GIS at the International Livestock Research Institute (ILRI)

Data, Capacity Building, and Networking Needs for the use of Geographical Information Systems
in Agricultural Research; Position paper - IIMI

ICRAF - The International Centre for Research in Agroforestry - GIS background

GIS use at the International Potato Center (CIP)

An introduction to Population Data by Uwe Deichmann

Existing soil data sets at global, regional and national level in CGIAR priority areas.

CGIAR's system-wide Information Network for Genetic Resources (SINGER)
-Present status, immediate plans and future integration

Prepared for ARENDAL II Workshop on UNEP/GRID and CGIAR cooperation to meet requirements for the use of digital data in agricultural management and research,

Arendal, Norway, 9-11 May 1995

This paper examines the availability of climatic data, its management and its use in agricultural research. The timescale of records used within the CGIAR system ranges from very long term (that may be useful for relating to tree growth) to hourly or more frequent (that may be required for estimating leaf wetness and so predicting disease risk). To simplify presentation we will concentrate on long-term climatic normals (i.e., long-term average monthly data) and historic data (here, actual sequences of daily data). The paper provides an introduction to major sources and applications as well as recommending actions CGIAR, UNEP/GRID and other organizations can take to support the development and use of appropriate databases.

1. Long term climate normals

1.1 Availability

Most large databases of climatic data are now available in digital form, though there are a number of global or continental databases also available in printed form (e.g. FAO 1987). As well as global and continental databases, individual countries produce many publications listing long-term climatic means. For example, the "Climatic Averages, Australia" publication contains mean monthly temperature, relative humidity and rainfall data for over 900 locations (Bureau of Meteorology 1988). Data from rainfall stations are so numerous that information from many are never published. The Australian Bureau of Meteorology can supply rainfall data for about 15,000 stations, though only about 7,000 have been in use at any one time. For more than 20 years, data derived from global, continental and national sources have been collected in digital databases intended for agricultural and environmental research. Two of the largest global databases are maintained at the Centre for Resource and Environmental Studies (CRES), Australian National University, Canberra and the Centro Internacional de Agricultura Tropical (CIAT), Cali, Colombia (Jones, 1995). The CRES database has developed from the GLOCLIMEANMTH database which was begun during the 1970s and 1980s by Henry Nix's group at the CSIRO Division of Water and Land Resources (McMahon 1986). There has been considerable exchange of data between the CRES and CIAT databases and currently they both hold data for over 25,000 locations (see Figure 1).

A database of monthly mean temperature, rainfall and cloudiness has also been developed at the International Institute for Applied Systems Analysis (IIASA)(Leemans and Cramer 1991). This includes data for over 15,000 locations. The coverage in Europe and the US is better than the CRES and CIAT databases, but coverage in developing countries is poor. The lack of maximum and minimum temperatures in the IIASA database is a disadvantage for agroecological modelling. Other organizations, such as FAO, maintain large databases of climatic data that are used for agricultural analyses and many CGIAR centres maintain smaller climatic databases for areas of particular interest. Organizations such as CRES and CIAT provide climatic data to other research bodies, but their base data sets may not be immediately available to potential users, depending on media and software compatibility.

The widespread acceptance of CD-ROM in recent years has provided an excellent medium capable of delivering large volumes of data at low cost. For example, the World Weather Disk (Weather Disc Associates 1994) contains six global climate datasets including monthly mean maximum temperature, minimum temperature, relative humidity, rainfall, wind speed and several other factors for 5,717 airport locations around the world. The International Station Meteorological Climate Summary CD-ROM (Federal Climate Complex Asheville, 1992) provides similar monthly mean data for several thousand stations worldwide. It includes a good, map-based interface allowing the user to zoom in on areas of interest and see the meteorological stations available for a particular region. The Interactive Global Change Encyclopedia GEOSCOPE CD-ROM (LMSOFT with Geomatics Canada and the Canadian Space Agency, 1994) contains a wide variety of environmental data sets, amongst these are several global climatic data sets, including a world meteorological surface station climatology. The Carbon Dioxide Information Analysis Center at Oak Ridge National Laboratory distributes several climatic datasets on its "Numeric Data Package Collection" CD-ROM. This includes monthly mean temperature and rainfall data for 205 locations across China. Individual countries are also beginning to provide climatic data on individual CD-ROMs. The dataset of about 900 temperature and nearly 15,000 rainfall recording stations for Australia is available on the Climate Data Compact Disk (CDCD) (Space-Time 1993). But it should be noted that costs of climatic data on CD-ROMs vary widely from several thousand US dollars for the Australian CDCD to about US$140 for the International Station Meteorological Climate Summary CD-ROM. Some research data sets are distributed at no charge.

