- To compare two models (SIWARE and TUNIN), which describe the regional aspects of salt and water balance, and two models (BIWASA and SWAP), that describe coupled salt and water flow in the soil profile, in terms of accuracy and sensitivity to input data and system schematisation;
- To assess the practical applicability of these models in the context of such different water administration practises as in Argentina and Egypt.
In the preparatory phase of the investigation, a conceptual framework for model evaluation and intercomparison was developed, identifying criteria for the evaluation and describing the function of such tools in the context of water management procedures. Particularly the different needs of planning water allocation and of operation of irrigation schemes were analysed. The attributes of both on-farm and regional models were described. Two categories of users were described : Irrigation Water Boards (upstream users) and Farmers (downstream users).
- The on-farm model BIWASA, developed at Cairo University simulates water and solute flow in unsaturated-saturated porous media based on the Richard's equation that is solved using a finite-element integration scheme in two-dimensions. Prior to this investigation, the model had never been applied to real situations and evaluated against observations. The numerical scheme was improved and modified to allow for the specification of different types of boundary conditions. This was necessary to reproduce the characteristics of the selected experimental site, particularly the position and operation of the on-farm drainage system. The model was applied first using the data collected at the Zankalon experimental farm. This led to identify required improvements, particularly to simulate the effect of drains (position and depth). The numerical scheme had to be improved to deal with the number of nodes and elements necessary to describe the field situations. The model was also applied to study the situation at Argentinean site in Lavalle, Mendoza. Comparison with field measurements of soil water content and solute concentration showed that the agreement was satisfactory, although the results were rather sensitive to the soil water retention and the hydraulic conductivity curves.
- The on-farm model SWAP, developed at the Winand Staring Centre in Wageningen simulates, as BIWASA, water and solute flow in unsaturated-saturated porous media based on the Richards'equation. It includes diffusion, convection and dispersion of solute and calculates the amount of adsorbed salts. Only one dimension (depth) is considered, however, and the equation is solved using a finite difference integration scheme. This model has been applied to a study a number of cases world-wide not only by the Staring Centre, but also by scientists and practitioners in several countries. The only improvement required was the use of a coefficient in the function describing root water uptake to account for the sensitivity of crops to salinity. The model was evaluated against measurements of the solute concentration in the soil solution. Samples of water in the unsaturated zone were collected using microtensiometers and a vacuum pump. These measurements were collected prior to irrigation and during infiltration and redistribution. Such detailed observations proved more effective for model validation than measurements collected at regular but longer time intervals (i.e. weekly). The model has been used to evaluate irrigation schedules. To summarise the detailed information provided by the model (i.e. time series of vertical profiles of soil water content and solute concentration) a set of water management indicators has been defined. The model was applied first to the case of Lavalle and then of Zankalon. The Lavalle simulation study showed that a one-dimensional model as SWAP may not be adequate to describe flow processes under conditions of significant lateral heterogeneity as the vineyard in Lavalle. Furrow irrigation and the row lay-out combine to establish two-dimensional patterns in the filtration velocity, as well as solute concentration and soil water content.
- The regional model SIWARE, developed by the Winand Staring Centre in Wageningen and the Drainage Research Institute in Cairo describes water management processes in an irrigation and drainage system. Water and solute balance at farm level are calculated using a bucket-type model, which also gives amount and solute concentration of drainage water. To perform the intended case study on the Tunuyan irrigation district a considerable data collection effort was required. Some of the SIWARE input date, such as the crop-specific parameters of the response functions to water and saline stress, are not commonly observed in Argentina. To measure the temporal profiles of fractional vegetation cover for the main crops a novel approach was developed base don the use of a model aircraft to collect low altitude photographs. This gives the possibility of performing on-farm surveys at very low cost. Satellite data were used to determine irrigated areas. Very limited adaptations were required to apply SIWARE to the Tunuyan district. A significant amount of attention was dedicated to the calculation of crop water requirements.
Notwithstanding the significant efforts of a large community of agronomists and hydrologists over the last forty years commonly used procedures still give rather large differences. Such differences add to the uncertainty of water balance calculations, i.e. as done with regional models, and limit to some extent their operational use. The problem remains of the difference (relative to Egypt) in the way drainage and groundwater are considered in the model. SIWARE has been developed to study water management in the Nile Delta where drainage water is used to supplement surface water and groundwater is a minor resource. In the Tunuyan groundwater complements' surface water, while drainage water is nearly negligible (1% of surface water). This difference has not been dealt with and in the SIWARE study on the Tunuyan the amount of groundwater used is prescribed instead of being calculated on the basis of deficit. The final report includes a review of the simulation studies performed with SIWARE in Egypt to illustrate the full range of applications of this model.
