A Brief History of * \

The Robert E. Horton Lecture

Rafael L. Bras Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts

1. Introduction 2. From the Greeks to Horton

I am grateful to the American Meteorological So- My family and I returned to the af- ciety for bestowing on me the honor of being the ter a year's sabbatical in 1983. A not-so-friendly im- Horton Lecturer. The award honors Robert E. Horton, migration officer in New York detained us for over 40 a truly multifaceted individual whose curiosity and minutes wondering what had we done over the past genius led him to blaze the trails in hydrology and 12 months, suspicious of the trips to China, Europe, that many of us have followed. and South America. I patiently explained that I was a Horton wrote about , runoff production, hydrologist and lectured in all those places. A quizzi- basin response, and fluvial geomorphol- cal look was followed by the question, "What is a hy- ogy, evaporation, and hydrometeorology (Horton drologist?" After a carefully crafted explanation she 1932,1933, 1940,1945). Everything he did was pro- gave me an incredulous look and asked, "Why would found and thorough, setting standards, hypotheses, and anybody care about that?" theories that are still debated. But Horton was not an Many have cared about hydrology, dating back to ivory tower . All his work was motivated by great civilizations in China, the Middle East, Greece, and the problems of society. For almost 30 years I have Rome. Early thinkers and philosophers did not under- followed Horton's delight in the variety of challenges stand three basic hydrologic principles (Eagleson 1970): that hydrology poses. Like him I have dabbled in prac- tically all elements of physical hydrology. And like 1) conservation of mass, him, I have had a wonderful time! Hydrology has 2) evaporation and condensation, and changed and evolved dramatically over my career. 3) infiltration. The following will be a short history of hydrology. It will also be a biased history, colored by my experi- They were worried about how gets up to the ences, interests, and work. I make no claim of com- mountains, flows down to the , and fails to raise the pleteness or of neutrality. I do claim pride in being a level of the latter. Because of what may be called lim- small part of what I think have been the best years of ited spatial awareness, they could not see rainfall as a hydrologic science. sufficient source of flow. To account for ob- served water behavior, underground reservoirs (be- neath mountains) were hypothesized. Water was believed to be pushed up the mountains by vacuum Corresponding author address: Dr. Rafael L. Bras, Room 1-290, forces, , or "rock pressure," surfacing Department of , Massachusetts Institute of Tech- as stream flow. These underground reservoirs were nology, Cambridge, MA 02139. E-mail: [email protected] replenished by the sea. In final form 12 March 1999. , during the first century B.C., stated that ©1999 American Meteorological Society the mountains received that then gave rise

Bulletin of the American Meteorological Society 1151

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC to stream flow. A filtration process by which water tion of river basin behavior. Horton's ideas on infil- percolated into was also acknowledged by tration, soil-moisture accounting, and runoff are still Vitruvius and later by da Vinci. recognized by present-day hydrologists. It was in the seventeenth century that Perrault proved by measurement that precipitation could ac- count for stream flow in the Seine River, France. 3. Thoughts on the last 30 years Similar quantitative studies were made by Mariotte and Halley during this historical period. At this stage, By the time I arrived on the scene, as a student in the mass balance concept was pretty established, 1968, generations of engineering hydrologists had although questioning of it continued well into the been educated using the pioneering textbooks of twentieth century. The eighteenth century saw advances in hydraulics and the mechanics of water movement by Bernoulli, Chezy, and many others. The nineteenth century saw experimental work on water flow by people like Darcy and Manning. The above names are familiar to stu- dents of and surface-water movement. Until the 1930s hydrology remained a science filled with empiricism, qualitative descriptions, and little overall understanding of ongoing processes. At that time, people such as Sherman (1932) and Horton (1940) initiated a more theoretical, quantitative ap- proach. Sherman's unit concept still re- mains with us as the most successful (but not necessarily the best) and most well-known explana-

FIG. 2. High-resolution in situ observations of the soil-moisture field in the small (0.1 km2) Tarrawarra experimental catchment; southeastern Australia demonstrates the role of competing hydro- logic processes in defining spatial structures in the soil-moisture field (Western and Grayson 1998). (top) The volumetric moisture content within the top 30 cm of soil during a typical wet season day (27 Sep 1995). In this case the soil-moisture pattern is orga- nized by elevation. The wetter zones follow elevation con- tours and depressions. High of moist allows significant redistribution along topographic gradient. During a typical day in the dry season (22 Feb 1996; conditions shown in the bottom panel), there is no discerned cor- respondence between topography and . Dry soil moisture has lower hydraulic conductivity and lateral redistribu- tion is less dominant as a factor in organizing soil moisture pat- terns. The random spatial variability of may be a more critical factor in explaining any spatial variability in the relatively dry soil moisture field. Lateral redistribution in the saturated and FIG. 1. Moisture content vs depth profiles for (a) the Horton unsaturated zones, groundwater and surface-water interaction, mechanism and (b) the Dunne mechanism. Overland flow , macropore flow, and soil texture heterogene- generation for (c) the Horton mechanism and (d) the Dunne ity are among the competing factors that influence the spatial or- mechanism (Freeze 1980). ganization of surface soil moisture fields.