For detailed studies of individual countries or particular regions, efforts should be made to obtain more local data. The INFOCLIMA publications (WMO, 1989 and WMO, 1992 supplement) provide a useful first check of data sources and describe 1,031 data sets available from 268 data centres in 112 countries. However, there is no substitute for working closely with colleagues in the particular country or region of interest, as climatic data may be gathered by several different organizations. For example, the FAO (1987) publication on agroclimatological data for Asia contains mean monthly temperature and rainfall data for only 64 locations in Thailand. In a recent study to assess forestry potential, mean monthly data were obtained for 181 temperature recording stations and for 1460 rainfall recording stations. These data were obtained from the Meteorological Department, Royal Irrigation Department, Electricity Generating Authority of Thailand and Royal Forest Department (Viriyabuncha et al., in press). In some countries old colonial records may also be useful.

1.2 Data Management

When obtaining data from in-country sources, it is important to make clear to the owners of the information the use to which their data will be put. This is usually not a major problem with climate normals but needs careful handling. The maintenance of long-term, mean climatic data is often not a high priority with meteorological services.

1.3 Interpolation

Most meteorological stations are located in cities, towns or villages that may be some distance from sites of interest for agroecological studies. So we need to interpolate to such areas. A good digital elevation model (DEM) is a prerequisite for most climatic interpolation studies. Typically, elevation point data and contour data are digitized and the data analyzed to produce estimates of elevation on a regular grid. The ANUDEM package (Hutchinson 1989a) was written to support the development of DEMs and routines from the package have recently been included in the ARC/Info geographic information system.

Given we have a DEM, a number of alternative methods are available for climatic interpolation. Pawitan (1994) has compared inverse distance, kriging and minimum curvature algorithms. Inverse distance or inverse squared distance (Jones et al, 1990) are reasonably good interpolators. They have the advantage of speed for large data sets with limited computational capacity. More complex interpolation algorithms (e.g. Laplacian splines) are better interpolators, but need more computing power. The major difference between the techniques used by CIAT (Jones et al 1990) and CRES (Hutchinson 1989b) is how the lapse rate correction is applied. The CIAT method uses a standard lapse rate applied over the whole dataset; the CRES method uses a 3-dimensional spline algorithm to determine a local lapse rate from the data. The former introduces error where the local lapse rate deviates from the standard lapse rate function. The latter suffers where there are erroneous data or insufficient data range in the local area, resulting in a spurious correction for elevation. The ANUSPLIN package provides a powerful set of programs especially designed for climatic analysis.

Climate surface relationships are available from several sources. In addition to CRES, climatic interpolation work is being carried out at CIAT (e.g. Jones et al. 1990), the International Centre for Research in Agroforestry (ICRAF), CSIRO Division of Forestry and several other centres. Climate surface relationships should not be viewed as static datasets. The relationships change as the underlying data are corrected, enhanced and updated. Recording the processing history of climate surface relationships is therefore as important as recording the history of records from individual meteorological stations. An example of good documentation is provided with climate images included on the Interactive Global Change Encyclopedia GEOSCOPE CD-ROM (LMSOFT with Geomatics Canada and the Canadian Space Agency 1994). As interpolated datasets proliferate there is a need for a metadatabase indicating who has developed surfaces for particular regions, the quantity and quality of data used, and the availability of fitted surfaces and/or gridded data.

