Site Specific Nitrogen Management Simulated by CropSyst Model under Different Inputs of Nitrogen Fertilizer

Site Specific Nitrogen Management Simulated by CropSyst Model under Different Inputs of Nitrogen Fertilizer (Y Wijayanto): Site Specific Nutrient Management (SSNM) has been suggested as the only means for increasing productivity of crops and minimizing the environmental impacts. Despite of this, it is also widely recognized that compared to uniform application, SSNM provides a significant challenges related to the level of management. This is due to the fact that SSNM relates to the management of field / site (or fields / sites) and considers also the spatial and temporal component of factors leading to crop production. A method is urgently required and the most appropriate one is crop model. This study was aimed at using  CropSyst to model yields due to the difference in N applications  and its implementation for SSNM. The study area was located at Jenggawah Village, Sub-District Jenggawah, Jember Regency. Thirty soil samples were taken and six farmer’s fields were chosen for the purpose of modeling. Interview was conducted to obtain the information about the management of farmer’s fields. Yields in each farmer’s fields were used as an integrated indicator. The results suggested that the predicted yields at farmer’s fields were in agreement with those in reality. Simulated yields  based on  different amount of N inputs showed yields were proportional with different N inputs. This study concluded that there do exist a significant amount of potential applications of CropSyst for Site Specific Nitrogen Management


INTRODUCTION
considering soil-plant systems with the aim at optimizing the supply and demand of nutrient for that particular crop. The dynamic nature of management on the basis of site specific provide the most important component which differentiate SSNM to uniform application. However, implementing the nature of being dynamic and site-specific at field level leads to a significant amount of challenges for management of nutrient. Variability of soil properties have been underscored by previous studies as study conducted by Mueller et al. (2001) and this has been the obstacles for site specific nutrient recommendation with regard to how to conduct site specific recommendation, especially for Nitrogen, as shown by Dobermann et al. (2003) and Pierce and Nowak (1999). The dynamics of factors controlling crop production (climate, soil, and field-management) can incr ease the complexity of site specific recommendation. Tool is needed, which can fulfill criteria of being: (a) able to integrate the dynamic nature of nutrient management in one side, (b) integrate factors controlling the plant production and (c) this tool is expected to the management aids for making decision. Crop model is the only tool which can accomplish these three criteria. Confalonieri et al. (2006) has strongly underscored the uses of crop model for studying the dynamic nature in agricultural system, who claims that crop models have been increasingly used to study the behavior of complex agricultural system and to understand the interaction between soil and plant under different meteorological conditions.
Crop model for simulating crop yields and biomass under different condition can be found in some studies by Cavero et al. (2000), Confalonieri et al. (2006) and Yang et al.(2004), which used different crop models. Despite a quite number of the application of crop models for studying the dynamic nature of agricultural systems, there is a few evidence on the uses of crop model for Site Specific Nutrient Management (SSNM). The reason is most likely due to the greater consideration of spatial and temporal variability in the SSNM than other studies, and the temporal variability provides the most challenging management component for SSNM (Pierce and Nowak, 1999). Due to the capabilities of crop model to analyse dynamic nature of agriculture production, spatial and temporal aspects of SSNM can potentially be managed. This study used CropSyst model as dynamic and mechanistic model (Stockle et al. 2003) for simulating yield due to the difference in Nitrogen (N) fertilizer input at field level. Yields was simulated in this study because the difference in soil, moisture, nutr ients and other factor s contr olling crop productions are ultimately realized in crop yields. Therefore, the main aim of this study is on the use of CropSyst to model yields as a result of the difference in N applications and its implementation for SSNM.

Soil Samples
Thirty surface soil samples were taken at July to October 2008. The location were determined by using Global Positioning Systems (GPS) using the Universal Transverse Mercator (UTM) coordinate system as listed in Table 1.
Grid sampling method was employed. The analysis of soil samples was, then conducted in the laboratories (Soil Physical Laboratory and Soil Chemistry and Fertility Laboratory at the Soil Department, Faculty of Agriculture, University of Jember). The standardized methods of soil samples analysis were employed, consisting: soil texture, bulk density, cation exchange capacity and pH. All data were used for CropSyst modeling. The results of laboratory analysis for thirty samples were then input into Geographical Information Systems (GIS). The interpolation techniques (kriging) within GIS was then employed to determine the values of each soil property at un-sampled areas. Detailed information about interpolation techniques within GIS can be found in Mueller et al. (2001) were then used to determine the values of each soil property within the farmer's fields.

Farmer's Fields
Seven farmer's field were determined as samples. These seven fields represents the variation of field managements in the study area. The values of soil properties within farmer 's fields as a results of interpolation were then determined by averaging the values of each soil properties for every interpolated sites within each farmer field. The values of each soil properties needed for CropSyst modeling can be seen in Table 2. These values was then input into CropSyst. Another information regarding crop management within farmer's field was determined by interviewing six farmers, consisting of (a) dates of some important crop management and performances: sowing, harvest, fertilizer applications, irrigation, phenological stages; (b) the amount of application (irrigation, fertilizer N). Ten years climatic data necessary for conducting modeling was collected from averaging the interpolated records of twenty one rainfall stations in Jember.