- The regional model TUNIN was developed by the Winand Staring Centre in Wageningen and the Centro Regional Andino of INCYTH in Mendoza, Argentina. The model describes water conveyance in the irrigation system using so called path-efficiencies, i.e. coefficients and/or functions determined by measuring volumes of water per unit area at primary, secondary, etc., offtakes. The ratio of these volumes along a specific path in the canal system indicates (at each level in the system) the fraction of irrigation water reaching that level in relation to the amount allocated at higher levels in the canal system (e.g. from project intake to the head of a primary canal). A bucket-type model is used as in SIWARE to calculate soil water and solute balance at farm level, with the significant difference that rotational schedules are taken into account in the TUNIN procedure. Since actual data collection to apply TUNIN to the case of the Nile Delta was not feasible within the scope of the investigation, the SIWARE input data and the calculated discharges were used to construct the TUNIN input data. This approach has the additional advantage of eliminating differences due to the determination of similar variables (i.e. soil water retention capacity) with different observations and procedures. Major improvements were required to develop a new version of TUNIN that could be compared to the observations of groundwater table depth and of drainage discharge and which could be applicable to the case of the Nile Delta. Deep percolation and drainage discharge are calculated using simple Darcy equations. Depth of shallow groundwater table depth is also obtained from the water balance of the soil reservoir. A simple representation of a drainage system was introduced in the model. Notwithstanding the crude schematisation of these processes, agreement of model calculation with observations was good. The original version of TUNIN was coded in outdated FORTRAN IV, so the code was adapted to FORTRAN 77, streamlined and documented. Two user interfaces were developed to handle input and output in an efficient way. The first one was programmed using FOXPRO and the second one, which includes a graphic module to display data in map form, in DELPHI. The latter one does also allow for data conversion from a general purpose data base to TUNIN input data format. Water conveyance in the irrigation system is described in TUNIN and SIWARE in a fundamentally different way : the former relies on the concept of path-efficiency, while the latter uses simple hydrodynamic equations. To assess the applicability of TUNIN to the case of the Nile Delta, the path-efficiencies for the irrigation system in the Eastern Delta were calculated for the months of January, March, July and September. The yearly averaged path-efficiencies were subsequently used to re-calculate the water volumes at all nodal points in the SIWARE network. The agreement was excellent, indication that water conveyance and allocation calculated as in SIWARE (i.e. through detailed modelling of sub-processes) was rather similar to the result obtained with the simpler TUNIN procedure (conveyance determined by the path-efficiencies and allocation proportional to irrigated area).
- The relevance of the models described above to irrigation practice depends on how well they fit in current irrigation management practices. Simulation models may be applied to fine-tune, i.e. irrigation schedules to control the concentration of dissolved salts in the soil. Practical application of such model-based guidelines, however, requires a significant level of flexibility in irrigation management and in the irrigation and drainage systems. The contribution of IRYDA, Spain to the project was to review systematically such practices in Argentina, Egypt and Spain to evaluate the practical relevance of models. This analysis was based on written materials, including legislation and administrative regulations on irrigation, and interviews, both collected during repeated field visits in Egypt and Argentina. The analysis of legislation and of administrative regulations was not limited to the present situation and covered both the past evolution and the expected trends. In the case of Mendoza, particular attention was dedicated to the enforcement of the original legislation (1884) on water allocation, which is intended to be proportional to the actual irrigated area. With time, the area having water rights (a permanent attribute of farms) replaced the notion of actual irrigated area, with the consequence of water being often allocated on the basis of outdated figures. More recently, the original notion was re-established, giving the possibility to the Irrigation Water Board to withheld water to farmers not actually using their land. This has also the consequence of requiring a significantly more flexible planning and operation of the irrigation system. The analysis of current legislation and future trends showed a clear evolution towards more accurate planning and operation of irrigation systems. The new legislation on water use in Egypt does already include the principle of allocation based on actual crop water requirements, while in Mendoza proposed new legislation on water prescribes water allocation on the basis of crop water requirements, soil type and hydrological conditions (i.e. shallow water table). The review performed by IRYDA did indicate useful improvements in the models described above, while concluding that the major restriction to operational use of such models in the current rigidity of planning and operation of irrigation and drainage systems. Given the problems that such rigidity is causing, i.e. low irrigation efficiency, waterlogging and salinisation, we can expect that increased flexibility will be considered as an urgent need, thus increasing the scope for operational use of simulation models.