1152 Vol. 80,, No. 6, June 1999

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC Linsley et al. (1958, 1982). Their Hydrology for En- empirical results or data may change regionally or gineers was one of the few, and overwhelmingly the geographically, the fundamental principles governing dominant, hydrology textbooks in the United States. It hydrologic processes, when defined in mathematical is probably the historical best seller in the field, by far. terms, do not vary, and therefore have general appli- Ven Te Chow's (1964) encyclopedic Handbook of cation." The era of pure empiricism was over. Finally, Applied Hydrology seemed to codify a "mature" field. in 1970 Peter S. Eagleson wrote Dynamic Hydrology. To students like me, the field was lacking new ideas. A best-seller it was not, but its impact on the genera- Luckily, my impressions were wrong; a revolution tions of hydrologists to follow was enormous. was brewing and it raised its head around 1970. Eagleson's work screamed science and emphasized Crawford and Linsley's (1966) work on the Stanford that the hydrology of the land was intertwined with Watershed Model and Harley et al.'s (1970) MIT the atmospheric phenomena. Catchment Model showed that digital computers There were other key actors and other key events, but offered hydrologists the opportunity to integrate pro- to me the above were pivotal to launching the hydrologic cesses to simulate complicated behavior in a system- revolution of the last 30 years. Things have changed so atic, integrated fashion. Work at the Harvard Water much that in a recent speech (1998) I joked that it seemed Program and at Colorado State University by that all hydrology I learned at MIT was wrong. An ex- Yevjevich and students began viewing natural hydro- aggeration, but it gets the point across! Several new logic processes as random phenomena and represent- principles, realizations, concepts, tools, and method- ing them in that fashion. The International Hydrologic Decade had been on-going for five years and a wealth of new data and group efforts were becoming avail- able. One such result, The Handbook on the PRIN- CIPLES of Hydrology (capitalization added by the author), edited by Donald M. Gray (1973), stated in its foreword "Hydrology is only an 'infant' in growth in the modern-day family of sciences . . . our ancient forebears, if they could but see us now, would be shocked to find how lax we have been in neglecting the study of water." It further claimed that "although

FIG. 4. Each of a number of different river basins is presented in the plot by two points: a circle for 1987 data and a triangle for 1988 data. The abscissa represents "model error," i.e., it represents the average runoff generated for the basin by nine land surface parameterizations (LSPs) minus the observed runoff there. The LSPs were forced with observed precipitation, radiation, and other atmospheric variables for 1987-88 as part of the Global Soil Wetness Project. The ordinate shows the density of gauges within the basin, a measure of the accuracy of the precipitation FIG. 3. Typical input and output of the distributed model by forcing used by the LSPs. The salient feature of the plot is the Garrote and Bras (1995a,b) illustrating the state of various significant reduction in model error as the density of rain gauges hydrologic variables at a particular time in a simulation over the increases. This suggests that much of the overall error in model Little Washita River Basin. Shown are (a) rainfall, (b) water runoff may not be related to flows in the model depth, (c) soil moisture content above , (d) and runoff parameterizations but to an inadequate coverage of rain gauges production. (Okietal. 1999).

Bulletin of the American Meteorological Society 1153

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC ologies that have their roots in this revolution have domi- nated hydrology over the last 30 years and will domi- nate it in the foreseeable future. These are the following.

Hydrologic processes are not only incredibly vari- able in time and in space but the properties and parameters of the media in which they evolve are also extraordinarily variable. The old representa- tions of hydrologic behavior within idealized ho- mogeneous media, in zero or one dimension, can lead to serious misconceptions and errors. Hydrologic phenomena can and should be repre- sented, and interpreted, as products of stochastic dynamics. Uncertainty is inherent due to extreme variability and issues of scale (see below). This thinking goes way beyond traditional statistical analysis of data. Process representation and scale are inseparable. Some processes are scale dependent; measurements always are. Other processes are scale independent, FIG. 5. Four days of derived soil moisture from microwave data. Note the progressive drying and the complicated patterns resulting exhibiting various degrees of self-similarity. from the interaction of precipitation and surface properties Complexity is the rule, not the exception. This (Jackson et al. 1999, manuscript submitted to IEEE Trans. Geosci. complexity can be exhibited even in simple hydro- Remote Sens.). logic systems, as nonlinearities and feedbacks be- tween interrelated phenomena are acknowledged or discovered. • The atmosphere- and the , includ- ing the land hydrologic phe- nomena, are inseparable. Changes in one will affect the other. • Hydrology is global. The river basin is no longer the only unit of interest. Acknowl- edging the close relationship between land hydrologic pro- cesses and the atmosphere is acknowledging that the im- portant hydrologic cycle is the global cycle. • The behavior and impact of plants on local, regional, and global hydrology remain our largest unknown. Quantify- ing the relationship between the biosphere and the physi- FIG. 6. Tropical Rainfall Measuring Mission (TRMM) photo (http://trmm.gsfc.nasa.gov) cal-chemical hydrology is for Hurricane Mitch. Clouds retrieved from the visible and spectrum, along with the precipitation structure derived from TRMM. The height of the rainfall structure is shown our biggest challenge. as height in the image. The color is used for the surface rainfall intensity (blue and green for must be- lower intensity, yellow and red for higher intensity). come the monitoring tool of

1154 Vol. 80,, No. 6, June 1999

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC FIG. 7. Very small changes in initial temperature and other variables in one or more locations lead to divergence in the spatially averaged precipitation produced by a very high resolution numerical model of the atmosphere.

choice. It is our only hope for viewing hydrology with the necessary time and space coverage to meet the challenges posed above. • Combined with remote sensing, advances in soft- ware and hardware demand a rethinking of our simulation, modeling, and data-gathering activities.