Modeling
After inputting the necessary data for CropSyst, modeling was then conducted. Modeling was conducted by using CropSyst version 4.12. Three scenarios of the amount of fertilizer application were constructed. These scenarios were (a) the amount of fertilizer applied by farmers; (b) the amount of fertilizer applied based on general recommendation, that is 300 kgha -1 ; (c) No fertilizer N applied (Table  3).
Calibration and validation of CropSyst model was then conducted. In order to calibrate model, the values of parameters obtained from previous studies (Bellocchi et al. 2000 and;Kiniry et al.(1989) were used. The values of some parameters was then adjusted. The calibrated parameters and their corresponding values were then established (Table  4).

RESULTS AND DISCUSION
The results of modeling clearly shows that the values of predicted yields were in agreement with those in observed ones (EF equal to 0.97). Figure 1 shows the comparison between the predicted and the simulated yields, providing the evidence that CropSyst can be used for predicting the yields. Only slightly differences were observed between the observed and the predicted values.
The yield differences amongst the farmer's fields are an interesting phenomenon in this study. As shown, although the study area is a small area (± 40 Ha), there do exist the yield differences which strongly related to the differences in the management of fields. Date of sowing, fertilizer applications, irrigation were different amongst them. The amount of fertilizer applied (especially Urea) varies significantly among farmers (Table 2), although a small differences in soil characteristics was observed (Table 2). Therefore, site specific Nitr ogen management is required for the study area. This results seems to agree with the Attanandana and Yost (2003) in Thailand which suggested that for the purpose of increasing the nutrient use efficiency in maize, it is necessary to implement site specific nutrient management. Figure 2 shows the simulated yields in each farmer 's fields based on the suggested N recommendation (300 kgha -1 Urea for uniform applications) and its correspondence values for unfertilized fields. As can be seen, for every field, the low values of simulated yields were observed in the fields without N fertilizer. Considering the yields differences between the fertilized and unfertilized yields suggested that there do exist the yield gap. Interestingly, the yield gap was different for the different field, which is most likely due to the differences in soil responds and management conditions by assuming that the management used for simulating yields as shown in Figure 2 was similar to those shown in Figure 1.
Comparing Figure 1 and Figure 2, it is apparent that by using the recommended dosage of Fertilizer N, yields were lower than current farmer practices. The only reason for this was most likely that farmers used more N fertilizer than those recommended (that is about 138 kg ha -1 Urea), whereas, farmers applied more that this recommended N, as shown in Table 2.
The results of the analysis of yields have clearly indicated that there do exist factors contributing to yield differences for each farmer's field. CropSyst was evidently able to model yields quite accurately for every field. Evidently, yields simulated by CropSyst were in agreement with the observed ones. Therefore, CropSyst is one of potential tool for Site Specific Nutrient Management (SSNM).
SSNM consider two main impor tant components: space and time in the management of field, as stated by Pierce and Nowak (1999). From the above discussion, it was clear that each field has a different yield. In other words, yields was different spatially, due to the differences in farmer 's management and this is likely to relate to the differences in soil characteristics, as shown in Table  1. The capabilities of CropSyst to model within daily time step was the evidence that temporal (time) component has been considered. Therefore, the intent of SSNM to manage a field rather than whole fields and within a particular time can take a significant benefits of CropSyst model. The peak nutrient uptake demand of N as claimed by Jones and Jacobsen (2003), as for instance, can be determined by using CropSyst based on modeled daily N uptake. Consequently, splitting time of N application can be tracked by using CropSyst. Binder et al. (2000) claimed that the application of N fertilizer must consider the highest demand of N by maize and by using CropSyst the highest demand of N fertilizer can be determined. This shows that CropSyst can determine when the N deficiency occurs. By simulating the dosage of N fertilizer, the most appropriate amount of N can be determined to obtain the optimum yields . Because of the fact that fertilizer recommendation must take into account "how much'', "what form of" and "when" to apply nutrient inputs, as claimed by Yost et al. (2000), CropSyst is a valuable tool for establishing N recommendation for Site Specific Nitrogen Management. This study has shown the merits afforded by CropSyst model for studying Site Specific Nitrogen Management. However, more studies using CropSyst are required for the fact that this study demonstrated only the use of this model for a single crop in high yielding season of maize. The applications of this model to low yielding season, crop rotation and long term simulation need to be undertaken.

CONCLUSIONS
This study has shown the benefits of crop model (CropSyst) for studying Site Specific Nitrogen Management. The results of this study suggested that each famer's field has a different characteristic. The inherent soil and management characteristics within each farmer's field has led to the yield differences amongst them. This study has ascertained the benefit of CropSyst to model maize yields under different condition of fields. The capabilities of CropSyst for simulating yields under different farmer's fields have proved that the yields were strongly affected by the integrated factors, and CropSyst allowed the integrated analysis of these factors. The results of this study clearly revealed the substantial supports of CropSyst model for Site Specific Nitr ogen Management.