1.4 Use

Before computers became widely available, agroecological analyses often made use of classifications and printed maps (see Cramer and Leemans, 1993 for a summary). The Koppen (1918), Holdridge (1947) and Thornthwaite (1948) classifications are examples of some of the more popular systems. The climatic databases that are now becoming available allow these systems to be calculated and mapped precisely. As well as these general classifications, we can now use the databases to develop custom-made classifications for particular applications. The BIOCLIM program developed by Nix, Busby and Hutchinson (Busby, 1991) uses climatic interpolation relationships to develop descriptions of the climatic requirements of particular taxa. BIOCLIM applications have used up to 36 climatic factors.

A simpler, six-factor approach has been developed at the CSIRO Division of Forestry to assist tree species selection (Booth 1991a). Figure 2 shows an example of output from a climatic mapping program for China which includes data for nearly 100 000 locations. Researchers at CIAT have also developed problem specific classifications, a "Seeds of Hope" project map was developed that shows areas of Central Africa climatically similar to different Rwandan regions. This is helping to select seed to distribute to farmers to replace stocks lost during the civil war. Agroecological classifications developed using climate as well as other environmental data can be used to assess the likely benefits of recently developed technologies or to anticipate the relative impact of proposed work (Wood and Pardey, in press).

Detailed process-based models have been developed for most of the world's major agricultural crops, but these models require much more detailed information than monthly mean data. Hackett (1988) developed a simple model (Plantgro) to predict the performance of the hundreds of lesser-known plants used in agricultural, horticultural and forestry systems around the world. It provides semi-quantitative predictions of growth and uses information on 12 important soil factors including pH, nitrogen and phosphorus plus monthly, ten-daily or weekly data for maximum temperature, minimum temperature, rainfall and solar radiation. The model uses a simple water balance sub-model as well as light and temperature sub-models. A simplified version of Plantgro has been incorporated into PC-based simulation mapping programs that use soil data derived from maps along with interpolated climatic estimates for thousands of locations (Booth 1991b).

2. Historic Climatic Data

2.1 Availability

Climatic variability plays a vital role in determining the year-by-year performance of particular enterprises. Several CD-ROMs include actual daily time series data. The World Weather Disc (Weather Disc Associates, 1994) includes daily maximum and minimum temperature as well as rainfall data for 222 US stations, with most data extending beyond than 70 years. Monthly time series data for mean temperature and rainfall are included for 3,293 stations worldwide with some records going back more than 100 years. The CDCD for Australia (Space-Time 1993) includes daily rainfall data for many locations with some records going back 150 years. The World Climate Disc produced by the University of East Anglia, Climate Research Unit presents a global coverage of stations with historic mean monthly series of temperature, precipitation and barometric pressure.

The National Climate Data Center has produced a global daily summary (GDS) CD-ROM which includes daily maximum and minimum temperatures and rainfall for about 10,000 stations for 1977-1991. NCDC provides access to the latest month's daily data for about 8,000 sites around the world via anonymous ftp on the internet. Contact NCDC on http://www.ncdc.noaa.gov/ or email NCDC at orders@ncdc.noaa.gov). INFOCLIMA publications (WMO 1989, WMO 1992) also provide a summary of historical data.

Climate change is important when dealing with time series data. Many will be familiar with global warming and its predicted effects on agriculture (Parry 1990). There is evidence that certain areas of the world, such as the Sahel, have had significantly reduced rainfall in recent years (Hulme 1992). Great care should be taken in interpreting conclusions made using time series data from such regions. Teams studying climate change have somewhat similar data needs to modelers in the CGIAR system. The Weather Generator Project within the Biospheric Aspects of the Hydrological Cycle (BAHC) section of the International Geosphere-Biosphere Program (IGBP) is developing a system for using predictions from general circulation models to produce higher resolution climate data needed for ecological and hydrological research (BAHC, 1993). The timescales are mainly monthly or daily. However, the spatial resolution requested is generally 10 km. Much finer resolution is needed for agricultural management and research down to 1 km resolution or even finer.