4. Examples of the revolution

I was taught that occurred when the intensity of rainfall exceeded the capacity of the soil to absorb it. We called it Hortonian runoff (Horton 1933). Now we know that other mechanisms exist; in fact, what I was taught practically never occurs in parts of the world. Dunne and Black (1970) described a satu- ration from below mechanism for runoff production. Figure 1, from Freeze (1980), highlights the main dif- ferences in the runoff production processes when viewed from the perspective of behavior at a point. The fact, though, is that both mechanisms are valid and what is missing is proper accounting of the spatial variability of rainfall, soil properties, land cover, and

FIG. 8. A simple coupled model of the atmosphere and the land surface can produce a probability density function of soil moisture that is bimodal; two different states are preferred, (a) Unimodal; results from a weak and stable link between precipitation and evaporation. As the nonlinear feedback between precipitation and evaporation is increased, driven by a multiplicative noise factor, bimodality arises (b) and (c) (Rodriguez-Iturbe et al. 1991).

Bulletin of the American Meteorological Society 1155

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC spatial variability of soil mois- ture not only affects the mass balance in the basin but influ- ences the energy balance through albedo effects and the partition- ing of sensible and latent heat. The implications to subgrid pa- rameterization of models should be clear. Beven and Kirkby (1979) must be credited for raising con- sciousness about the importance of topography in moisture redis- tribution and runoff production. Their TOPMODEL concept pa- rameterized runoff in terms of basin areas more apt to saturate because of convergence of flow lines and small slopes. That idea begins the departure from the "homogeneous," "equiva- FIG. 9. This figure demonstrates both the impact of soil moisture on the numerical lent," or "lumped" models that I of the extreme events and the notion that surface soil moisture is equivalent in learned as a student and that are importance to sea surface temperature (SST) as boundary conditions for the climatic system, still prevalent in operations, (a) The observed precipitation over the continental United States for summers containing the extreme 1988 and 1993 events are presented as a difference field, (b), (c), where a system of conceptual (d) General circulation model results for the same differences between rainfall during the "buckets" are used to represent 1988 drought and 1993 flood events are shown, (b) Both SSTs and soil-moisture fields are in an aggregate fashion the vari- either observed or estimated for the specific period of simulation, (c) Estimates of soil ous storages and delays that lead moisture for the particular time period are used in conjunction with SST , (d) to the basin response. Such Observed SSTs are used as boundary condition for the simulation in conjunction with approaches were the result of dif- estimated soil moisture climatology. It is evident that precipitation extremes over the United States are more strongly affected by soil moisture fields than SSTs that are currently the ficulties obtaining spatially vari- principal fields used to extend forecast lead times (M. Suarez et al. 1999, personal able precipitation, topography, communication). and soil properties. Presently, we are quickly moving from a data- poor to a data-rich environment. topography. The Horton versus Dunne debate in run- We must substitute conceptualization with data. off production is an artifice of our zero- or one- Digital elevation maps are readily available every- dimensional traditional thinking. This spatial where, and are constantly improving in quality. component to basin infiltration was recognized even Precipitation over the United States is now measured in the early digital watershed models. The early with digital Doppler at very high resolutions in Stanford Watershed Model and its child, the operational time and space. remote sensing promises even Sacramento Model, attempt to account for variability more data on everything from soil moisture to precipi- in runoff production in the basin as a function of a sur- tation to vegetation coverage. The bottom line is that rogate of overall "wetness" over the region. The distributed hydrologic models are the future and are zero-dimensionality of the models did not allow them here to stay. Already we have systems that use digital to account for one key element in the runoff produc- elevation information and radar precipitation to drive tion: basin topography. Topography plays a major role distributed rainfall-runoff models that represent the in the redistribution of moisture, with hollows in the basin at resolutions of tens of meters. These models basin becoming wetter, and hence influences the na- provide a wealth of information on basin behavior, ture of runoff production. Figure 2, from Western and from soil moisture distribution to runoff distribution Grayson (1998), illustrates this redistribution. This in space to discharges at arbitrary locations, as Fig. 3

