Study 2 would concentrate on estimating the 

 improvements in forecasting that would be 

 achieved by adding the new sub-systems con- 

 sidered in Study 1, as well as the potential for 

 forecasting parameters that are not now regularly 

 predicted. It is in this area where the environ- 

 mental scientist must make his strongest contribu- 

 tion. Some work has already been conducted in 

 this field in connection with numerical (i.e., 

 computer) weather prediction,^ to estabhsh cri- 

 teria for required data density and accuracy, where 

 the question has generally been limited to the 

 acquisition of additional data similar to that 

 already acquired, as well as data observing ac- 

 curacy. Theoretical studies of the behavior of 

 errors (i.e., perturbations) in the initial data field 

 during forecast calculations have also been made.* 



Similar analyses have been conducted to esti- 

 mate the density and accuracy of observations 

 required to describe adequately the sea-surface 

 temperature structure. In one study' relationships 

 were obtained between desired accuracy of the 

 analysis, and the density and accuracy of observa- 

 tion. 



At present, much of the expectation for im- 

 proved long-range forecasts is based on the use of 

 data not now available. Attempts should be 

 made to simulate the effects on prediction per- 

 formance of postulated new data. Where it is 

 difficult to adequately test new hypotheses with- 

 out adequate data, data collection experiments 

 should be planned so that the impact of new data 

 on prediction performance can be tested on as 

 smaU a scale as feasible before making a com- 

 mittment to major new data-collection systems. 

 Among the serious questions to be considered is 

 the feasibility of modifying numerical forecasting 



^Alaka, M. A. and F. Lewis, Numerical Experiments 

 Leading to the Design of Optimum Global Meteorological 

 Networks, Technical Memorandum WBTM-7, U.S. Depart- 

 ment of Commerce, ESSA, Washington, D.C., February 

 1967, 14 pp; Panel on Observations Over Sparse Data 

 Regions, Plan for Meeting Meteorological Observation 

 Requirements Over Sparse Data Regions, Technical Plan- 

 ning Study No. 1, U.S. Weather Bureau, Washington, 

 D.C., 1963,51pp. 



^Thompson, P.O., Uncertainty of Initial State as a 

 Factor in the Predictability of Large Scale Atmospheric 

 Flow Patterns, Tellus, vol. IX, No. 3, 1957, pp. 275-295. 



'james, Richard W., Data Requirements for Synoptic 

 Sea Surface Temperature Analyses, Special PubUcation, 

 Naval Oceanographic Office, Washington, D.C., 1967, 29 

 pp. 



techniques to include unconventional data col- 

 lected by platforms now under development; for 

 example, satellite observations which would yield 

 average values of envirorunental parameters over 

 fairly large areas, as opposed to the conventional 

 point observations. 



In many cases, the combined outputs of the 

 first two studies would be extremely valuable. We 

 would be able to relate the projected ability to 

 forecast environmental parameters to proposed 

 characteristics of the expanded environmental 

 monitoring system. We would further be able to 

 make statements relating system performance to 

 increased cost. 



In Study 3 we are faced with the problem of 

 estimating benefits to various segments of the 

 economy resulting from environmental prediction 

 services. This is typically a difficult problem, and 

 the panel has found a limited number of ex- 

 amples.'" The strongest requirement is the de- 

 velopment of a detailed understanding of the 

 operation under study, rather than understanding 

 the behavior of the environment. Thus, to study 

 the impact of improved ocean temperature pre- 

 dictions on the fishing industry, it is more im- 

 portant to understand a particular fishing opera- 

 tion, and the dynamics of the fish population, 

 than it is to understand the physics of the ocean's 

 temperature structure. 



Such studies may be initiated before the results 

 of Studies 1 and 2 are available if reasonable 

 improvements in the environmental prediction 

 "products" are postulated. Such studies must 

 concern themselves with reahzable benefits in a 

 given operation, as opposed to potential benefits. 



Although the considerable difficulty in achiev- 

 ing benefit estimates is recognized, it is equally 

 clear that proposals for increased major invest- 

 ments associated with the expansion of the Na- 

 tional Environmental Monitoring and Prediction 

 Service must be supported by estimates of the 

 type outlined here. Decisions regarding deploy- 



'"a. J. Russo et al.. The Operational and Economic 

 Impact of Weather on the Construction Industry, The 

 Travelers Research Center, Inc.. Hartford, Conn., 1965; 

 Kolb, L. L. and R. R. Rapp, Utility of Weather Forecasts 

 to the Raisin Industry, The RAND Corporation, Santa 

 Monica, California, 1961; Demsetz, H., Economic Gains 

 from Storm Warnings: Two Florida Case Studies, The 

 RAND Corporation, Santa Monica, California, 1962. 



11-66 



