in course of time their tradition and wisdom was formalized and recorded in simple agricultural hand-books. In the first century BC, for example, a Chinese
∗Tel.:+44-131-445-4343; fax:+44-131-445-3943.
agronomist, Fan Sheng-Chih, wrote an account of his experience that was eventually translated into English under the title On Fan Sheng-Chih Shu — An
agricul-turalist book of China (Shih, 1974). Not surprisingly,
the contents are qualitative rather than quantitative but there are many examples of what we now refer to as ‘agricultural meteorology’. For example, farmers were advised to compact snow with rollers during winter and early spring in order to maximize water stored in the soil after the snow melted. Thereafter, they could look forward to the time when “. . . after thawing, the breath of earth comes through so the soil breaks up for the first time. With the summer solstice, the weather begins to become hot and the yin breath strengthens. . .”
It is salutary to remember that the branch of science we now refer to as Agricultural Meteorology has its roots in centuries of experience gathered and organized by farmers. The first descriptive stage of the subject, already illustrated by the writing of Fan Sheng-Chih, dominated the subject for centuries and was still
inant when Edward Mawley, President of the Royal Meteorological Society in 1898, gave his Presidential Address on ‘Weather Influences on Farm and Gar-den Crops’. He wrote: “There are few sciences so in-timately connected with each other as Meteorology, Agriculture and Horticulture” and went on to suggest that “Of all the influences brought to bear on vegetable life by the atmosphere, that of temperature is the most powerful and far-reaching” (Mawley, 1898).
1.2. The nineteenth century
Twelve years later, another agriculturally-minded President, Henry Mellish, spoke on a similar theme (Mellish, 1910). Like Mawley, however, he made no attempt to introduce quantitative relations be-tween weather, productivity and management. The emergence of agricultural meteorology as a science was not far off, however. R.H. Hooker, President in 1921–1922, advocated the use of statistics to corre-late yields with weather variables. He appreciated that “There is an enormous amount of work to be done before we can determine the effect of a given change in any one phenomenon at any given period of the plant’s growth upon the ultimate yield of the crop” (Hooker, 1921). On a more positive note, he observed that “quite recently, a few writers on the Continent of Europe. . . have made contributions to the subject, and the general subject of ‘agricultural meteorology’ is there beginning to attract attention, particularly in Italy”. The quotation marks that appear round agricultural meteorology in Hooker’s paper are a reminder that this was a new term in the early 1920s, introduced to describe a field of research that was essentially statistical.
In the period from 1920 to 1960, agricultural meteo-rology and the closely-related subject of micrometeo-rology took root and began to flourish after the Second World War in several European countries, in North America, in Australia and in China and Japan. One important stimulus was the emergence of new instru-mentation for measuring and recording the physical environments of both crops and livestock and their re-sponses to microclimatic factors. Initially, there were few relevant text-books apart from Geiger’s ‘Climate Near the Ground’ (Geiger, 1927, in German), still a good introduction to the subject but with a bias to-wards forestry rather than agriculture. Rudolf Geiger
died in 1981, but a Fifth Edition containing new mate-rial was published posthumously (Geiger et al., 1995). Eventually, in 1968, Elsevier launched a new jour-nal — ‘Agricultural Meteorology’ — which flour-ished and was expanded to ‘Agricultural and Forest Meteorology’ (hereafter AFM) in 1984. The titles of papers published in this international journal through the years reveal areas where the most significant work has been done and how enthusiasm for specific top-ics has waxed and waned over the years. Some early papers were characteristic of the descriptive and sta-tistical phases of agricultural meteorology; but over the years a third mechanistic phase became dominant, stimulated by the development of first-class instru-mentation both for recording and for processing data.
2. Agricultural (and Forest) Meteorology
Table 1
Distribution of papers in Agricultural and Forest Meteorology journal
Fields December 1964–June 1972 (Vols. 1–9; 187a) June 1996–June 1998 (Vols. 80–91; 185a)
No. % No. %
Climate 5 2.9 4 2.4
Microclimate
Field 35 20.2 27 16.4
Glasshouses 3 1.7 8 4.8
Animal houses 3 1.7 0 0
Storage of products 0 0 0 0
Crops
Growth 2 1.2 12 7.3
Yield 1 0.6 1 0.6
Development 1 0.6 4 2.4
Temperature 16 9.2 7 4.2
Radiation 12 6.9 14 8.4
CO2exchange 2 1.2 10 6.1
Water balance 39 22.5 41 24.8
Pests and diseases 8 4.6 1 0.6
Other damage 3 1.7 0 0
Livestock
Climatic stress 1 0.6 0 0
Pest and diseases 4 2.3 0 0
Instrumentation 36 20.3 4 2.4
Modelling 2 1.2 32 19.4
aTotal number of papers.
work has been done (though not necessarily the work that has been of most immediate benefit to farmers and foresters!); and the way priorities in research have changed with time.
