Wallis, Jim

From History of Hydrology Wiki
Jump to: navigation, search


Jim Wallis


James R. (Jim) Wallis, 1928 (Montreal, Canada) - 2016 (Florida, USA)


Jim Wallis, former Hydrology Section President, AGU Fellow, and the force behind foundational work in synthetic hydrology and flood frequency analysis (most of which appeared in WRR) passed away in Florida on February 13, 2016. Jim was born of English parents in Montreal. His family moved back to England in the 1930s, where he grew up in wartime London before being evacuated to the countryside during the Blitz in the early 1940s. He returned to Canada in 1946 to pursue an interest in forestry, which he studied at the University of New Brunswick, graduating in 1950. In the early 1950s, he worked as a logger in the Queen Charlotte Islands of British Columbia (there being no jobs for degreed foresters). In 1953, he entered the MS program in Forestry at Oregon State University, and went on to do his Ph.D. studies at UC Berkeley from 1958-65, working on the side at Pacific Gas and Electric in San Francisco, where he was motivated to learn about operations research and the use of digital computers. He became interested in the erodibility of forest soils (Wallis and Stevan, JGR 1961, “Erodibility of some California wildland soils related to their metallic cation exchange capacity”) which became his dissertation topic. Following completion of his Ph.D., he was awarded a Bullard Fellowship at Harvard University (a postdoc in today’s terms), from 1965-66 with Harold E. Thomas. He then joined IBM’s Thomas J Watson Research Center in Yorktown Heights, New York in 1966 as part of its nascent Environmental Sciences Program, where he stayed until he retired to join the Yale Climate and Energy Institute.

Hydrological Achievements[edit]

At IBM, where Jim spent most of his career, he was essentially given the charge “do something interesting, ideally having to do with water and/or the environment, and that uses IBM computing.” That was basically the extent of the constraint, hard that it is to believe in today’s world, where even at centrally funded government research labs, there is a fair amount of top-down imposition of “priorities.” Notwithstanding that his Ph.D. research had dealt with forest erosion, shortly after joining IBM, he began to interact with Benoit Mandelbrot, who had been working on a theory of fractals. His interest was piqued by H.E. Hurst’s work on the Nile River, and his book which came out in 1965. Hurst’s work showed that the rescaled range of cumulative departures from the mean of annual Nile flows exhibited a fundamental behavior different than what would be expected from a purely random sequence of flows, or sequences coming from other models (like a low lag Markov) that exhibited “short term memory”. In a seminal 1968 WRR paper “Noah, Joseph, and Operational Hydrology”, Mandelbrot and Wallis showed why the synthetic streamflow models then in use (largely as an outcome of the Harvard Water Project) were unable to reproduce Hurst-type behaviors. They went on to develop a class of self-similar models which they termed Fractional Gaussian Noise which reproduced Hurst-type behavior in synthetic computer experiments. In another seminal WRR paper (1969; “Some long-run properties of geophysical records”) they showed that the Hurst-type behavior was present not only in streamflow sequences, but in many other geophysical records. Their work remains important today (although largely unrecognized in the climate community) – current generation global climate models are unable to reproduce long-term persistence of the type exhibited by key geophysical observations such as precipitation and streamflow.

In 1973, Jim Wallis took a position as advisor at the IBM Scientific Center of Pisa in Italy. In the two years he was at Pisa, he dealt with rainfall-runoff modeling as part of the River Arno hydrological model, and later assumed a major role in a WMO inter-comparison of hydrological conceptual models (published 1976).

In the mid-1970s, he turned his attention to flood frequency estimation. A 1975 WRR paper “Regional skew in search of a parent” with USGS scientists Nick Matalas and Jim Slack showed that the relationship between the mean and standard deviation of regional estimates of skewness for annual maximum streamflow data from the western United States could not be explained by corresponding relationships for the conventional frequency distributions. Instead, the relationships for the observations exhibited what they termed the Condition of Separation’, a characteristic of heavy tailed behavior. He investigated the behavior of the Wakeby distribution (a new distribution suggested by Harvard’s Harold Thomas) that could mimic the Condition of Separation. The Wakeby distribution is expressed in inverse form, and hence does not lend itself to parameter estimation using conventional moments or maximum likelihood methods. Jim worked with statistician J.A. Greenwood, as well as Matalas and USGS scientist J. M. Landwehr, to develop the method of Probability Weighted Moments (PWMs), which was attractive in that the fitting method was based on order statistics, and had good small sample properties, as demonstrated in two 1979 WRR papers with Landwehr and Matalas. The Generalized Extreme Value (GEV) distribution, then in use in the UK, conveniently also lent itself well to PWM estimates, and working with J.R.M. Hosking at the U.K. Institute of Hydrology in the mid-1980s (and later at IBM T.J. Watson Research Center, to which Hosking moved), they developed regional estimation procedures based on L-Moments, which are linear combinations of PWMs. Their work fundamentally changed the field of regional frequency analysis to the extent that a 1993 WRR paper by Vogel and Fennessey was titled “L-moment diagrams should replace product moment diagrams“, on the basis that product moments were subject to substantial bias and variance. A 1997 book by Hosking and Wallis provides a complete treatise on regional frequency analysis that remains the key reference on the subject, both in research and practice. L-moment methods are now used in the U.K. (e.g., the Institute of Hydrology’s 1999 Flood Estimation Handbook), by the U.S. Army Corps of Engineers in its U.S. National Drought Atlas, and by NOAA in its ongoing upgrade of U.S. precipitation Intensity-Duration-Frequency relationships.

