Costs and cost effectiveness

In addition to determining the effectiveness of SAPHYDATA it is also necessary to determine costs. The ultimate decision to implement a particular SAPHYDATA in a given situation may be guided by the ‘cost effectiveness ratio’. In effect, this describes the cost per unit of information in the total system. The steps to evaluate this ratio follow in sequence those of the costing procedure:

1. Compute implementing costs for the SAPHYDATA;

2. Determine operating costs;

3. Determine total costs;

4. Compute cost effectiveness ratios for each system;

5. Rank systems according to cost effectiveness.

2.5.4.1 Compirting cost efectiveness

The design of the programme should be based on specific goals including acceptable accuracy levels selected to ensure an efficient relation between cost and information.

Regional differences in these goals may be expected because of the variation in hydrologic conditions and the need for information. The goals and the items listed under each goal as

used herein can be modified to fit conditions in a particular region (see Table 2.3).

The goals should include acceptable accuracy levels because the accuracy goals not only govern cost and techniques used in providing information, but also provide a measure of the attainment of specific goals. Ideally, the accuracy goals would be set by knowledge of the worth of accuracy of data in design and planning and the cost of attaining a certain level of accuracy in the data programme. Lacking this knowledge, it is still useful to arbitrarily set accuracy goals by comparison with the accuracy that could be obtained from a streamflow record of reasonable length.

For a general range of conditions a higher degree of accuracy can be justified for current- purpose data used in operation of water systems than for data to be used in design of water development projects. Design is based on the probability of future occurrences with the associated wide confidence intervals due to the relatively short-time sample available.

O n the other hand, the operation of water systems deals with known volumes of water and higher accuracy can be of great economic benefit.

Guidelines for evaluating and planning the surface-water data programme in a region are given in this section. The primary elements of such a study are (a) setting the goals of the programme; (b) evaluating the existing information in relation to the goals, and (c) considering alternative action to improve the existing programme. The guidelines are organized by ‘type of data‘ and follow the pattern that could be used in a typical study.

At many sites a continuous record of current information on streamflow is needed for day-by-day decisions on water management. The stations at these sites which are here called ‘current-purpose streamflow data’ stations should be identified as the first step in the study and the purpose of operation clearly stated. In general, this part of the data programme cannot be designed, and its character changes frequently in response to changing needs. Although the data obtained at many of these stations may not have general hydrologic significance, the data are needed to manage the water and to assess the effects of development. The accuracy goal will depend on the requirement of the particular management system, and can be met by intensified observations or by more sophisticated instrumentation as needed. Important uses of current purpose data are listed on p. 51.

TABLE 2.3 Tentative design. National Water Data Network (United States)

Accuracy

Objective Accounting objective

of mean)

procedures (percentage

Stream discharge (a) Sites:

To record streamflow moving from region to region, from one accounting unit to another and, leaving each accounting unit, to bordering nations or coastal waters

1. Stations on streams draining each accounting unit, and principal diversions from and to, where

appropriate, or

stations, correlate ungauged streams with gauged streams.

2. where infeasible to maintain required

(b) Determine, for at least 90 per cent of the area of each unit:

1. Mean annual discharge 5 3 2. Mean monthly discharge +5

Description of different systems characteristics

I. Assessment of current water conditions. (Examples: the monthly forecasts of stream- flow and assessments of extreme drought of flood conditions.)

2. Operation of storage reservoirs for hydroelectric power generation, flood control, water supply, and maintaining water levels for recreation or navigation. (Examples: the allocation of water among users from a stream or reservoir for water supply, the programming of release of water for power production at various dams in a stream and reservoir system, and the storage and relessz of flood water.)

3. Forecasting flood peaks, low flow, and runoff from snowmelt. (Examples: forzcasts of flood discharges and stages, forecasts of low flow on monthly or seasonal basis, and

‘verification’ of forecasts of runoff from snowmelt.)

4. Disposal of contaminants. (Example: release of waste from storage lagoon in propor- tion to flow of stream.)

5. Monitoring water quality. (Example: a simultaneous record of discharge is needed at sites where water quality is continuously monitored for current purposes.)

6. Measurement of sediment loads. (Example: a simultaneous record of discharge is needed at sites where sediment loads are currently measured.)

7. Meeting treaty or legal requirements. (Example: accounting for annual delivery of water under international agreements.)

8. Research or special studies.

The need for current information requires continuous measurement of streamflow. The monitoring of lake and reservoir levels, diversions, and return flow may also be required.

Although there is usually only a limited degree of latitude for selecting locations, and for control of observation periods, some gauges operated to provide current information will provide data useful in regional definition of certain streamklow characteristics. The statistical characteristics that are unaffected by regulation or diversion should be identified at each station operated primarily for current information. For example, it may be noted that flood flows are representative of natural conditions, but that flows below a certain value are not. Such notations serve to identify those stations that can be used in regional studies of streamflow characteristics.

Designers and planners of water-related facilities increasingly utilize the statistical characteristics of streamflow rather than flow at specific times.