- The work described above led to investigate in more detail how the simulation models could be used to support irrigation water management. This was done considering separately Farmers and Irrigation Water Boards. The use of models to support farmers-driven irrigation scheduling was investigated building upon the results of a previous investigation on farmers' perceptions of and preferences on water allocation and distribution. This study provided clear indications of preferred irrigation schedule, rotational interval and flowrate for three categories of farmers. Such preferences may not be consistent with available water or hydrological constraints. To evaluate the feasibility of irrigation schedules preferred by farmers, a regional simulation study with the model SWAP was performed for the same Tunuyan district considered in the TUNIN study. The preferred irrigation schedules were evaluated in two different ways :
- simulating the schedules preferred by the three categories of farmers to establish whether these schedules could have side effects such as rising water tables or salinisation. The irrigation schedules resulting from farmers' preferences were simulated. The results indicate that there are no hydrological restrictions to the implementation of the preferred schedules;
- simulating a larger set of irrigation schedules for a set of hydrological conditions, soil type and crops and ranking the schedules using multi-objective optimisation. The regionalisation of the one-dimensional calculations with SWAP was done considering all combinations of the following factors : three soil types (sand, sandy loam and loamy sand), three crops (grape, onion and peach), three initial groundwater depths and drainage/seepage flux and two initial solute concentrations. This gave 54 different situations to represent actual spatial variability. Eleven irrigation schedules were considered in combination with the 54 situations. The results of the resulting 594 simulations were summarised using the water management indicators mentioned in the preceding paragraph on SWAP. This procedure associates the values of say five indicators to each simulation (1 year, daily calculations). Multi-objective optimisation was used to rank the 594 cases using the indicators as attributes. This approach was also used to rank the cases according to different criteria by constructing different objective functions. After ranking the cases, so-called utility functions were determined by considering each attribute separately and the sequence of values of this attribute when going from the best to worst rated strategy in the ranking. The utility functions provide a measure of the relative weight of each factor in determining the rank of the strategy and of the importance of keeping a factor at a specific value. As regards the solute concentration, for example, the utility functions indicate the optimal concentration. The procedure was also applied by clustering the situations into groups having similar characteristics (i.e. same crop).
- The relevance of the simulation models to Irrigation Water Boards may be enhanced by embedding the models in an information system with a wider scope and range of functions. The information system can provide date and/or analyses useful for planning and operation of an entire irrigation scheme. In this context, simulation models become tools to analyse complex date sets. This was done by linking three components : TUNIN, a data base and the crop response functions calculated with SWAP (see previous paragraph). The information system was developed for the Tunuyan district. Diverse data have been collected during the last 20 years about this area. Effective use of this information was hampered by the diverse organisation of each separate data set and by the lack of an efficient data base. The latter was created by leaving each data set in its original format (assumed to be convenient for further data collection) and procedures developed (DBASE IV) for retrieval and analysis. Likewise, procedures were developed for extracting the TUNIN input data in the required format. This was also done for the data set containing the listing of water rights, i.e. the data used by the Irrigation Water Board for water allocation. In combination with the TUNIN user interfaces described above, the information system allows to perform evaluations of water allocation strategies. The TUNIN model is used to calculate water conveyance and distribution, drainage and deep percolation. Given the TUNIN results (i.e. water available on-farm and the rotational schedule), the SWAP response functions (per crop and TUNIN calculation unit) give a measure of the effect of water allocation on crop yield. The information system was also used to evaluate broadly defined scenarios on conjunctive use of ground- and surface water.
- Intercomparison experiment of models BIWASA and SWAP
Field and laboratory experiments were carried out to obtain soil hydrological properties; matrix and osmotic pressure head were measured separately. Results of simulation experiments were compared with field measurements to assess accuracy. Simulation experiments were executed to assess sensitivity of model output to errors in input data and to system schematisation, i.e. compartments and nodal points.
- Intercomparison experiment of models SIWARE and TUNIN
The two models were considered as being comparable on the basis of conveyance efficiency, when they would give the same results in terms of water delivered to the on-farm distribution system. Accuracy of models' results was evaluated using measurements of drainage discharge, depth of shallow groundwater table and salt concentration in drainage water and soil.
- Analysis of water management practices, with special emphasis on salinisation-prone areas, in Spain Argentina and Egypt
Simulation models of salt and water balance in irrigated areas are potentially useful tools to support management of water and land. The physical, technical and administrative context of irrigation in Argentina, particularly in Mendoza, Egypt and Spain was reviewed to identify constraints and opportunities for the practical use of simulation models. The study provided a basis to assess the practical applicability of these models within the context of irrigation management in the three countries considered in this investigation.
Funding SchemeCSC - Cost-sharing contracts