tellers to the editor' Inhomogeneity of Satellite-Derived Climatological Data The era of global meteorology, and specifically of GARP, has highlighted the challenge that our science faces in two of its main objectives: 1) an increase in the time scale of detailed (transient) weather forecasting, and 2) an improve- ment in understanding of climate and climate variability. In each of these objectives the maximum homogeneity of data is important. But whereas the first objective demands spatial homogeneity and relegates temporal homogeneity (at least over a period of years) to minor importance, the second objective reverses this order of priority. The purpose of this letter is to draw attention to certain needs in the second context. Sea surface temperature has long been recognized as per- haps the most important single index and in many respects a generator of climatic anomalies and variations on all time scales. Innumerable past correlative and other descriptive studies, which have relied heavily on this index, have of practical necessity been based on a fraction of the data and therefore may well have suffered loss of accuracy. Thus the publication of the results of the World Meteorological Or- ganization's Historical Sea Surface Temperature Data Project must be eagerly awaited. In this vast project the main mari- time nations of the world are effectively pooling their data resources so that analyzed monthly values of sea surface temperature will be available as far back, and over as much of the world's oceans, as is deemed to be justifiable. Since until quite recently the technique of measurement has changed but little, these more or less homogeneous data should prove a most valuable tool in the study of climatic variability over the last hundred years or so. However, in the case of satellite-derived sea surface tem- perature (SST) data (Brower et al., 1976), several changes in analysis procedure have been made. These relate to the pro- cedure for correction for atmospheric attenuation and that for the retrieval of true surface temperatures in the presence of cloud-contaminated data. Such changes have undoubtedly led to improved absolute accuracy of the current products and to greater compatibility with concurrent and historical conventional data. Unfortunately, they have also introduced more or less systematic changes with time and geographical location that are not real. Changes in the vertical temperature profile radiometer (VTPR) methodology have also occurred, notably concerning the methods by which "first guess" distributions have been derived. These include climatologically based regression re- lationships, statistical relations generated from contemporary conventional sounding data, and forecasts provided by opera- tional numerical prediction models. Although these changes should not substantially affect limited vertical resolution products such as 1000-500 mb thicknesses, systematic differ- ences and biases may well be introduced where implicitly i This section of the BULLETIN is made available to mem- bers who wish to express opinions about problems of concern to the AMS. (For guidelines followed in accepting letters, see "Minutes of the Council," BULLETIN OF THE AMS, 51, p. 40, F4; 51, p. 434, ^7.) The opinions expressed in "Letters to the Editor" are those of the writers and do not represent the official position of the American Meteorological Society. high-resolution data such as standard level temperatures are derived. We therefore urge that attention be given to the need for the retrospective reprocessing of satellite-derived SST and VTPR data by uniform methods so as to obtain the maxi- mum possible length of historically homogeneous data. We also suggest that the raw data, or some intermediate product not sensitive to analysis procedures (such as clear-column brightness temperatures), should be permanently archived for future reprocessing if and when new methods are introduced. There is obviously a cutoff point at which the absolute accuracy of the satellite-derived product is too poor to justify retrospective reprocessing, which would involve only minor corrections. We are not, however, confident that such a point has been reached and are concerned that potentially valuable recent historical data may be destroyed and/or inhomogene- ous data series used without sufficient awareness of the inhomogeneities. Historical climatology is sufficiently difficult without the progressive introduction of new inhomogeneities through the valuable and growing satellite-derived data base. Reference Brower, R. L., H. S. Gohrband, W. G. Pichel, T. L. Signore, and C. C. Walton, 1976: Satellite derived sea-surface tem- peratures from NOAA spacecraft. NOAA Tech. Memo. NESS 78, 74 pp. C . H . B. P RIESTLEY AND A. B. PLTTOCK CSIRO Environmental Physics Research Laboratories Mordialloc, Vic., Australia Radiation constants Readers of the paper on the blackbody radiation function by Widger and Woodall (1976) might note that values of the radiation constants more recent than those given in the paper are available from the National Bureau of Standards (1974). NBS lists the first radiation constant, 2irhc 2 , as 3.741832(20) x 10" 18 W m~ 2 . Accordingly, 2he 2 is 1.191062(6) X 10~ 18 W m -2 sr -1 . The second radiation constant, hc/k, is given as 0.01438786(45) m K. The numbers in parentheses are the one-standard-deviation uncertainties in the last digits. Those who may not wish to do the integration suggested by Widger and Woodall should note that Sargent (1972) refers to a convenient table in which the integration is al- ready done. Although the AMS has not published the table, it does appear on pages 68-71 in the book by Duffie and Beckman (1974). One of Widger's earlier contributions concerned the Wien displacement law (Widger, 1968). The power of the scientific pocket calculator allows one easily to find improved values of the coefficients in these laws. By iteration of a set of seven steps on the HP-35 (after turn on and entry of any number >0, the seven steps are: 1, interchange xy, change sign, e x , —, 5, X), we find the number in Widger's (1968) Eq. 6 to be 4.965114232. . . . When starting with a number larger than the answer, the convergence takes 14 cycles. When starting with the given value 4.965, convergence takes only 5 cycles. By putting 3 in the set of steps in place of 5, we find the 258 Vol. 58, No. 3, March 1977 Unauthenticated | Downloaded 12/04/21 08:39 AM UTC