Appendix A: The integrated modeling system--framework and approach.
The Integrated Modeling System (IMS) developed for purposes of this study consists of three sub-components--HadRM3 climate data, a hydrological model (SWAT), and an agro-met simulation (EPIC) model--and their functional links. These sub-components are, in turn, linked to the economic model, described in Appendix G. The modeling toolkit and database binds these sub-components into single modeling tool, in a simple and interactive way. The basic components of the integrated modeling system include a control unit, database management, model- and knowledge-base management, and user interface.
The underlying flow of data and information in the Integrated Modeling System is illustrated in figure A.1. The starting point for the IMS is the generation of climate data based on the IPCC emissions scenarios. The resulting climate data is then used in the hydrological model, SWAT, to generate surface water data, which are required as inputs to run the agro-meteorological model, EPIC. The latter integrates water and climate data into an agricultural output estimation framework. Detailed information about SWAT and EPIC is provided later in this annex. Both SWAT and EPIC, are process-based deterministic, each is governed by a set of modeling equations. The spatial resolution used in this study is Mandal/Block resolution. However, calibrations were done with the help of local scientists--to know more about EPIC and its calibration.
The IMS requires a computer running on, at least, Windows 2000. The minimum platform configuration is a Pentium or equivalent processor running at 100 megahertz with 64 megabytes of memory, at least 500 Mb of free disk space, and a display resolution of at least 1024 x 768. For optimal performance, a Pentium processor running at 400 megahertz, or faster, with at least 128 Mb of memory, 1Gb of free disk, and a display resolution of 1280 x 1024 is recommended.
[FIGURE A.1 OMITTED]
The Graphical User Interface (GUI) for IMS-EPIC is a Windows-based tool that communicates with the internal database for data entry, editing, and data validation. The GUI was designed to overcome most of the significant hurdles faced by EPIC users, particularly the lack of a user-friendly interface. The development environment of the GUI is based on Arc View Avenue and Visual Basic; both applications operate under Microsoft Windows. Figure A.2 provides an example of the appearance of the interface which will pop up during installation.
[FIGURE A.2 OMITTED]
Once the application is installed on the computer, the Arc View application is launched. Using the toolbar one can open a view to show the regions of interest (figure A.3).
[FIGURE A.3 OMITTED]
In order to run projections and create charts depicting the projected parameters, one makes a connection to the preferred climate scenario database by choosing the scenario: Baseline, A2, or B2. The dataset is uploaded by the system and the data corresponding to the block or region desired will be displayed in a series of interfaces showing the block, the crop(s) being analyzed, the type of soil, the weather, and the crop management information. It is also possible for the user to edit such parameters. Finally, crop yields are displayed using bar charts and layouts as shown in figures A.4 and A.5.
[FIGURE A.4 OMITTED]
[FIGURE A.5 OMITTED]
Hydrological Modeling (SWAT Model)
SWAT, which stands for Soil and Watershed Assessment Tool, was developed to predict the impact of land management practices on large, complex river basins or watersheds. SWAT2000 is capable of performing continuous, long term simulations for watersheds composed of various sub-basins with different soils, land uses, crops, topography, weather, etc. Being a physically-based model, SWAT2000 requires specific inputs to model any river system rather than use of regression equations to describe relationships between the inputs and outputs. The driving force is water balance. Another important characteristic of SWAT2000 is that it provides relative accuracy as well as absolute accuracy. This model is best used to predict long-term outcomes of management practices.
SWAT2000 can handle hundreds of sub-basins. The soil profile for each of these sub-basins can be divided into ten layers. The movement of runoff, sediment, nutrients, and pesticide loadings to the main channel in each sub-basin is simulated considering the effect of several physical processes that influence the basin's hydrology. SWAT2000 requires data relating to daily precipitation, maximum/ minimum air temperature, solar radiation, average daily wind speed, and relative humidity. This information can come from observed data or it may be generated from a weather generator database.
The precipitation may be homogenous for the entire watershed; however, spatial variability may lead to unique climate conditions for the various sub-basins in the model.
Box A.1 Stochastic Weather Generator A stochastic weather generator allows the generation of synthetic daily weather data for any number of years using the statistical properties of a weather variable. In this study, a stochastic weather generator was used to generate scenarios of future climate at block level from the low-resolution RCM-derived scenarios. The climate scenarios developed from the outputs of the RCM include not only changes in the mean of the climate but also changes in its variability. The historical annual cycle of means, standard deviations, probability of wet/dry days, and number of rainy days at block level were computed using the daily rainfall data at block level and other daily weather data obtained from the India Meteorological Department (IMD). The stochastic weather generator was used to simulate daily weather for 50 years for historical (Baseline), A2, and B2 scenarios in each block of the selected districts of each study region. The weather generator is designed to preserve the dependence in time, the internal correlation, and the seasonal characteristics that exist in the actual weather data. Precipitation and wind are generated independently of the other variables. Maximum temperature, minimum temperature, and solar radiation are generated subject to whether the day is wet or dry. A first-order Markov chain is used to generate the occurrence of wet or dry days. When a wet day is generated, the precipitation amount is based on a skewed normal distribution. With the first-order Markov chain model the probability of rain on a given day is conditioned on the wet or dry status of the previous day. The procedure to generate the daily values of maximum and minimum temperature and solar radiation is based on the weekly stationery generating process given by Matalas (1967). The wind component of the model provides for generating daily values of wind speed and direction as described by Richardson (1982a). Source: RMSI, 2006b.
