Description of the Weather Data Processing


The Nitrogen and Phosphate Models take input from a weather file each time they are run. The Internet version of the Models have access to a series of weather files from different sites throughout the World (prepared as described below). The Internet user of the Models selects a climatic region and a corresponding weather file is automatically selected for input to the chosen Model. At the same time, if the user has opted for 'Advanced Input', he/she is presented with rainfall data from the selected weather file. The user can then alter this rainfall data so that when he/she commands the Model to run, a new weather file containing the user's rainfall modifications will be automatically created and input to the Model instead of the file originally selected. This automatic selection of weather files and the modification procedures outlined below are all carried out by programs within the Web System which supports the Model programs on a server computer.

 
USER MODIFICATION OF RAINFALL

Because the weather files contain daily data and the user only modifies monthly averages, his/her modifications must be expanded to provide daily values of rain. This expansion depends on a set of rules which are intended to preserve as much realism as possible within the modified weather files. They are based on the evidence in the APPENDIX. The rules are as follows.

  1. If the user reduces the monthly mean to zero, then the rainfall on each wet day is reduced to zero, i.e. every day of the month is a dry day.
  2. When the user adds rain or irrigation, the total amount is added evenly on day 1, day 6, day11, day 16, day 21 and day 26 irrespective of whether there had been any rain or not on that day 

PREPARATION OF STANDARD WEATHER FILES

These were produced so as to approximate to the average weather at each of many sites throughout the world. The crop response model requires daily values of the mean air temperature (degrees C), precipitation (rainfall in mm)and potential evaporation from an open water surface in mm.

Sites in different parts of the world outside the UK

These values were prepared at all sites, with the exception of those in the UK ( see below) by the following procedures.

First, for each site a file was produced of the mean monthly values of precipitation and the number of days on which precipitation occurred and also of the average monthly daily maximum and daily minimum temperature. The data was obtained from the booklets, details of which are given in the REFERENCES under Meteorological Office in the REFERENCE Section at the end of this document . Although the data was old it generally summarized measurements over a period of at least 30 years.

Daily values of maximum and minimum temperatures had to be generated for each day throughout the year. They were generated from the cosine model described in Peiris & McNicol (1996). The model was calibrated for each site with the monthly data by fitting, using an option in GENSTAT (Genstat5 Committee, 1987) for non uniform errors. These values together with altitude and the latitude of the site were used to generate daily values of evaporation from an open water surface (pan evaporation). The procedure is described by Linacre (1977). If, however, the average temperature was less than 4 degrees C the model was, according to the author and our experience, unsatisfactory. At these low temperatures, daily potential evaporation was set to 0.01 mm.

Daily rainfall in each month was generated from the total rainfall for that month and the number days in the month when rain fell. The procedure was method 2 described in EPIC(1997) with the X parameter set at 1.3 a value that has been found to be usually satisfactory in the USA (EPIC, 1997). It involved the generation of random numbers. And for each month the procedure was run with different randomizations to produce ten separate distributions of daily rainfall. The distribution with total rainfall that was closest to the measured average rainfall was selected for inclusion in the weather file for the site.

Sites in the UK

With the exception of daily potential evaporation from an open water surface the standard weather files for the IK were generated as described above. These were calculated from the average values over a 47 year period at Wellesbourne (near the centre of England) by assuming that potential evaporation during any month in the UK is approximately proportional to the ratio of average monthly temperature at the site divided by that at Wellesbourne.. ( See APPENDIX for the reasoning behind the approach.)

APPENDIX

Modification of rainfall.

Methods of adjusting daily rainfall for user inputs of monthly rainfall were based on analysis of following data:

  1.  
    1. Average yearly values of rainfall for each of 96 sites over a 29 year period for the UK.
    2. Average monthly rainfall for all different UK sites averaged over years.
    3. Single year monthly values for different years at Wellesbourne, Kirton and Woburn.
    4. Daily values of rainfall at sites throughout Europe.
  2.  
    1. Average yearly values for different sites in Spain.
    2. Average monthly values for different sites in Spain averaged over years.

The data for (a i & ii) and for (b i & ii) were obtained from (Meteorological Office, 1972 ,0.856c ). The data for (a iv) were kindly provided by A Walker of HRI.

For each of the (a) data sets total rainfall was far better correlated with rain per rainfall day than with number of rain days. Regression of total rainfall versus rainfall per rain day removed 83% of the variance in (ai) whereas regression against number of rain days removed only 46%.

The entire data set in (b) showed that rainfall was far better correlated with number of rain days than with rainfall per day. Regression of total rain v number of rain days removed 72% of the variance in (bi) whereas regression against rainfall per day only removed 16%.

Thus in the wetter regions total monthly rainfall is much better correlated with rainfall per day than with number of rain days while the converse holds in the drier regions. The change-over seems to take place at about 13 days.

Generation of UK weather.

The procedure was devised because it appeared to give a better estimate of UK daily potential evaporation throughout the year than could be achieved by Linacre's method (Linacre, 1977).

 

REFERENCES

EPIC (1997)Documentation\precipitation - method 2 equation 99- at Internet site:
http://brcsunO.tamu.edu/epic/documentation/precipitation.html

Genstat 5 Committee (1987) Reference Manual, Clarendon Press, Oxford

Linacre (1977) A simple formula for estimating evaporation rates in various climates using temperature data alone. Agricultural Meteorology 18, 409-424

Meteorological Office (1958) M.O.617a Tables of temperature , relative humidity, and precipitation for the world . Part 1 North America, Greenland and the North Pacific Ocean. London:HMSO

Meteorological Office (1958) M.O.617b Tables of temperature , relative humidity, and precipitation for the world . Part 2 Central and South America ,The West Indies and Bermuda. London:HMSO

Meteorological Office (1972) Met 0.856c Tables of temperature,relative humidity, precipitation and for sunshine for the world. part 3 Europe and the Azores. London HMSO

Meteorological Office (1958) M.O.617d Tables of temperature, relative humidity, and precipitation for the world. Part 4 Africa, the Atlantic Ocean South of 35 degrees N and the Indian Ocean. London:HMSO

Meteorological Office (1958) M.O.617e Tables of temperature, relative humidity, and precipitation for the world. Part 5 Asia. London:HMSO

Meteorological Office (1958) M.O.617f Tables of temperature , relative humidity, and precipitation for the world . Part 6 Australasia and the South Pacific Ocean. London:HMSO

Peiris D R & McNicol J W (1996) Modelling daily weather with multivariate time series. Agriculture and Forest Meteorology 79, 219-231.