Other factors, such as the flexibility offered by simulation modeling in representing different seasonal scenarios, would also contribute to the possibility of obtaining improved cane yield forecasts. Kiker of the School of Bioresource Engineering and Environmental Hydrology, University of Natal for guidance in using the CANEGRO-DSSAT model. Working rules used in predicting soil type and depth based on soil parent material and other information (Mann, Meyer, & Hellmann, 1997).
Regression Modelling
Using MontWy averages to supplement climate data, as was done in Walker's (1989) study, provides accurate late-season yield predictions. The rainfall totals for each of the months of October through March (i.e. the maize growing season in South Africa) for the period from 1931 to the year before the season that was forecast were examined to determine which years had months with rainfall totals within the given boundaries. forecast category, viz. The technique for downscaling precipitation forecasts was of primary interest in this pilot study and to test the technique, forecast categories were selected based on the assumption of a perfect forecast, i.e. the actual categories that took place.
Comparison of Techniques
When forecasting yields, observed climate data is used to fill in the seasonal record up to the date of forecast. The general indication regarding the use of growth models in crop yield forecasting is that the models hold promise as a tool for the future. If improvements are made to the data entry, these will have to be weighed in terms of the accuracy achieved.
Discussion
Thompson sugarcane yield model
The experiments were predominantly irrigated, so the influence of water stress on the ratio would be minimal. In arid conditions such as those in the Eston area, water scarcity is inevitable and must be taken into account when estimating cane yield. This seasonal total then forms the input to Thompson's equation, allowing sugarcane yield predictions to be made.
CANEGRO-DSSAT Model
Processes represented within the CANEGRO model
It is calculated as the dry matter component distributed per stem multiplied by the stem population. Glover, 1972) to derive the cane yield (wet mass) measured in the sugar industry. The partitioning of sugarcane stalk dry matter fractions into brix and juice purity is also empirically described in the CANEGRO model using multiple regression equations (Inman-Bamber, 1991b).
The demand for crop water is simulated in the energy balance by taking into account potential evaporation (demand for atmospheric evaporation). The energy and water balances are closely linked in CANEGRO, as plant water use is controlled by the energy balance when soil water content is high, and by the water balance when the water supply is limited (Van Antwerpen, Meyer and Inman-Bamber, 1993 ) . Crop water stress is believed to occur when the amount of water required for energy balance exceeds the amount that the roots can absorb (Inman-Bamber et al., 1993).
In dividing maximum evaporation into soil transpiration and evaporation, the fraction of soil evaporation is calculated using soil water content and a simulated LA. These leaves are included because they continue to cover the ground for a significant period of time (Inman-Bamber, 1991a). green leaves) is also used in the carbon balance, where daily crop light interception for photosynthesis is modeled as a function of LA. The simulated fraction of total plant biomass in roots decreases with increasing plant age (increasing total biomass), but is always greater than 12% of total plant dry mass (Inman-Bamber, 1991a, based on data of Van Dillewijn, 1952).
Root characteristics affect the water balance for crop water supply, with the rate of supply being limited by energy balance or water balance depending on soil water content.
Model applicability
- Temperature
- Solar radiation
Temperature values for each of the farms in the MSA were determined using the spatial temperature estimation technique (Schulze and Maharaj, 1998). Direct measurements of solar radiation in Eston MSA were not available for the period considered. Two sources of soil information were available to develop soil inputs for farms in the Estonian MSA.
Not all farms indicated on the maps are under cane (including the individual farm with a TAM less than 20 mm). Growth cycles are input into the yield models by varying the start and harvest dates of the crops. A modeling strategy was required to account for the growth cycles practiced in the Eston MSA.
In the third part, the suitability of the models for dividend prediction is commented on. The oversimulation of yields becomes less important as the yields for the different years are expressed as fractions of the yield obtained in the previous year. The chapter begins with an evaluation of the seasonal precipitation forecasts used in the model.