Assumptions and Limitations
A watershed may be broken down into several sub-basins, each consisting of different Hydrologic Response Units (HRUs). The HRU is the primary modeling unit for the SWAT model. Within a sub-basin, an HRU often consists of areas with the same hydrology, soil type, and management practices. The model assumes that the hydrological routing paths are the same for all the areas that belong to the same HRU, even if the areas are distributed in different parts of the sub-basin.
Since the model is designed to simulate and predict the impacts of long-term land management practices in watersheds on the water quality in receiving water bodies, it cannot properly be applied to simulate detailed, single-event flood routing. The stream flow network in SWAT is designed as one directional flow, which routes runoff from upper stream/reach to down steam/reach. It cannot simulate backflows.
The basic model components simulated by SWAT2000 include weather, surface runoff, return flow, percolation, evapotranspiration, transmission losses, pond and reservoir storage, crop growth and irrigation, groundwater flow, reach routing, nutrient and pesticide loading, and water transfer.
Agrometereological Model (EPIC)
The EPIC (Erosion Productivity Impact Calculator) model was developed by scientists from the US Department of Agriculture's Agricultural Research Service (ARS), Soil Conservation Service (SCS), and Economic Research Service (ERS). The model was selected for use in the IMS because it provided a more coherent modeling environment and because there was experience available in the application of EPIC in relevant parts of India.
EPIC was originally designed to assess the effect of soil erosion on productivity. It simulates the effects of management decisions on soil, water, nutrient, and pesticide movements, as well as their combined impact on soil loss, water quality, and crop yields for areas with homogeneous soils and management. Some of the important components of EPIC are: weather generator (WXGEN); hydrology, erosion and sedimentation: nutrient cycling; crop growth; tillage; economics; and plant environment control.
The IMS' model resolution is aimed at "blocks"; results are aggregated at "district" level, since this is the level of resolution of the agro-met model. All efforts are made to collect data at basin/district level and the results are generated at this level. For regional levels, the results can be aggregated from the districts/blocks. Apart from this, one major driver of this study is HadRM3 (a Regional Climate Model), which has a resolution of 0.44 x 0.44 degrees (approximately 50 kilometers cell-size) on ground covering the average size of a typical Indian district/sub-basin.
The model was validated through rigorous testing involving the comparison of reported crop yields to simulated yields based on similar conditions to those prevailing at the time that the real-world reported data were published. Efforts were made to provide accurate and realistic data-input files to close the gaps between simulated and reported crop yields.
The validation exercise showed that the observed crop yields and those simulated by the IMS were very close.
Box A. 2 Regional Climate Models vs. Global Climate Models The simulation of seasonal rainfall as well as its spatial and temporal variability over the Indian subcontinent has remained rather poor in most Global Climate Models (GCMs). This is mainly due to the fact that GCMs have course horizontal resolution, which restricts the representation of the topography's and coastlines' complexity. It also limits the parameterization of sub-grid scale processes. Therefore, GCM scenarios fail to capture the local details needed for conducting impact assessments at the regional level. In addition, GCMs cannot capture extreme weather events and their intensities such as cyclones or heavy precipitation events. The alternative method to obtain detailed predictions of the future climate is to use a high-resolution version of GCMs known as Regional Climate Models (RCMs). An RCM has a high resolution (typically 50 kilometers, compared to 300 kilometers in a GCM) and covers a limited area of the globe (typically 5,000 kilometers x 5,000 kilometers; roughly the size of a box around Australia). It is a comprehensive physical model, usually made up of the atmosphere and land surface, containing representations of the important processes in the climate system (e.g. clouds, radiation, rainfall, soil hydrology). At its boundaries, RCM is driven by atmospheric winds, temperature, and humidity output from a GCM. RCM predictions of ideally 30 years (e.g. the period 2071-2100) are needed to provide robust climate statistics, e.g. distributions of daily rainfall or intra-seasonal variability. The third-generation Hadley Centre RCM (HadRM3) is based on the latest GCM, HadCM3. It has a horizontal resolution of 50 kilometers with 19 levels in the atmosphere (from the surface to 30 kilometers in the stratosphere) and four levels in the soil. In addition to a comprehensive representation of the physical processes in the atmosphere and land surface, it also includes the sulphur cycle. This enables it to estimate the concentration of sulphate aerosol particles produced from S02 emissions, which have a cooling effect as they scatter back sunlight and also produce brighter clouds by allowing smaller water droplets to form. Thus, regional models can take account of the effects of much smaller-scale terrain than GCMs. In spite the RCMs' advantages, their level of resolution is not high enough to assess the impact of climate change on natural resources. In order for this assessment to be as accurate as possible, the resolution would have to be even better, that is, approximately 10 kilometers by 10 kilometers. In an ideal, best case scenario it would be as high as 1 kilometer by 1 kilometer. Thus, there is a mismatch between what climate models can supply and what natural resource impact models require. Since the technology to create even lower-resolution climate models is not yet available, other alternatives are available to produce a similar effect. These alternatives involve the manipulation of the climate data fed to the models using an approach called "Statistical Downscaling". This means developing high-resolution climate data using the IPCC future climate scenarios and the observed data of rainfall, temperature, solar radiation, relative humidity, and wind speed. Source: RMSI, 2006b.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Climate Change Impacts in Drought and Flood Affected Areas: Case Studies in India|
|Publication:||Climate Change Impacts In Drought and Flood Affected Areas: Case Studies In India|
|Date:||Jun 1, 2008|
|Previous Article:||6. A way forward.|
|Next Article:||Appendix B: Methodology used for the design and analysis of household surveys and data.|