2.2 Use

About a dozen major crop plants, including rice, maize and wheat, dominate global agricultural production. Ritchie (1994) and Jones et al. (1994) have reviewed some of the most important published dynamic crop models. Most widely used models need daily weather data: maximum and minimum temperatures, solar radiation and rainfall. Model users often need simulated daily weather data. This involves research on two fronts: stochastic simulation of weather and the capacity to spatially interpolate the weather emulation. The Decision Support System for Agrotechnology Transfer (DSSAT) incorporates weather simulation routines based on markov rainfall simulation (IBSNAT 1993). Recent work (Jones and Thornton, 1994) has shown that the markov simulation in tropical climates is considerably more demanding than that incorporated in DSSAT. It requires a third order model with probability resampling to account for annual variance. Work continues on the interpolation of the third order markov model for use in spatial analysis of risk and yield forecasting. A recent example is its use in Burkina Faso (Jones et al. 1994) to produce real time estimates of millet yields based on FEWS (Famine Early Warning System) rainfall estimates. The markov model parameters were spatially interpolated throughout the country then used at each time interval to predict the final yield by modelling forward from the FEWS rainfall estimate to the harvest date. The simulation provided a useful estimate of actual regional yields compared with the long-term mean yields long before actual harvest.

3. Discussion

CGIAR centres need access to reliable climatic data for any location in the developing world. How do we deliver this effectively and efficiently with no duplication of effort? The following recommendations would fulfil this aim:

a) Developing a moderate resolution global digital elevation model

The most readily available global DEM is ETOPO5 (Edwards, 1986), found on the Global Ecosystem Database (NOAA-EPA 1993) and GEOSCOPE (LM-SOFT 1994) CD-ROMs. It has a resolution of 5 minutes (approximately 9 km spacing) and known errors. An improved 5-minute data set is included on the TerrainBase CD-ROM recently released by the National Geophysical Data Center, but a much finer resolution grid is needed. It is a sad comment on humanity that the potential use of DEMs for military purposes has held up the release of high resolution DEM information. A global 1km grid would be enough for many agroecological purposes while having only limited military applications. The CGIAR should work with NASA's Mission to Planet Earth/Earth Observing System project, GRID and any other appropriate partners to develop a global DEM at 1km resolution and make the information available on CD-ROM as soon as possible.

b) Development of improved interpolated climatic data

One of the main problems in tropical and sub-tropical climes is how to interpolate the scarce data that are available. At present, interpolated estimates are simply based on latitude, longitude and elevation. A new approach may be needed using detailed DEMs and wind direction information as well as the present interpolation algorithms to provide a better simulation of topographic effects on rainfall. We recommend that UNEP/GRID and CGIAR seek funds to support research to improve rainfall interpolation in which CGIAR centres and institutions like CRES have an interest.

CRES, with financial support from ILCA, have developed an interpolated (1/20th of a degree grid) climatic data set for Africa which is planned to be released on CD-ROM. We suggest that CGIAR support funding requests for similar projects to make interpolated databases available for all the developing world. The information could be stored on CD-ROM either as actual gridded data or as programs and interpolation relationships that would generate gridded data (up to 1km resolution) on host computers for selected areas. This work should be carried out by agencies such as CRES and CIAT with experience in the methods. As well as developing interpolations for continents and major regions, agencies such as CRES and CIAT, should also provide training to users in particular countries who wish to develop and use interpolated climatic data. CRES has provided initial training to users in 40 countries as part of the Climatic Impacts Assessment and Management (COMCIAM) Program (AIDAB 1994).

c) Developing standards for documentating data and interpolating relationships The CGIAR should encourage the development of standards for documenting long-term climatic records (including history of detecting and correcting errors) and to develop standards to describe the status of climatic interpolation relationships. The UNEP/GRID should maintain a metadatabase indicating the existence of interpolated climatic relationships for particular regions, their quality and availability. This would complement work already done by GRID regional centres in collecting and disseminating climatic information.

d) Developing programs for climatic analysis

Interpolated climatic data can be analyzed using sophisticated geographic information systems such as ARC/Info. But many potential users in developing countries cannot afford access to expensive GIS programs or workstations. Access to climatic data and generic models need to be provided via inexpensive, easy-to-use, PC-based programs (e.g., Booth 1991 a, b; Hackett 1988). The CGIAR should support developing such programs and providing training for potential users.