Not nearly as well known in hydrology was his early forest growth modeling work with Botkin (then at Yale). Yet his most cited paper (Botkin, Wallis and Janak, J. Ecology 1972, “Some ecological consequences of a computer model of forest growth”, cited over 800 times) was on that topic. The paper describes the first computer model to reproduce the population dynamics of trees in a mixed-species forest stand. A companion paper, also in 1972, also with Botkin and Janak, further described technical aspects of their forest growth model.

Jim was President of the Hydrology Section from 1980-82. Among his many contributions to the Section, two stand out. The first relates to the Horton Research Grants. Jim was aware that on his death in 1945, Robert E Horton had made a substantial donation to AGU that was intended to promote hydrology interests. Yet AGU had essentially comingled the funds, which were not identified with the Section. Spirited discussions with then Executive Secretary Fred Spilhaus resulted in an agreement that if the Section were to have a set of bylaws that directed how the money would be spent, AGU would separate out the Horton donation and designate it to the Section. Jim and others set about to draft Section bylaws (now posted on the Section web site) that included a provision for the Horton Research Grants, the first of which was awarded in 1983. A second area where Jim made lasting contributions was to the protocols for selection of AGU Fellows. At the time, the election of Fellows was more or less that if Joe and Harry were sufficiently influential, and they said Sam was a good person, Sam was elected. Almost invariably, in addition to the fact that Harry, Sam, and Joe were all males, they came from areas other than hydrology. Jim was vociferous in his role as an AGU Council member that a more even handed process had to be implemented. The result of those efforts was a set of requirements for the nomination and consideration of Fellows that is the predecessor to those used today. In 1982, five hydrologists were elected as Fellows, in comparison with a cumulative total of 13 in all prior years!

Together with Denis Lettenmaier and Eric Wood, Jim also put together the first collection of quality assured (as far as was possible) open source data set on precipitation and streamflow for the United States.

Reference Material[edit]

Taken from the obituary prepared by Dennis Lettenmaier, Enda O'Connell, Ezio Todini, and Eric Wood for the AGU Hydrology Section Newsletter and Italian Hydrological Society Newsletter

Major Publications[edit]


Hosking, J.R.M. and Wallis, J.R., 2005. Regional frequency analysis: an approach based on L-moments. Cambridge University Press.


Mandelbrot, B.B. and Wallis, J.R., 1968. Noah, Joseph and Operational Hydrology. Water Resources Research, 4(5), pp.909-918.

Mandelbrot, B.B. and Wallis, J.R., 1969. Some long‐run properties of geophysical records. Water Resources Research, 5(2), pp.321-340.

Mandelbrot, B.B. and Wallis, J.R., 1969. Computer experiments with fractional Gaussian noises: Part 1, averages and variances.Water Resources Research, 5(1), pp.228-241.

Mandelbrot, B.B. and Wallis, J.R., 1969. Computer experiments with fractional Gaussian noises: Part 2, rescaled ranges and spectra. Water Resources Research, 5(1), pp.242-259.

Mandelbrot, B.B. and Wallis, J.R., 1969. Computer experiments with fractional Gaussian noises: Part 3, mathematical appendix. Water Resources Research, 5(1), pp.260-267.

Mandelbrot, B.B. and Wallis, J.R., 1969. Robustness of the rescaled range R/S in the measurement of noncyclic long run statistical dependence. Water Resources Research, 5(5), pp.967-988.

Wallis, J.R. and Matalas, N.C., 1970. Small sample properties of H and K—Estimators of the Hurst coefficient h. Water Resources Research, 6(6), pp.1583-1594.

