Water Resources Modelling of the Worfe River Catchment and Perak State Using Remote Sensing and GIS
Total Page:16
File Type:pdf, Size:1020Kb
HydroGIS 96: Application of Geographic Information Systems in Hydrology and Water Resources Management (Proceedings of the Vienna Conference, April 1996). IAHS Publ. no. 235, 1996. 605 Water resources modelling of the Worfe River catchment and Perak State using remote sensing and GIS MD AZLIN MD SAID School of Civil Engineering, University Sains Malaysia, Perak Branch Campus, Seri Iskandar, 31750 Tronoh, Perak, Malaysia Abstract Water resources modelling using spatial data inputs is capable of giving good representation of the hydrological processes involved. Two case studies are presented in this paper, one in which an established model was utilized and the other a model being developed. In both case studies, satellite images were used to provide land use/cover information. In the first case study, the model for the Worfe River catchment was used. Landsat MSS data, GIS techniques and Digital Terrain models were utilized to provide the necessary spatial information, such as land cover/use, rainfall and permeable soil distribution to estimate the recharge estimates. Several modelling cases were run which produced mixed groundwater and river flow results. The second case study is to evaluate the water resources availability for the State of Perak in Malaysia. A water resources model is currently being developed to integrate various spatial data inputs. Several problems were encountered in developing this model will be discussed in this paper. INTRODUCTION The use of mathematical models for the simulation of hydrological processes with water resources has become a common approach. In most cases, field data measurements from known locations in the study area are obtained and applied as input to the models. The accurate representation of the processes depends upon the quality of the data inputs, especially its spatial variation. In most cases, the spatial characteristics of hydrological and geographical data were not maintained. It is not the case that spatial data were not available, but problems of data handling proved to be the main problem. Recent advancements in information technology, especially in sophisticated data management have provided modellers the opportunity to improve the accuracy and efficiency of their models. Spatially distributed data can be used in estimating model inputs, updating the model and for calibration purposes (Peck et al., 1982). In this paper, these new techniques were applied for two case studies. Using a tried and tested recharge model developed by Miles (1979), based on the method of Rushton & Ward (1979), techniques were developed to acquire the input data needed for a model in a spatially distributed format. A water resources model is currently being developed for the Perak State, in which spatially distributed information can be used as input. The practicality and adequacy of both case studies are assessed. 606 Md Azlin Md Said CASE STUDY 1 Study area The study area is located in the West Midlands of England. It is a relatively small catchment of 260 km2 and forms part of the Shropshire Groundwater Field. The River Worfe is a left bank tributary of the River Severn. In general, it is an agricultural area with crops, dairy and cattle as the main farming activity. The study area is underlain by Bunter Sandstones, impermeable drift and low permeability strata. Data availability There are altogether nine rainfall stations, a nearby Potential Evaporation station, and 11 borehole locations. Continuous river flow records were available for two sites. Government land use records are available for a number of parishes during the period covered by the model. One cloud-free Landsat MSS image was available for the period of interest. Several blocks of digital elevation data were purchased under licence from Ordnance Survey (UK). Geological maps were available from the British Geological Survey. Recharge model The recharge model was developed as a component in a comprehensive surface and sub surface numerical model developed by Miles (1979). The model is a simple water balance in which recharge is estimated as the difference between precipitation and évapotranspiration. Actual évapotranspiration is based upon the prevailing soil moisture deficit and the root constant function. The calculations are based on daily basis, then aggregated into monthly recharge values. In the original recharge model, average values for land use, impermeable soil area and rainfall were assumed to be representative for the whole catchment. However, to represent the spatial variation of recharge, each node of the surface and sub-surface numerical model was scaled by a factor denoted as the recharge distribution factor. Data processing It was decided that all information would be organized on a rectangular 50-m grid Transverse Mercator projection. This would then provide the best compatibility with the basic data sets. The various layers of data are arrayed in a GIS, in which facilitated the abstraction of information