The quality of a numerical model or computer calculation can be assessed objectively by determining its conformance to the specified requirements, that is, the ‘fitness for purpose’ of the model or calculation can and should be formally checked, approved and documented at all stages of its design and application. These concepts are the basis of quality assurance requirements which will be contained in contract specifications more and more frequently in the future. QA procedures will make mandatory, within any contract, rigorous levels of documentation, formal checking of methodology and comprehensive validation tests. Non-conformance with procedures, as identified by a technical audit, will result in a work stoppage with consequential losses.
This chapter aims to draw attention to these requirements by describing a system of organisation for program development which is based on the relevant British Standards. This system is designed to form the basis for QA procedures, allows for quality control and fits in with good project management methods.
If such a system is followed, then adequate documentation will be generated, clear logical code will be produced and contractual requirements will be fulfilled with the minimum of wasted (i.e. loss-making) effort.
An additional benefit arises from the application of this system in the case of a complex numerical model which, by its very nature, cannot be validated for all circumstances. In such cases, the limited validation possible plus the comprehensive documentation, complete methodology statements and clear code will assure the quality auditors that all possible care has been exercised in its development.
This chapter is procedural, not technical. It attempts to define the meaning of quality as applied to software models of the environment, and why the concept should be widely used. The chapter outlines how quality can be achieved and when quality methods should be applied. It is important also to point out who needs to take the initiative to improve quality, and how quality methods may effect the competitiveness of the UK service and software industries in engineering and the environment sciences. The main thesis is that quality methods should be more widely applied in scientific software because this will benefit both the funding agency (the client) and the scientist or engineer (contracted supplier of software). The benefit will come in several ways: in financial terms, from the increased reputation of the supplier and from the confidence placed in the methods used for the computer modelling of the environment.
The benefit of adopting quality methods in the software industry generally are becoming increasingly recognized (Miller, 1986); these benefits are also available to the producers and users of numerical models of the environment.
By definition, ‘quality’ appears to be a qualitative concept which cannot readily be quantified. However, quality, as used in this chapter, implies ‘fitness for purpose. Quality is ?the totality of features and characteristics of a product or service that bear on its ability to satisfy a given need’ (BSI, 1979a).
The quantification of the quality of a product is achieved by formal.