For many purposes, a forecast of actual production for a season, i.e. the quantity of sugar cane, is needed. It has been shown that the period of rain between the forecast dates and the harvest dates has little influence on the yields predicted by the ACRU-Thompson model. gave rise to small differences in predicted yields; and .. in the small differences found between yield forecasts derived from actual and perfect rainfall forecasts. h) Point g) above indicates that for the Eston MSA: .. the crops are well established when yield forecasting begins and the observed available rairi decline is sufficient for the ACRU-Thompson model to provide a good representation of seasonal yield; and .. the influence of winter towards the end of the season ensures that growth is less significant, as a result of lower rainfall and lower temperatures during this period. i). It was shown that rainfall occurring between forecast and harvest dates exerts more influence on SRM yields at longer lead times, due to greater sensitivity to rainfall influences. j) The findings in g) and h) above may not necessarily be applicable in areas with shorter growing cycles, where crops are more dependent on the rainfall of a single summer season, such as in rain-fed coastal areas. k).
Practical Application
The research project on which this thesis is based required that recommendations be made for the practical implementation of a yield forecasting system in the South African sugar industry. If land type information is not readily available, soil inputs can be derived from information about the soil parent materials and knowledge of the soils that occur on these parent materials. Slope position, slope gradient and MAP can be taken into account in the prediction of the soil types and depths occurring on the different parent materials. e) The dominant growth cycles in the supply area should be identified and reflected in the modeling strategy.
In yield forecasting, seasonal rainfall forecasts can be used to develop suitable daily rainfall data sets to: fill a season's climate data between forecast and harvest dates. To translate categorical seasonal precipitation forecasts into daily precipitation values required by a yield simulation model, analog years in the historical time series that resemble the corresponding precipitation category for the period of interest can be identified. If there is a strong relationship between precipitation and temperature, daily temperature values recorded in selected analog precipitation years can be used to supplement temperature (and thus evaporation) data for the remainder of the season (the data are available).
Graphs or tables showing the expected yield as a fraction of the previous year's yield remove any systematic error related to the impact of yield management. To aid decision-making, crop forecasts .. should be presented in a way that indicates the range of likely yield in a season. Results can be displayed in the form of graphs or tables for sub-areas of the MSA and/or the entire MSA. k) There is a possibility of linking remote sensing with a functioning crop forecasting system.
Remote sensing can be used as an aid in developing model inputs that are spatially representative of the area.
Future Research
This assessment can then be used to ascertain whether the yield model simulations are likely to represent the spatial patterns in crop yields harvested during the season.
11 REFERENCES
South African Sugar Association Experiment Station, Mount Edgecombe and Department of Agricultural Engineering, University of Natal, Pietermaritzburg, RSA. Advances in Understanding Mesoscale Climate Variability in Climate Risk in Crop Production: Models and Management for the Semiarid Tropics and Subtropics. Bulletin of the Statistical Climate Studies Research Group, South African Weather Bureau, Pretoria, RSA Vo13 No.
Simulation of sugar cane yield at the scale of a factory supply area. Proceedings of the South African Sugar Technologists' Association. An assessment of the potential for predicting sugarcane yield using seasonal rainfall forecasts and crop yield models. Effect of irrigation scheduling on water use efficiency and yield. Proceedings of the South African Sugar Technologists' Association.
The spatial temperature estimation technique (Schulze and Maharaj, 1998) involves taking into account the proximity and relative elevation of surrounding climate stations available to estimate daily temperatures at a point of interest, such as a farm. If no observed data can be found for a given day at any station, the daily temperatures are derived from a harmonic analysis of the corresponding monthly long-term averages of the daily maximum and minimum temperatures of the most suitable station. Selection of suitable analogue years (corresponding to the 50th percentile of rainfall) for the remaining pre-harvest periods.
Selection of appropriate analogue years (corresponding to 50th percentile rainfall) for the remaining rainfall periods until harvest. case of the relevant 50th percentile rainfall amounts.