1156 Vol. 80,, No. 6, June 1999

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC illustrates (Garrote and Bras 1995a,b). There are still challenges of calibration, but these I believe will be resolved as we in- sist on using remotely sensed data to drive the models of the future. When I was a student we were happy if we found one every 2000 to 10 000 km2 over the United States. It looked strange then; it looks stupid now. Today we can blanket the land with me- teorological radar and measure rainfall at resolutions of 2 km2 and sometimes that is not enough! There is little question in my mind that errors in precipitation overwhelm other parameterization and model errors in flow and flood predic- tions [see Fig. 4 from Oki et al. (1999)]. Equally important to all hydrologic ap- plications is our increasing ability to uti- lize space-based remote sensing to obtain estimates of variables like soil moisture (Fig. 5), depth and extent, FIG. 10. Experiments performed using a mesoscale atmospheric model (MM5) surface fluxes, and even precipitation. driven by the land cover of the Rondonia region of the southeastern Amazon Many of us share the excitement pro- region, (top left) The land cover is shown, illustrating the linear pattern of duced by the successful Tropical Rain- deforestation. This pattern is, under certain conditions, reflected in sensible heat, fall Measuring Mission, which for over vertical velocity and cloud water shown, respectively, in the remaining three panels (J. F. Wang and R. L. Bras 1998, personal communication). a year has produced impressive estimates of precipitation from space. This plat- form flies an active down-looking radar and passive hydrologic variables, was a unimodal function with microwave radiometers. The future of remote sensing clear central tendency. Now I know that extremely of precipitation lies in a systemwide integration of simple representations of the land-atmosphere inter- measurements from various space-borne instruments action leads to bimodal PDFs of soil moisture, imply- and of ground observations of various types (Fig. 6). ing that the system has two preferred states and can Predicting precipitation and was our Holy oscillate between them: the biblical seven years of Grail. In some ways it still is, but today it is tempered floods and seven years of drought (Fig. 8)! by something called "." In a nutshell, Back in the early 1970s we saw the land as the hydrologic phenomena are very nonlinear. That means passive recipient of the atmospheric forcing, an inter- that a small change somewhere can lead to a very large action like the one described previously was never change elsewhere in ways that are not fully foresee- given serious consideration or, at best, was relegated able beyond a certain horizon in the future. So pre- to others to think about. Today the play dictability is very much limited. Figure 7 illustrates an important role in the climate change debate, not the enormous sensitivity on initial conditions to a only as potential sinks or sources of carbon but as model predicting precipitation (Li et al. 1995; Islam important factors in the and thermodynamics etal. 1993). of the circulating atmosphere. It is no longer just oceans Nonlinearities and feedbacks between states can and atmosphere that matter, so does the land surface. occur in apparently simple hydrologic systems result- This interaction clearly enhances or inhibits hydro- ing in unusual, unexpected, and unpredictable behav- climatological anomalies like floods or . ior. Back in my student years, if asked to guess at a Figure 9 illustrates that land surface conditions are as representative soil moisture in a region, I would argue important as sea surface temperatures in defining for the climatic mean. I believed that the probability weather and climate over the United States. At the lo- density function (PDF) of moisture, and most natural cal level, deforestation of relatively small areas of rain

Bulletin of the American Meteorological Society 1157

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC predictable impacts on global circulation and associated mass and energy balances. Robert Houze (1969) classified storm events in cells, small mesoscale, large mesoscale, and synoptic scale. Indeed, this useful taxonomy was the basis of many mathematical formulations for simulating rainfall (Rodriguez-lturbe et al. 1986,1987). As useful as that con- cept of dominant scales has proven to be, the present debate is whether or not rain- fall and other atmospheric phenomena exhibit scaling, that is, scale inde- pendence, at least over some regimes (Veneziano et al. 1996). The implica- tions of those results to representations of atmospheric phenomena, to measur- ing of precipitation and other fields, and to predictabilty are significant. This scal- ing is suspected in both time and space. Our challenge remains to link such behav- ior to the dynamics of the atmosphere. Horton was no stranger to the con- cept of scaling and self-similarity. His description of the regularity of treelike geometries has been proven to be consis- tent with a fractal, or self-similar, de- scription of the river basin. In fact Horton's classification of channel net- works by stream orders, with the small- est tributary basins having order one and FIG. 11. The left side of the figure shows an arbitrarily generated "river" network. The top left two panels are a plan view at two different resolutions of with order increasing with aggregation the initial river network. Below, in the same column, are the characteristics of the of tributaries, dominated the field of flu- "basin" in terms of bifurcation ratio (a real basin normally has a value of around vial until the mid-1980s 4); length ratio, which should plot as a straight line against order with a slope of and it is still used by some. The "Horton 2; the probability of exceedance of areas draining through a point, which should numbers" (the bifurcation ratio describ- plot as a straight line with a slope around -0.5; and the probability of exceedance of stream lengths, which should plot as a straight line with slope of about -1.9. ing the branching pattern and the length On the right side of the figure are the new networks generated after imposing ratio stating that the average length of optimal energy principles on the initial condition and its characteristics, showing is related to order) are still com- convergence to behavior observed in . The plot in the middle column shows mon in the literature, although we now how the total energy of the system goes down during the optimization procedure know that they are poor discriminators (Rodriguez-lturbe et al. 1992). of branching-tree structures. The beautiful patterns of drainage forest in the Amazon have the potential of producing that river basins form, I was taught, were the product mesoscale circulations with resulting impact on cloudi- of randomness. I learned a beautiful mathematical ness, radiation, and even precipitation (Fig. 10) (J. F. theory of topologically random trees (Shreve 1969). Wang and R. L. Bras 1998, personal communication). Today we know that the treelike organization of river This land-atmosphere interaction is not only local. basins and drainage results from well-enunciated prin- We are now familiar with teleconnections triggered by ciples of energy expenditure and nature's desire to do sea surface anomalies like El Nino. It is to be expected its job efficiently. Through these ideas we relate the that large alterations of the land surface will have un- fractal nature of the basin to underlying physical prin-