Matalas, N.C. and Wallis, J.R., 1971. Statistical properties of multivariate fractional noise processes. Water Resources Research, 7(6), pp.1460-1468.

Botkin, D.B., Janak, J.F. and Wallis, J.R., 1972. Some ecological consequences of a computer model of forest growth. The Journal of Ecology, pp.849-872.

Botkin, D.B., Janak, J.F. and Wallis, J.R., 1972. Rationale, limitations, and assumptions of a northeastern forest growth simulator. IBM Journal of Research and Development, 16(2), pp.101-116.

Matalas, N.C. and Wallis, J.R., 1973. Eureka! It fits a Pearson type: 3 distribution. Water Resources Research, 9(2), pp.281-289.

Matalas, N.C., Slack, J.R. and Wallis, J.R., 1975. Regional skew in search of a parent. Water Resources Research, 11(6), pp.815-826.

Slack, J.R., Wallis, J.R. and Matalas, N.C., 1975. On the value of information to flood frequency analysis. Water Resources Research, 11(5), pp.629-647.

Wallis, J.R., Matalas, N.C. and Slack, J.R., 1977. Apparent regional skew. Water Resources Research, 13(1), pp.159-182.

Landwehr, J.M., Matalas, N.C. and Wallis, J.R., 1978. Some comparisons of flood statistics in real and log space. Water Resources Research, 14(5), pp.902-920.

Greenwood, J.A., Landwehr, J.M., Matalas, N.C. and Wallis, J.R., 1979. Probability weighted moments: definition and relation to parameters of several distributions expressable in inverse form. Water Resources Research, 15(5), pp.1049-1054.

Landwehr, J.M., Matalas, N.C. and Wallis, J.R., 1979. Probability weighted moments compared with some traditional techniques in estimating Gumbel parameters and quantiles. Water Resources Research, 15(5), pp.1055-1064.\

Landwehr, J.M., Matalas, N.C. and Wallis, J.R., 1979. Estimation of parameters and quantiles of Wakeby distributions: 1. Known lower bounds. Water Resources Research, 15(6), pp.1361-1372.

Hosking, J.R.M., Wallis, J.R. and Wood, E.F., 1985. Estimation of the generalized extreme-value distribution by the method of probability-weighted moments. Technometrics, 27(3), pp.251-261.

Hosking, J.R.M., Wallis, J.R. and Wood, E.F., 1985. An appraisal of the regional flood frequency procedure in the UK Flood Studies Report. Hydrological Sciences Journal, 30(1), pp.85-109.

Wallis, J.R. and Wood, E.F., 1985. Relative accuracy of log Pearson III procedures. Journal of Hydraulic Engineering, 111(7), pp.1043-1056.

Hosking, J. and Wallis, J.R., 1986. The value of historical data in flood frequency analysis. Water Resources Research, 22(11), pp.1606-1612.

Hosking, J.R.M. and Wallis, J.R., 1986. Paleoflood hydrology and flood frequency analysis. Water Resources Research, 22(4), pp.543-550.

Hosking, J.R. and Wallis, J.R., 1987. Parameter and quantile estimation for the generalized Pareto distribution. Technometrics, 29(3), pp.339-349.

Lettenmaier, D.P., Wallis, J.R. and Wood, E.F., 1987. Effect of regional heterogeneity on flood frequency estimation. Water Resources Research, 23(2), pp.313-323.

Hosking, J.R.M. and Wallis, J.R., 1988. The effect of intersite dependence on regional flood frequency analysis. Water Resources Research, 24(4), pp.588-600.

Hosking, J.R.M. and Wallis, J.R., 1993. Some statistics useful in regional frequency analysis. Water Resources Research, 29(2), pp.271-281.

Guttman, N.B., Hosking, J.R.M. and Wallis, J.R., 1993. Regional precipitation quantile values for the continental United States computed from L-moments. Journal of Climate, 6(12), pp.2326-2340.

Lettenmaier, D.P., Wood, E.F. and Wallis, J.R., 1994. Hydro-climatological trends in the continental United States, 1948-88. Journal of Climate, 7(4), pp.586-607.

Handcock, M.S. and Wallis, J.R., 1994. An approach to statistical spatial-temporal modeling of meteorological fields. Journal of the American Statistical Association, 89(426), pp.368-378.

Hosking, J.R.M. and Wallis, J.R., 1995. A Comparison of Unbiased and Plotting‐Position Estimators of L Moments. Water Resources Research, 31(8), pp.2019-2025.


AGU Hydrology Section Obituary