Skip to main content

PDF


Title: Testing and Further Development of RIVPACS: Stage 4
Author: D D Hornby
Author: R T Clarke
Author: J F Wright
Author: F H Dawson
Author: Environment Agency
Document Type: Monograph
Annotation: Environment Agency Project ID:EAPRJOUT_1605, Representation ID: 10, Object ID: 1340
Abstract:
RIVPACS requires information on a small suite of environmental variables in order to make predictions of the macroinvertebrate fauna to be expected at a given site with stated environmental features in the absence of environmental stress. The Option 1 suite of variables has remained the same in RIVPACS II, (available in 1990) and in RIVPACS III/RIVPACS III+ (used in the 1995/2000 General Quality Assessments). Some of the current variables including slope of site, altitude of site and distance of site from source, together with mean annual discharge category are acquired manually from maps. This is a time-consuming process which can be prone to error. In this package, we have examined the feasibility of developing procedures for acquiring accurate values of these predictor variables from a Geographic Information System (GIS). We have also investigated whether additional site variables acquired from the GIS are capable of increasing the accuracy of RIVPACS predictions. The software ArcView, a Windows-based GIS, was used in conjunction with the ablue linea river system of Great Britain, as shown on 1:50000 scale Ordnance Survey Landranger maps to create a novel procedure for the extraction of environmental data. This involved two separate processes, each one of which was very time-consuming. First, it was necessary to check and edit a variety of errors in the original digitised blueline network for each individual hydrometric area throughout Great Britain. Second, the decision was taken to aprocessa the entire network by incorporating information on the routes to the source and mouth of each river from any given arc on the network. This information is essential for calculating such features as distance to source and slope. By undertaking this mammoth task prior to routine use of the purpose-built system for extracting site information, the query process to automate estimation of the values of the various variables became much more efficient. Some problems remain. At present, site grid references are only given to 100m resolution; this is not always sufficiently accurate to automatically associate the site with the correct river location on the blue-line network. Therefore each new biological site needs to be manually assigned to the correct river location on the network before the GIS can be used to derive environmental data for the site. In a small number of lowlying areas of the country, the grid-like drainage pattern can not currently be resolved by the GIS in order to calculate distance to source, slope etc. Hence environmental variables at 49 of the 614 RIVPACS reference sites were unobtainable using GIS, and the original map-derived RIVPACS values for these variables were used to maintain a full dataset in subsequent multiple discriminant analyses (MDA). Values of current RIVPACS predictor variables acquired from the GIS and from maps were compared for the RIVPACS reference sites. Careful manual re-checking found that the use of GIS to obtain altitude and distance from source was generally more reliable, in addition to being quicker. Different procedures, often involving different distances up- and down- stream, were used to obtain slope of site by the GIS and the manual map method, resulting in a limited number of substantial changes in slope estimates. There were further practical problems in obtaining discharge category from the GIS, because this layer was supplied at a different scale to the blue line and it had to R and D TECHNICAL REPORT E1-007/TR1 ii be loaded on top of the blue line network before manually selecting the site and hence the discharge category. However, over 90% of sites were within one discharge category of the original map-read category. The GIS-derived values of these four variables were then used with the remaining RIVPACS variables in MDA to determine whether, overall, they gave improved ability to predict the biological group of each of the 614 RIVPACS reference sites. The GISderived values are assumed to be estimated with greater accuracy, but they failed to improve on current ability to predict the biological group of the sites. GIS software procedures were developed to derive new environmental predictor variables for any river site. These were: altitude of source, slope to source, a measure of stream power, upstream catchment area, the proportion of the upstream catchment covered by each major RHS solid and drift geology class and the solid and drift geology class of the 1 km square containing the site. When assessed on the RIVPACS reference sites, none of these GIS variables provided any significant improvement in ability to predict site group and hence expected fauna when used with the existing RIVPACS predictor variables. There was only some minor improvement when adding stream power (a function of discharge, stream width and slope at site). However, the automated GIS procedures identify an incorrect upstream catchment for a small percentage of sites (5-10%); developing GIS procedures to amanuallya correct such site positions is beyond the scope of this project, but merits further investigation. In the long-term, it would appear that the use of GIS-derived values of some RIVPACS predictor variables will bring benefits in terms of speed of acquisition, greater accuracy and a modest increase in predictive capability. However, the consequences of implementing such a change on the RIVPACS software itself and on all Ecological Quality Index (EQI) and other results generated by RIVPACS are such that it should be carried out as a single operation and only when there is clear evidence that progress in generating reliable outputs from the GIS is complete. We also propose that there is a need for a Windows version of RIVPACS before additional changes are made to the current software. R and D TECHNICAL REPORT E1-007/TR1 iii CONTENTS ACKNOWLEDGEMENTS i EXECUTIVE SUMMARY ii LIST OF FIGURES v LIST OF TABLES vi 1. 1.1 1.2
Publisher: Environment Agency
Subject Keywords: Geographical information systems; Biological monitoring; RIVPACS; Macroinvertebrates; Environmental variables
Extent: 62
Permalink: http://www.environmentdata.org/archive/ealit:4392
Total file downloads: 9

Download PDF    Display PDF in separate tab