1158 Vol. 80,, No. 6, June 1999

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC ciples (Fig. 11). This is something to keep in mind every time we alter the without heeding those principles. Ultimately nature will prevail: ask those who suffered the 1993 Mississippi floods. We have learned that the topological, planar de- scription of the basin is not independent of the third dimension: the relief or energy in the system (Fig. 12). We have also learned that the evolution of the river basin, its statistical characterization, is intimately re- lated to the hydrologic processes that form it. We have built numerical models that mimic the evolution of the basin. In a sense, we can play creator and let our arti- ficial nature evolve an organized structure from dis- organized origins (Fig. 13). The outcome of our experiments is very much a function of the processes in our artificial nature. Figure 14 shows how radically different basins result from experiments with or with- out mechanisms for hill-slope erosion by landsliding. Possibly one of the most important concepts we have learned in recent fluvial geomorphology work is that it is futile to think of the river channels inde- FIG. 12. The networks shown are generated from a featureless initial condition by simply imposing a relationship between slope pendently of the hill slopes that feed them and vice 9 versa. Channel and hill slope are part of a whole and at a point and contributing area at that point of the form, S ~ A" . Clearly the planar expression of organization is very much a control and shape one another (Fig. 15). function of the relief of the system. In nature 0is generally around When I learned hydrology, vegetation was treated 0.5 (from Tarboton et al. 1998). almost as a nuisance term. For example, its effects on land surface fluxes were lumped into a catchall evapo- transpiration term, which was in turn crudely param- noring the strong links is a recipe for disaster, evident eterized. In the late 1970s and early 1980s Eagleson as we seek to preserve the quality and quantity of our suggested the then radical idea that the biosphere, the . climate, the hydrology, and the soil were in a syner- We believed that disposing of contaminants in the gistic waltz trying to reach a point of "ecological was safe, since flow in a homogeneous soil was optimality" (Eagleson 1978a-g; 1982). Hydrologists so slow that contaminants could not possibly move too have certainly awakened to the realization that vegeta- far. How wrong we were! Today we suffer the conse- tion affects all we do and cannot be summarily treated. quences of our ignorance with thousands of contami- Some of the work in land-atmosphere interactions dis- nated sites endangering humans' health and the cussed previously certainly considers vegetation a key environment because water does move through pre- element of the analysis. In my opinion, though, we re- ferred high-conductivity areas or cracks and fractures main woefully ignorant and need to focus more on in soil and rock. Even Hollywood and John Travolta these issues. have discovered that excitement and mystery of the The study of subsurface flows and transport have field! been a dominant subject in hydrology over the last 30 Studies of groundwater hydrology led our forays years. I will do it injustice by limiting my comments. into the impact of spatially variable medium properties For one, my "short history" is becoming too long and on hydrologic behavior, particularly flow and trans- furthermore my own work in the field is far more lim- port. The work of Dagan, de Marsily, Gelhar, and ited. But do not be misled; advances in the analysis of Neuman developed the tools presently used to study subsurface flows have been extraordinary both from heterogeneity in subsurface and other hydrologic sys- the science and the applications point of view. Years tems. The field also led in the solutions of inverse ago I was led to believe that surface , like problems and parameter estimation, much of it pres- and , and were practically indepen- ently reappearing in the context of modern assimila- dent systems, with weak links. Now we know that ig- tion approaches in and surface hydrology.

Bulletin of the American Meteorological Society 1159

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC it useable in the computer. Today my students go into the World Wide Web and practically instantly obtain digitized elevation data of anywhere in the world: no time, no sweat, and practically no cost.

5. The next 30 years

Water and the water environment are quickly becoming global and political issues, with implications to the security of nations and human health, develop- ment and worldwide. As the science of water, hydrology has a bright and exciting future. But I believe we are at a threshold, one similar to the one I encountered in 1968. A major push is needed and more bright young minds must come into the field to create new paradigms and new knowledge. A group of us (see references), led by my younger colleague, Professor Dara Entekhabi, have produced a document entitled "An Agenda for Land-Surface Hydrol- ogy Research and a Call for the Second International Hydrologic Decade" (Entekhabi 1999), presently available FIG. 13. The evolution of a river basin as simulated with a numerical model. httpV/web mit edu/darae/WWW/

The initial condition is a flat, uplifted, region with one outlet (from Moglen and H dro html and wiU in an J Bras 1994). . . r , „ r , coming issue of the Bulletin of the American Meteorological Society. I would be remiss if I ended this narrative without The time is ripe for a new hydrologic decade. The mentioning the impact that computers have had in the last one, 1965-75, was a major element in launching field. For my bachelor and masters theses, I found the exciting developments that I have summarized myself pushing computer technology of the time to previously. It created a revolution of thought. It was the limit. I developed a numerical model of water flow the seed on which many of us have tried to build a new in an urbanized area. The best computer technology science from a mature engineering discipline. The of the time took the whole night (and I had to be time is ripe because we have identified many avenues present!) to run the model. The computer took close of inquiries; because the outstanding questions are im- to 1000 square feet of space and the instructions were portant to society; and because technology, particu- written on cardboard cards (or even paper tape). To- larly in computation, communications, and remote day cards are museum pieces; I believe my children sensing, is ready and capable to help us through the have never seen one. The program I developed is now outstanding obstacles. It is ripe because, in the era of surpassed by one hundredfold and available for any- interdisciplinarity, we are ready and eager to find new body, and it runs in seconds in any desktop personal partners, to create new alliances, and to continue the computer. evolution of new models for the education of the hy- Only a few years ago, if I wanted to get the topog- drologist of the future. Finally, what could be more raphy of a region, I had to obtain a map on paper and, exciting than to launch such global effort with a prom- with some luck and a lot of effort, digitize it to make ising new millennium?

