 |
NEO
Postbus 2176
NL-3800 CD Amersfoort
Tel: + 31 (0)33 463 74 33
Fax: + 31 (0)33 463 74 40
Website: www.neo.nl
Email: 
|
|
|
MODELLING APPROACH FOR SOIL WATER AND CROP RESPONSE
|
|
Concepts in soil water balances for crop water requirements
|
|
In crop modelling the growth of a plant is simulated, expressing the accumulation of biomass as a function of responses to temperature, sunlight, nutrients and water requirements. Both crop water and nutrients are provided through the soil as an intermediary.
In the project area neither light nor (low) temperature are factors that limit plant growth. Nutrients and water however are. In this stage of the project variations in nutrient availability have not been considered, but it will be necessary to consider this aspect in future. Crop water requirements are defined as the depth of water (e.g. in millimetres) needed to meet the water loss through evapotranspiration (which is the combined loss of water form soil and plant to the air from a surface). This factor and a number of others that also are required to model the growth of a plant are crop specific (Doorenbos and Kassam, 1979).
In this project sorghum, millet, maize and cowpea have been considered. These crops are the principal staple crops for the area.
The rainwater that drops on the surface of the soil only in part becomes available to the growth of a plant. It evaporates, runs off or runs on from somewhere else or infiltrates. The infiltrated water, that does not seep into the groundwater and which is not too strongly attached to the soil particles, is the soil water available for the plant growth. This available soil water is of course also a function of the rooting depth of the plant, which increases rapidly during the first weeks after germination. This available soil water can also be expressed as a depth of water for a certain soil type and location. The methods used to simulate the behaviour of water in a soil and its external drainage are called 'a soil water balance'.
In this project various modelling approaches have been adopted to first of all rigorously test the scatterometer data as a workable measure of soil moisture for crop modelling. Secondly to develop a model to utilise the scatterometer data as an alternative to conventional rainfall-evapotranspiration driven models to predict crop performance such as IRSIS (Raes et al., 1988), CROPWAT (Smith, 1990) and CRIWAR (Bos et al., 1996). Three aspects are addressed: planting dates, early season crop establishment, growing season stress and yield reduction.
Planting dates and crop establishment were predicted for different regions using the following algorithms: (1) probability/frequency analysis of historical rainfall plotted against evapotranspiration rates (agro-ecological zone approach), (2) rainfall sequence simulation for 1991-1995 and (3) rainfall-evapotranspiration based water balance simulations. These three planting date and crop establishment algorithms were compared to different criteria imposed on the Ms. A Boolean analysis of all the criteria in terms of 'sow and grow', 'sow and fail' and 'no sow' was reported to correspond well to the farmer's strategy of early planting and their subsequent assessment of crop establishment. In the 1998 growing season a detailed analysis was made on the calendar date of planting in different areas, which showed the practical use of Ms based planting date criteria. The final product of this work is the Planting date Estimator, PDE.
A scatterometer based water balance model was developed and compared to a rainfall-evapotranspiration model in a Visual Basic/Spreadsheet environment. Both models adopted a field capacity approach to soil water modelling with boundary conditions being field capacity and wilting point respectively. The comparison of the two models has been made for historic years over which scatterometer data and meteorological data are available, for different combinations of crops and soils. Secondly, the model has been run for the 1998 growing season for the four crops using actual SWI-values and the digitised soil map of Mali.
The performance of a crop is reduced if there is a discrepancy between the crop water requirements and the available water in the soil. A Crop Performance Index (CPI) was established to simulate periods of crop stress. A similar index defined periods of yield reduction.
For selected cases of crop and soils combinations using historic meteorological records and scatterometer data, conventional 'Et0 or evapotranspiration driven' models have been compared with the spreadsheet-model in which the SWI is used as an input. Secondly, the model has been run for the 1998 growing season for the four crops, using actual SWI-values and the digitised soil map of Mali. The CPI indicates the level of satisfaction as the ratio between modelled crop water requirements and plant available water in the profile.
Results of the works on the historic data series, field work and evaluation
The analysis of historic data series has provided the following results:
* the years 1991-1997 can be considered as representative years from a climatological point of view;
* there is a positive correlation between rainfall measurements and SWI-calculations for the 1991-1997 years, but the comparison most of all illustrates the problem of comparing point and area measurements.
Field validation took place during the 1998 growing season between May and November 1998. In this field work data were collected on planting dates, crop phenology, yield, localised crop failures, farmer's strategy, etc. Weekly SWI and DSWI-maps were sent to Mali and checked during the fieldwork, which lasted throughout the rainy season, which also involved some 7,500 kms of driving in the project area, that permitted less than formal map checking. The preliminary results were discussed and evaluated at a seminar in the beginning of February 1999 in Mali, in the presence of 10 Malinese researchers and government officials involved in crop monitoring, research, drought early warning and agro-meteorology. In general, the rainy season in 1998 in Mali was not excessive, nor deficient. However, upon an early start in June, a very dry July followed. This dry spell was specifically noted for parts of the Regions of Segou and Kayes. In September a dry spell took place in the region of Sikasso, stretching well into northern Ivory Coast.
|
|
Case studies