To fulfil their mission, CGIAR centres need to be able to predict the performance of many different land uses at many locations. To do so they need access to high quality climatic data. UNEP/GRID and CGIAR have important roles to play in ensuring that such data are used to their full advantage. This implies an ongoing collaboration between meteorological data suppliers, CGIAR and expert centres such as CRES and the end users. These are the farmers of the developing countries.

References

AIDAB 1994. Climate Impacts Assessment and Management Program for Commonwealth Countries. Australian International Development Assistance Bureau, Canberra, Project Completion Report 1991-94. 2 vols.

BAHC (Biospheric Aspects of the Hydrological Cycle, Core Project Office) 1993. BAHC Focus 4 : The Weather Generator Project. Activity 09 4.1 Data requirements for ecological and hydrological studies and related management purposes. BAHC Core Project Office, Institute for Meteorology, Freie Universitat Berlin, 36 p.

Booth, T. H. 1991a. Where in the world? New climatic analysis methods to assist species and provenance selection for trials. Unasylva, 165, 51-57.

Booth, T. H. 1991b. A climatic/edaphic database and plant growth modelling system for Africa. Ecol. Model., 56, 127-134.

Bureau of Meteorology. 1988. Climatic Averages Australia. Australian Government Publishing Service, Canberra, 532 pp.

Busby, J. R. 1991. BIOCLIM - a bioclimate prediction system. In: Margules, C. R. and Austin, M. P. (eds.) Nature Conservation - Cost Effective Biological Surveys and Data Analysis. CSIRO, Melbourne, p. 64-68.

Cramer, W. P. and Leemans, R. 1993. Assessing impacts of climate change on vegetation using climate classification systems. In: Solomon, A. M. and Shugart, H. H. (eds.) Vegetation Dynamics and Global Change. Chapman and Hall, London, 190-217.

Edwards, M. H. 1986. Digital image processing of local and global bathymetric data. MSc Thesis, Washington University, Dept. of Earth and Planetary Sciences, St. Louis, Missouri, 106 p.

FAO (Food and Agriculture Organization). 1987. Agroclimatological Data for Asia. Volume 2 K-Z. FAO Plant Production and Protection Series No. 25, FAO, Rome.

Federal Climate Complex Asheville. 1992. International Station Meteorological Climate Summary CDROM. Jointly produced by Naval Oceanography Command Detachment Asheville, USAFETAC OL-A and National Climate Data Center.

Hackett, C. 1988. Matching Plants and Land. CSIRO Division of Water and Land Resources, Natural Resources Series No. 11, 82 pp.

Holdridge, L. R. 1947. Determination of world plant formations from simple climatic data. Science, 105, 367-8.

Hulme, M. 1992. Rainfall changes in Africa: 1931-1960 to 1961-1990. Int. J. Climatol., 12, 685-699.

Hutchinson, M. F. 1989a. A new procedure for gridding elevation and streamline data with automatic removal of spurious pits. J. Hydrology, 106, 211-232.

Hutchinson, M. F. 1989b. A new objective method for spatial interpolation of meteorological variables from irregular networks applied to the estimation of monthly mean solar radiation, temperature, rainfall and windrun. CSIRO Division of Water Resources Tech. Memo. 89/5, 95-104.

IBSNAT (International Benchmark Sites Network for Agrotechnology Transfer). 1993. The IBSNAT Decade. IBSNAT, Department of Agronomy and Soil Science, College of Tropical Agriculture and Human Resources, University of Hawaii, Honolulu, Hi,

178 pp.