1160 Vol. 80,, No. 6, June 1999

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC The "agenda" presented is limited to land surface hydrology; we are confident that complementary documents will come forth to complete our vision of the future. I would like to take the op- portunity to summarize the science pri- ority questions that the document poses. I urge you to read the complete docu- ment for detail. The science questions are the following.

• What are the physical mechanisms and process pathways by which the coupling between surface hydrologic systems and the overlying atmo- sphere modulate regional weather and climate variability? • What are the mechanisms and the timescales of interactions between the formation of terrain, soils, vegeta- tion ecotones, and hydrologic re- sponse? • Are there critical scales at which spa- tial variations in surface properties should be explicitly represented in models of land-atmosphere exchange? • Under what conditions can effective FIG. 14. (top) A land surface generated only by surface erosion and diffusional parameters be used to represent processes like and rain splash at small scales, (bottom left) The result macroscale hydrologic processes, and when a pore-activated landsliding mechanism is operational, (bottom right) The does the upscaling of microscale pro- result of a dominant saturation from below runoff production mechanism (from cesses depend on the process and lead Tucker and Bras 1998). to changes in the form of the govern- ing equations? models that are forced with spectral observations • Does lateral redistribution significantly or retrieved fields of spatial data. affect large-scale soil-vegetation-atmosphere ex- Hydrologists need to evolve from passive recipi- change processes? ents of limited remote-sensing observations to act- • How can the effects of human activity on hydrologic ing as a unified scientific community that is en- response be distinguished from natural climate vari- gaged in supporting the definition, design, and ability in a range of physiographic environments? implementation of suborbital and space-borne instruments. The agenda goes on to develop the need for re- Validation databases (over regions or basins) con- search in monitoring systems and their integration with sisting of in situ and remote-sensing measurements the science questions and modeling needs. I want to need to be established so that instruments and re- emphasize the requirements for fulfilling the promise trieval algorithms may be quantitatively and defi- of remote sensing that the document identifies. nitely evaluated.

• Classic ideas and associated existing models that are optimized to function with sparse in situ ob- 6. To conclude servations of precipitation accumulation, stream , and surface-air micrometeorology I hope I have been able to project the excitement, should be replaced with reformulated hydrologic the fast pace, and the importance of hydrology. To me

Bulletin of the American Meteorological Society 1161

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC important professional legacy. I thank them all. (Their names are provided in the acknowledgments.) Finally, I want to thank my parents, Amalia and Rafael: what can I say, they gave me all. Last but not least, I am blessed with a wonderful partner and friend, Pat, and two absolutely wonderful children, Rafael and Alejandro. Frankly, after them the rest is icing on the cake.

Acknowledgments. Throughout my career I have been blessed with the sponsorship of many groups and institutions. They include agencies such as NSF, NASA, NOAA/NWS, DOD/ARO/ ONR, DOE, DO, Department of State, and EPA. The government of Italy, particularly the National Research Council, has generously funded my re- search. I must also mention the Bacardi Corpora- tion, as well as the Stockholm Water , whose generosity endows my professorship at MIT. This paper was prepared with the on-going support of the following grants and programs: U.S. Army Research Office, "Fractal Properties of the 3-Dimensional Terrain" (DAAH04-95-1- 0181); U.S. Army Construction Engineering Re- FIG. 15. Landscape generated by a model that includes a meandering channel, search Laboratory (CERL), "An Integrated (top) The landscape and a section of river, (bottom left) The triangular grid used Hillslope and Channel Evolution Model as an in the numerical solution. The region of high density of grid points corresponds Investigation and Tool" (DACA88- to areas where the stream has meandered and altered the terrain (the solution mesh 95-C-0017); Consiglio Nazionale delle Ricerche, is dynamic), (bottom right) Color codes drainage areas. Bluish regions drain the Cooperative Agreement in Climate Change and most area (from Lancaster 1998). Hydrogeological Disasters; U.S. Army Research Office (ASSERT), "Development of a High Resolution Simulator of SpaceTime Rainfall for the last 30 years have been extraordinarily rewarding Mobility Assesment" (DAAH04-96-1-0099); NASA/SSA professionally and personally. Let me quote Albert Goddard Space Flight Center, "Amazon Deforestation and Re- Einstein: gional Climate" (NAG5-3726); National Science Foundation, "Scaling Approach to Hydrological Extremes and Their Depen- dence on Climate and Basin" (CMS-9612531); Army Research A hundred of times every day I remind myself Office, Equipment for DOD-Sponsored Topographic Research; that my inner and outer life are based on the Alliance for Global Sustainability, "Natural Hazards and Climate labors of other men [and women], living and Change: The Case of Extreme Droughts and Floods;" NASA/ dead, and that I mustj exert myself in order to Goddard Space Flight Center, "The Value of Distributed Mod- give in the same measure as I have received and eling to " (NAG5-7475). am still receiving. Finally, thank you to all my students: Adeyinka E. Adenekan, Babar Bhatti, Alice Brown, Ross Buchanan, M. Cabral, Fabio I have benefited tremendously from my associa- Castelli, Siu-On Chan, Siong H. Chua, Raul Colon, Jose R. Cordova, K. Curry, Mario Diaz-Granados, Elfatih Eltahir, D. W. tion with MIT and the Department of Civil and Envi- Ephraim, Ying Fan-Reinfelder, Mark French, Auroop Ganguly, ronmental Engineering. I have been mentored and Luis Garrote, Nicole Gasparini, Konstantine Georgakakos, Maria tutored by giants in the field who are also some of my Gini, Ede Ijjasz-Vasquez, Shafiqul Islam, Edward R. Johnson, very personal friends. Doctors Peter Eagleson, Ignacio Diana Kirshen, Peter K. Kitanidis, Steven Lancaster, David E. Rodriguez-Iturbe, and Donald R.F. Harleman: thank Langseth, Qihang Li, Glenn Moglen, M. Moughamian, Jeffrey Niemann, Angelos L. Protopapas, Carlos Puente, Jorge Ramirez, you for everything. Pedro Restrepo, Scott Rybarczyk, Dong-Jun Seo, David G. Standing on the shoulders of giants was not Tarboton, Gregory E. Tucker, Ivanov Valeri, Laurens van der Tak, enough; I continue to rely on the sweat of an extraor- M. Wagle, Jingfeng Wang, Garry Willgoose, J.D. Wyss, Li Yu, dinary group of students. They are, after all, my most and Yanlong Zhang.