for the recharge model. Data processing was conducted using ERDAS software. Topographic information was extracted from the digital elevation data. Three types of information were extracted, i.e. elevation, slope and aspect. A relationship between rainfall and topography was examined using a multiple linear regression of long average annual rainfall (LARF). It produced the following relationship: Water resources modelling using remote sensing and GIS 607 LARF = 578 + 0.760 (Elevation) + 2.09 (Slope) - 0.098 (AspectXl) r2 = 0.90; r2(adj) = 0.84; s.e.e. = 10.13 mm The Landsat image was classified using the supervised maximum likelihood classifi cation technique. Ground-truthing was established from a postal survey of farmers in the area, the Ordnance Survey maps and visits to the area. Interpretation of ground-truthing was made to facilitate the land cover classification and land use distribution. Pre- and post-processing techniques were extensively used to achieve a sufficient reliability in the results, but unfortunately also introduced the element of subjectivity into the data layer. Classification results, given in Table 1, showed good comparisons. Coupled with close correspondence of classification and ground-truthing gave a high degree of confidence in the use of the classified data to provide estimates of land use at each nodal point. A similar process was performed in classifying the bare soils visible in the bare fields using British Geological Maps as control. The classification showed that the distri bution of impermeable soils are patchy and somewhat different from those shown in the soil maps. The final result indicated that the recharge area is 15 % less than that assessed originally. This finding was not confirmed by a more detailed soil study of the area. Modelling procedures In modelling recharge estimates using spatial databases, three modelling schemes were used, i.e Area, Node and Cell. For Area category, averaged values for the whole catchment were used. For Node category, two schemes were used; Node 1 and Node 2. For type Node 1, values for each 50-m cell were taken and then aggregated to represent the average values for each nodal point. Whereas, for type Node 2, the land use and impermeable soil data layers were integrated to produce a land use/impermeable soil layer. This method of integration eliminates the errors attributed in averaging the individual data cells. In this case, the values for each land use/impermeable soil cell was taken and aggregated to produce an average value for each node which is truly contributing to recharge. For Cell category, each cell value was extracted and modelled. The recharge model was then utilized to model seven modelling cases, based on the three data schemes, to provide external inputs to the surface and subsurface model. In order to investigate the effectiveness of spatial data inputs, the recharge model was also integrated into the surface and subsurface model, in which the recharge distribution factors were removed. Table 1 Land cover classes and permeable soil distribution (Md Azlin Md Said, 1991). Class type Miles Study Grass 32.5% 36.4% Cereal/root-crop 57.1% 52.4% Forest 7.4% 8.0% Urban/water/bare soil 3.0% 3.3% Permeable soil 180 km2 153.9 km2 608 Md Azlin Md Said Modelling results The results for the seven cases for several areas are given in Fig. 1. For Stanmore Cottage and Harrington the results indicate that when recharge was calculated externally, the groundwater levels followed the trend associated with the original model. This is generally due to the optimization process of the model and therefore would not effect the modelling process. However, results for Case 4 showed that the groundwater levels tend to be higher than expected. When the recharge factors were removed and recharge calculated internally in the model, there were some distinct changes in the groundwater levels, as shown in Fig. 2. The results showed some improvements in approximating the trends and patterns of groundwater flow, especially for Stanmore Cottage and Harrington. This indicated that by removing the optimized recharge distribution factor, the results can be improved by using input data obtained from spatial datasets. Modelling cell data inputs was found to be time consuming but gave recharge values that vary from location to location. CASE STUDY 2 Introduction The State of Perak in Malaysia is currently in the process of rapid development as envisaged in aims of Vision 2020. It was estimated that the population in the year 2020 would increase by 61.56% from 1990 census estimates. Similar trends would also be expected for the industrial and agricultural sectors. The expected water demands in the year 2020 was expected to rise by approximately 42% from 1991 figures. To meet the expected demands, a detailed study of the availability of water resources in the state is required. Model development With this in mind, a water resource model was developed for the Perak State Water Resources System, as shown in Fig. 3. The model configuration was developed based on the present water demand and water sources.