Major topics and trends can be summarized as follows:
(a) The impact of weather on field crops constitutes the bulk of papers in both periods (a) and (b), with a very strong bias in favour of water balance and evapo-ration in temperate climates (ca. 40%). Interest in the agricultural implications of rising carbon dioxide con-centrations has increased over the period covered but is still not a major theme.
(b) Microclimatic studies in the field constitute al-most 20% of all papers. A decrease in this fraction with time has been balanced by an increase in papers exploring the physics of glasshouse microclimates.
(c) In contrast to the literature on carbon dioxide, AFM has published relatively few papers on the ways in which global increases of temperature are likely (i) to increase crop growth rates in many environments; but (ii) to increase rates of development and thereby
to shorten the growing warn of determinate crops; and (iii) to shorten the life cycles of pests and pathogens. (d) A few papers on microclimatic aspects of animal housing were published in early issues but none have appeared recently. Likewise, the response of livestock to climatic stress has received little attention.
(e) The microclimate of stored produce, a particu-larly important issue in the tropics, is unrepresented in both periods.
(f) Early interest in the development of instrumen-tation appears to have been replaced by an enthusiasm for modelling, which accounts for almost one fifth of the papers in recent issues. This trend is an inevitable consequence of the increasing availability of PCs and of relevant software.
3. Other issues
relation between weather and stress in animals. Con-versely there is much less emphasis on water balance. However, the most significant difference between the two sources is that Griffiths devoted about a third of his review to the impact of weather and climate on ‘Decision Making, Management and Economics’. This important field is almost unrepresented in AFM, which has a strong biophysical bias. The treatment in Griffith’s Handbook is therefore particularly welcome. An older source of information is Agrometeorolog-ical Methods, the proceedings of a UNESCO Sympo-sium held in Reading, UK, in 1966. This provides an overview of the state of the art 30 years ago, against which one can measure progress — or the lack of it! — in particular areas. It is salutary to read the introduction by Austin Bourke (1968) on ‘The aims of agrometeorology’. Having expressed some dissat-isfaction with contemporary definitions of agricultural meteorology, he framed his own in the following way. “The task of the agrometeorologist is to apply every relevant meteorological skill to helping the farmer to make the most efficient use of his physical en-vironment, with the prime aim of improving agri-cultural production, both in quantity and quality. . .
The agricultural meteorologist can be helpful only in so far as he inspires the farmer to organize and activate his own resources in order to benefit from technical advice.”
Looking through past volumes of AFM, it is diffi-cult to believe that Austin Bourke would be happy to find so much emphasis on experimentation and theory and so little on helping farmers to be more productive and more efficient. However, in contrast to the balance of material in AFM, WMO’s ‘Selected list of WMO publications of interest to agrometeorology’ reveals an emphasis on the broad response to weather of spe-cific crops and on the impact of pests and diseases on both crops and livestock. These topics are clearly appropriate for the agricultural arm of WMO because of its concern with the many direct and indirect ways in which weather affects agricultural production and food supplies.
Two other international meetings, co-sponsored by WMO, concentrated on tropical crops which received little attention from meteorologists before the estab-lishment of the International Agricultural Research Centres. In 1982 and in Hyderabad, ICRISAT held an International Symposium on The Agrometeorology of
Sorghum and Millet (Virmani and Sivakumar, 1984). This was followed in 1985 by a similar meeting on the Agrometeorology of Groundnut (Sivakumar and Virmani, 1986).
Complementing these two publications, INRA and others recently published ‘Agrometeorology of Mul-tiple Cropping in Warm Climates’ in which Baldy and Stigter (1997) bring together a wealth of tropical measurements in physically complex systems.
4. Models
In the 32 years since the UNESCO Symposium at Reading, agricultural climatology has advanced on several fronts, notably in the measurement of water vapour and carbon dioxide fluxes and of polluting gases; and in using PCs to build models that ‘simulate’ the response of crops or livestock to climate with the ultimate objective of improving the effectiveness of management.
The construction of contemporary crop models en-tails the combination of many algorithms for physio-logical processes and for the impact of environmental factors on process rates. Commonly, the values of model parameters are drawn from diverse sources but this procedure has two weaknesses that are avoided by the protocols of traditional science. First, infor-mation from different sources is usually difficult to be compatible but incompatibility is usually difficult to detect and its impact on model output is hard to quantify. Second, it is usually implicitly assumed that input parameters are devoid of error so that pre-dictions from models are impressively precise and untainted by the uncertainties associated with mea-surements! Field tests of models, preferably over a range of climates, are therefore essential to establish the validity of a model but relatively few rigorous examples can be found in the literature. Two of these have been chosen for contrasting success.