1162 Vol. 80,, No. 6, June 1999

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC References , and ,1995b: An integrated software environment for real-time use of a distributed hydrologic model. J. Hydrol., 167, Beven, K. J., and M. J. Kirkby, 1979: A physically-based, vari- 307-326. able contributing area model of basin hydrology. Hydrol. Sci. Gray, D. M., Ed., 1973: Handbook on the Principles of Hydrol- Bull., 24 (1), 43-69. ogy. Water Information Center, Inc. Bras, R. L., 1998: The Clarke Prize Lecture. National Water Re- Harley, B. M., F. E. Perkins, and P. S. Eagleson, 1970: A modu- search Institute, 10 pp. [Available from National Water Re- lar distributed model of catchment dynamics. Ralph M. Par- search Institute, 10500 Ellis Avenue, P.O. Box 20865, Foun- sons Laboratory for Water Resources and Hydrodynamics, tain Valley, CA 2728-0865.] MIT, Rep. 133, 537 pp. [Available from Ralph M. Parsons , 1990: Hydrology: An Introduction to Hydrologic Science. Laboratory, Department of Civil and Environmental Engi- Addison Wesley, 643 pp. neering, MIT, Cambridge, MA 02139.] Chow, V. T., 1964: Handbook of Hydrology. McGraw-Hill. Horton, R. E., 1932: characteristics. Trans. Amer. Committee on Opportunities in the Hydrologic Sciences, 1991: Geophys. Union, 13, 350-361. Opportunities in the Hydrologic Sciences. National Academy , 1933: The role of infiltration in the hydrologic cycle. Trans. Press, 348 pp. Amer. Geophys. Union, 14, 446-460. Crawford, N. H., and R. K. Linsley, 1966: Digital simulation in , 1940: An approach toward a physical interpretation of in- hydrology: Stanford Watershed Model IV. Stanford Univer- filtration capacity. Soil Sci. Amer. Proc., 5, 399^17. sity, Department of Civil Engineering, Tech. Rep. 39,210 pp. , 1945: Erosional development of streams and their drainage [Available from Department of Civil and Environmental En- basins: Hydrophysical approach to quantitative morphology. gineering Library, 2nd Floor, Terman Engineering Center, Bull. Geol. Soc. Amer., 56,275-370. Stanford University, Stanford, CA 94305-4020.] Houze, R. A., 1969: Characteristics of mesoscale precipitation Dirmeyer, P. A., 1997: The Global Soil Wetness Project areas. M. S. thesis, Department of Meteorology, Massachu- (GSWP)—Initial results. GEWEX News, 7 (2), 3-6. setts Institute of Technology, 77 pp. [Available from Dept. of Dunne, T., and R. D. Black, 1970: Partial area contributions to Meteorology, MIT, Cambridge, MA 02139.] storm runoff in a small New England watershed. Water Resour. Islam, S., R. L. Bras, and K. Emanuel, 1993: Predictability of Res., 6(5), 1296-1311. mesoscale rainfall in the Tropics. J. Appl. Meteor., 32, 297- Eagleson, P. S., 1970: Dynamic Hydrology. McGraw Hill, 462 pp. 310. , 1978a: Climate, soil and vegetation—Part 1. Introduction Lancaster, S. T., 1998: A nonlinear river meandering model and to dynamics. Water Resour. Res., 14 (5), 705- its incorporation in a landscape evolution model. Ph.D. the- 712. sis, Massachusetts Institute of Technology. [Available from , 1978b: Climate, soil and vegetation—Part 2. The distribu- Department of Civil and Environmental Engineering, MIT, tion of annual precipitation derived from observed storm se- Cambridge, MA 02139.] quences. Water Resour. Res., 14 (5), 713-721. Li, Q., R. L. Bras, and S. Islam, 1995: Growth and decay of errors , 1978c: Climate, soil and vegetation—Part 3. A simplified in a numerical cloud model due to small initial perturbations model of soil moisture movement in the liquid phase. Water and parameter changes. J. Appl. Meteor., 34, 1622-1632. Resour. Res., 14 (5), 722-730. Linsley, R. K., Jr., M .A. Kohler, and J. L. H. Paulhus, 1958: Hy- , 1978d: Climate, soil and vegetation—Part 4. The expected drology for . 1st ed. McGraw-Hill, 340 pp. value of annual evapotranspiration. Water Resour. Res., 14 (5), , , and , 1982: Hydrology for Engineers. 3d ed. 731-739. McGraw-Hill, 508 pp. , 1978e: Climate, soil and vegetation—Part 5. A derived dis- Moglen, G., and R. L. Bras, 1994: Simulation of observed topog- tribution of storm surface runoff. Water Resour. Res., 14 (5), raphy using a physically-based basin evolution model. Ralph 741-748. M. Parsons Laboratory, Tech. Rep. 340, 221 pp. [Available , 1978f: Climate, soil and vegetation—Part 6. Dynamics of from Ralph M. Parsons Laboratory, Department of Civil and the annual water balance. Water Resour. Res., 14 (5), 749- Environmental Engineering, MIT, Cambridge, MA 02139.] 764. Oki, T., T. Nishimura, and P. Dirmeyer, 1999: Assessment of land , 1978g: Climate, soil and vegetation—Part 7. A derived dis- surface models by runoff in major river basins of the globe tribution for annual water yield. Water Resour. Res., 14 (5), using total runoff integrating pathways (TRIP). J. Meteor. Soc. 765-776. Japan, in press. , 1982: Ecological optimality in water limited natural soil Rodriguez-lturbe, I., D. R. Cox, and P. S. Eagleson, 1986: Spa- vegetation systems. 1. Theory and hypothesis. Water Resour. tial modelling of total storm rainfall. Proc. Roy. Soc. London, Res., 18 (2), 325-340. 403A, 27-50. Entekhabi, D., 1999: An agenda for land-surface hydrology re- , , and V. Islam 1987: Some models for rainfall based search and a call for the second international hydrologic de- on stochastic point processes. Proc. Roy. Soc. London, 410A, cade. Bull. Amer. Meteor. Soc., in press. 269-288. Freeze, A., 1980: A stochastic conceptual analysis of rainfall-run- , D. Entekhabi, and R. L. Bras, 1991: Nonlinear dynamics off process on a hillslope. Water Resour. Res., 16 (2), 391— of soil moisture at climate scales: 1. Stochastic analysis. Wa- 408. ter Resour. Res., 27 (8), 1899-1906. Garrote, L., and R. L. Bras, 1995a: A distributed model for real- , A. Rinaldo, R. Rigon, R. L. Bras, E. Ijjasz-Vasquez, and A. time flood forecasting using digital elevation models. J. Marini, 1992: Fractal structures as least energy patterns: The Hydrol167, 279-306. case of river networks. Geophys. Res. Lett., 19 (9), 889-892.