Jones, P. G.; Robison, D. M., and Carter, S. E. 1990. A geographical information approach for stratifying tropical Latin America to identify research problems and opportunities in natural resource management for sustainable agriculture in Centro Internacional de Agricultura Tropical (CIAT). Agroecological Studies Unit, CIAT, Cali, Colombia, 52 pp.

Jones, P. G. and Thornton P. K.; 1994. A weather generator for agricultural applications in the tropics. Agric. and Forest Meteorol. 63:1,19

Jones, P. G.; Thornton P. K.; and Hill P. 1994. Agrometeorological models: crop growth and stress indices. Proceedings FAO expert consultation on Crop Yield forecasting. Nice, France, Oct 1994

Jones, P. G. 1995. Centro Internacional de Agricultura (CIAT) climate database version 3.41, Digital data tape, CIAT, Cali.

Koppen, W. 1918. Klassifikation der Klimate nach Temperatur, Niederschlag und Jahreslauf. Petermanns Geographische Mitteilungen, 64, 193-203, 243-48.

Leemans, R. and Cramer, W. P. 1991. The IIASA database for mean monthly values of temperature, precipitation and cloudiness on a global terrestrial grid. RR-91-18, International Institute for Applied Systems Analysis, Laxenburg, Austria, 56 pp.

LMSOFT with Geomatics Canada and the Canadian Space Agency. 1994. Interactive Global Change Encyclopedia GEOSCOPE User's Manual. MJB, Quebec, 122 pp.

McMahon, J. P. 1986. Climatic databases held in the Division. Research Report 1983-85, CSIRO Division of Water and Land Resources, Canberra, Australia, p. 60-63.

NOAA-EPA. 1992. Global Ecosystems Database Version 1.0. User's Guide, Documentation, Reprints, Digital Data on CD-ROM. USDOC/NOAA National Geophysical Data Center, Boulder, CO.

Parry, M. 1990. Climate Change and World Agriculture. Earthscan, London, 157 pp.

Pawitan, H. 1994. Climatic data generation and surfacing. In: Murdiyarso, D. and Booth,

T.H. (eds.) Evaluation of Climatic and Soil Data for Agriculture, Forestry and Conservation. Institut Pertanian Bogor, 43-68.

Ritchie, J. T. 1994. Classification of crop simulation models. In: Uhlir, P. F. and Carter,

G. C. (eds), Crop Modeling and Environmental Data. CODATA Monograph Series Vol. 1, CODATA, Paris, 3-14.

Space-Time Pty. Ltd. 1993. Climate Data Compact Disk Release 2 Users Manual. Bureau of Meteorology and Space-Time Pty Ltd, Melbourne, 183 pp.

Thornthwaite, C. W. 1948. An approach towards a rational classification of climate. Geographical Review, 38, 55-94.

Viriyabuncha, C.; Booth, T. H.; Jovanovic, T.; and Zuo, H. In press. Climatic Mapping for Thailand. In: Booth, T. H. and Haldane, J. (eds.) Matching Trees to Sites, ACIAR Proceedings, Australian Centre for International Agricultural Research, Canberra.

Weather Disc Associates. 1994. World Weather Disc User Manual, Weather Disc Associates Inc., 4584 NE 89th, Seattle, Washington, 125 pp.

WMO (World Meteorological Organization). 1989. INFOCLIMA - Catalogue of Climate System Data Sets. WMO/TD-No. 293, 508 pp.

WMO (World Meteorological Organization). 1992. INFOCLIMA - Catalogue of Climate System Data Sets. Supplement to the 1989 Edition. WMO/TD-No. 293, 31 pp.

Wood, S. and Pardey, P. G. In press. Supporting agricultural research policy and priority decisions: an economic-ecologic systems approach. In: Goldsworthy, P. B. and Penning de Vries, F.W. T. (eds.) Opportunities for Use of Systems Research Methods in Agriculture in Developing Countries. Kluwer Academic Publishers.

Figures

Fig. 1. Location of meteorological stations included in the CIAT database.

Fig. 2. Output from the climatic mapping program for China. Dark-shaded areas indicate locations climatically suitable for Acacia mearnsii