First, Hammer et al. (1995) obtained measurements of peanut yields (Y) from 16 trials in Australia, one in Indonesia and eight in the USA. Using local weather and soil information, yields were correlated with esti-mates (X) from PEANUTGRO, a model in the CERES family, to give the regression
This is a highly satisfactory result justifying the authors’ confidence that the model can be used to identify new sites suitable for development by the industry.
In contrast, Landau et al. (1998) attempted to cor-relate yields of wheat reported from 341 field trials in the UK with estimates from the Agricultural Re-search Council wheat model (AFRC WHEAT2), from CERES-Wheat and from SIRIUS, models developed in the UK, the USA and New Zealand, respectively. Weather was represented by maximum and minimum air temperature, radiation and rainfall. The respective correlation coefficients for the three models were 0.02, 0.00 and 0.04, but none of these figures were signifi-cantly different from zero.
Given such conflicting evidence, how far can we trust contemporary models, particularly those con-cerned with crop yields? Wide variations have been found in the yield predicted by different models for a specific crop in a defined environment (e.g. Matthews et al., 1995 for rice); or when the same crop model is combined with predictions of future climate ob-tained from different GCMs (e.g. Seino, 1995). On the whole, models of water use appear more reliable because they have a much simpler basis and have been tested and improved for much longer. Hopefully, yield models will eventually achieve comparable success.
5. Closing remarks
As we strive to maintain a sensible balance between measurements and models, between instruments and algorithms, we need to remember that food supplies ultimately depend on the skill with which farmers can exploit the potential of good weather and minimize the impact of bad weather. I am sure this is what Austin Bourke had in mind when he challenged participants at the Reading meeting in the following terms.
“The only valid test of an agrometeorological rule or formula is this: does it give practical results?
. . .does the procedure, on average, lead to practical help for the farmer? That this vital criterion should not be lost sight of, I suggest that throughout the coming week we should imagine that we are delib-erating in the presence of a jury of intelligent prac-tical farmers, with constant regard to their reaction
to our discussions and their probable verdict upon our work. If we do this, we shall not stray too far from the strictly utilitarian path. . . mapped out for us.”
References
Ashford, O.M., 1964. Technical assistance in agricultural meteorology. Agric. For. Meteorol. 1, 14–21.
Austin Bourke, P.M., 1968. Introduction: the aims of agrometeorology. Agroclimatological Methods: Proceedings of the Reading Symposium. UNESCO, Paris.
Baldy, C., Stigter, C.J., 1997. Agrometeorology of Multiple Cropping in Warm Climates. Institut National de la Recherche Agronomique, Paris, France; Oxford and IBH Publishing Co., New Delhi, India Science Publications Inc., Enfield, USA. Blanc, M.L., Smith, L.P., 1964. International agricultural
meteorology. Agric. For. Meteorol. 1, 3–13.
Geiger, R, 1927. Das Klima der Bodennahen Luftshicht. Vieweg, Brunschweig.
Geiger, R., Aron, R.H., Todhunter, P., 1995. The Climate Near the Ground. Vieweg, Braunschweig, Germany.
Griffiths, J.F., 1994. Handbook of Agricultural Meteorology. Oxford University Press, Oxford.
Hammer, G.L., Sinclair, T.R., Boote, K.J., Wright, G.C., Meinke, H., Bell, M.J., 1995. A peanut simulation model. 1. Model development and testing. Agron. J. 87, 1085–1093.
Hooker, R.H., 1921. Forecasting the crops from the weather. Quart. J. R. Meteorol. Soc. 47, 75–99.
Landau, S., Mitchell, R.A.C., Barnett, V., Colls, J.J., Craigon, J., Moore, K.L., Raybe, R.W., 1998. Testing winter wheat simulation models’ predictions against observed UK grain yields. Agric. For. Meteorol. 89, 85–99.
Matthews, R.B., Kropff, M.J., Bachelet, D., van Laar, H.H., 1995. Modelling the Impact of Climate Change on Crop Production in Asia. CAB International, Wallingford, Oxon., UK. Mawley, E., 1898. Weather influences on farm and garden crops.
Quart. J. R. Meterol. Soc. 24, 57–82.
Mellish, H., 1910. Some relations of meteorology with agriculture. Quart. J. R. Meterol. Soc. 36, 77–92.
Seino, H., 1995. Implications of climate change for crop production in Japan. Climate Change and Agriculture: Analysis of Potential International Impacts. American Society of Agronomy, Special Publication No.59, American Society of Agronomy, Inc., Madison, WI, USA.
Shih, Sheng-Han, 1974. An Agriculturalist Book of China. On Fan Sheng-Chih Shu. Science Press, Beijing, China (English translation).
Sivakumar, M.V.K., Virmani, S.M. (Eds.), 1986. Agrometeorology of Groundnut. International Crops Research Institute for the Semi-Arid Tropics, Hyderabad, India.