Bulletin of the American Meteorological Society 1163

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC Sherman, L. K., 1932: Stream flow from rainfall by the unit graph Veneziano, D., R. L. Bras, and J. D. Niemann, 1996: Nonlinearity method. Eng. News Rec., 108, 501-505. and self similarity of rainfall in time and stochastic model. J. Shreve, R. L., 1969: Stream lengths and basin areas in topologi- Geophys. Res., 101 (D21), 26 371-26 392. cal^ random channel networks. J. Geol., 77, 397-414. Western, A. W., and R. B. Grayson, 1998: The Tarrawarra data Tarboton, D. G., R. L. Bras, and I. Rodriguez-Iturbe, 1988: The set: Soil moisture patterns, soil characteristics, and hydrologi- fractal nature of river networks. Water Resour. Res., 24 (8), cal flux measurements. Water Resour. Res., 34 (10), 2765- 1317-1322. 2768. Tucker, G., and R. L. Bras, 1998: Hillslope processes, drainage density, and landscape morphology. Water Resour. Res., 34 (10), 2751-2764.

With the development of meteorological science and the continual refinement of the technologies used in its practical application, the need to produce a new edition of the International Meteorological Vocabulary (IMV) became evident (the original edition was published in 1966). This volume is made up of a multilingual list of over 3500 terms arranged in English alphabetical order, accompanied by definitions in each of the languages (English, French, Russian, and Spanish) and an index for each language. This new edition has been augmented with numerous concepts relating to new meteorological knowledge, techniques, and concerns. It should help to standardize the terminology used in this field, facilitate communication between specialists speaking different languages, and aid translators in their work. WMO No. 182, 784 pp., softbound, color-coded index, $95 (including postage and handling). Please send prepaid orders to: WMO Publications Center, American Meteorological Society, 45 Beacon St., Boston, MA 02108-3693. (Orders from U.S. and only.)

International Meteorological Vocabulary

1164 Vol. 80,, No. 6, June 1999

Unauthenticated | Downloaded 10/08/21 08:48 AM UTC