In order to appraise the relation between accuracy likely to be required and accuracy that can reasonably be obtained, the following description of goals relating to stations operated to define the statistical characteristics of streamflow can b: broken down into three goals. The first goal is for all unregulated streams, the second is for principal streams, and the third is for regulated streams. The rationale for this classification is based on the approach used to attain the goals. These approaches in corresponding order are (a) regionalization of data, (b) operation of gauging stations at selected sites, and (c) system studies.

C. H. Hardison (Hardison, 1968) proposed an accuracy standard which could be obtained from an equivalent of 10 years of record for the first and third objectives, depending on the size of the stream, and an equivalent 25 years of record for the second objective. Using this concept, the accuracy figures in percentages would vary from region to region, and would be determined by the theoretical relation of the standard error to an index of variability and number of years of record. As an example, the accuracy figures given are equivalent to what could be obtained from 10 to 25 years of record on a typical stream.

This concept of accuracy was proposed as an expedient in lieu of more sophisticated ways of obtaining uniform ratios between incremental information and incremental cost as developed by Matalas and Huzzen (1968).

The first objective of a water resources data programme might be to define the following streamflow characteristics within the stated accuracy for all natural unregulated streams in the region. The accuracy figures represent standard error in percentage of the item. The

accuracy shown is equivalent to that obtained from 10 years of measurement of a natural stream.

Item Accuracy

1.

2.

3.

4.

5.

6.

I.

Mean annual discharge

Mean monthly discharge (average)

Standard deviation of monthly and annual discharge 50-year flood

Median annual I-day low flow 20-year, I-day low flow 50-year, I-day high flow

(percentage) 10 25 20 30 15 25 25 2.5.4.2

The establishment of a water resources data processing centre involves the inventorying of an available in-country resource. These resources include existing data collection pro- grammes, user requirements for information, the availability of computer processing systems, and the current allocation of manpower to manage and maintain the system.

In addition, the inventory should include projections of future workload to meet national objectives. The allocation of resources to data collection programmes has a definite relation to the allocation of funds for water resources development. The planning of data information systems has to take place within the framework of total national resource development. The following examples of planning for a processing centre were abstracted from a W M O report (Tsherwood, 1969) and illustrate the application of cost-benefit techniques to arrive at the proper mix of manpower and hardware to perform the various data processing functions efficiently and effectively.

Evaluation of the manpower resources and availability of computers

2.5.4.3

The decision on where to start the application of a water resources information system is a difficult one to make. The following example illustrates one technique that was used by W M O to assist local authorities in the Republic of Kenya to improve part of their water resources data information system. A plan for an extensive and easy-to-use manual on data collection was elaborated; during this elaboration it became clear that much man- power was needed and that it was necessary to start thinking seriously about the possi- bilities of automatization; the investments necessary for automatization would, however, be large and the higher efficiency of automatization would only appear after several years;

the final solution adopted was an optimal mixture of manpower and computer hardware.

Several areas of processing streamflow data were considered and subjected to cost-benefit analyses. These included: plotting of discharge measurements, plotting of hydrographs, listing data for publication, correcting and estimating missing periods of records. Only one example of the cost-benefit analyses will be given in this report. The reader is referred to the full report for the reasoning and details of the development of the automated processing centre.

Planning and evaluating new injormation systems

2.5.4.4 Plotting of discharge measurements

Plotting discharge measurements is the first step in analysing stage discharge rating conditions at a river station. Although the manual process of plotting points is not a difficult one, the number of points to be plotted is large. At the rate of 12 measurements a year for an average of 10 years per station for about 200 stations, this amounts to more than 20,000 points. Although there is no true plotting equipment on the computer available

Descviption of ([ifleevent systems chavuctevistics

to the project, a technique can be used to make a rough plot using the computer line printer. Such a plot consists ofa series ofprinted characters (the character ‘x’ for example) in the print line and print position most nearly corresponding to a plotted point in the correct position.

The limitations are that the accuracy of the plotted point is only to the nearest print position (0.1 inch) in the horizontal range, and the nearest line (1/6 inch) in the vertical range. Even such gross plotting could be extremely helpful in the initial evaluation of the rating conditions of a river station. Justification for computerizing this particular process is shown below, It must be pointed out that the output product of the computer is not the same as that from manual plotting, but it essentially fulfils the same function.

Costs b.v cornpiiter methods U.S.$

Programme preparation and testing 150

Card punching 20,000 cards at $10 per 1,000 cards (20 columns punched per card) 200

Card verifying (same as punching) 200

Computer time-2 hours at $75 per hour 150

Miscellaneous (supplies, etc.) 200

Contingencies (corrections, reruns, etc.) 15

$975 Costs by iiiariiml methods

At 2 days of manual work to plot measurements for one station, 200 stations would Assuming 250 working days per year per man at $2,500 per year for Grade TI employees,

-

require 400 working days.

the total cost by manual methods is $4,000

This specific project will generate large masses of data, much of which needs some manipulation or processing to obtain meaningful figures. The resulting refined data must be published and prepared for use in later complex analyses. Therefore, it was recom- mended that digital computer techniques be used, both for the purpose of gctting more work accomplished within the funds available and for making it practical to do much more analysis of the data than could be done by manual methods during the life of the project.

However, many manual steps are still involved, so a summary is shown below to indicate where the computer processes fit in among the manual ones.

Item Muniial Computer