Testing an Urban Climate Simulator By GORDON M. GREENE THE INFLUENCES OF LAND USE AND LAND COVER IN CLIMATE ANALYSIS GEOLOGICAL _ »SURVE Y PROFESSIONAL _ PAPER - 1099- E An example of environmental analysis using land use and land cover information UNITED STATES GOVERNMENT PRINTING OFFICE, WASHINGTON : 1980 UNITED STATES DEPARTMENT OF THE INTERIOR CECIL D. ANDRUS, Secretary GEOLOGICAL SURVEY H. William Menard, Director Library of Congress Cataloging in Publication Data Greene, Gordon M. The influences of land use and land cover in climate analysis. (Geological Survey professional paper; 1099-E) Bibliography: p. Supt. of Docs. no.: I 19.16:1099-E 1. Urban climatology-Mathematical models. 2. Land use, Urban-Environmental aspects-Mathematical models. I. Title. II. Series: United States. Geological Survey. Professional paper; 1099-E. Q@C981.7.U7G73 551.6'9'1732 79-607092 For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, D.C. 20402 FIGURE 1. TABLE 2. 8. 4. 5 120° CONTENTS Page Abstract - -_. - JEI Introduction _____ 1 Review of urban climate modeling ________________ 1 Energy balance simulation in urban environments ____ 2 The model ..___.._ a 6 Application of the model osa. 7 Analysis of the model -=« {10 Conclusion _. 15 References cited ___ 16 ILLUSTRATIONS P P Page Line drawing showing silhouette ratio, which is found by dividing the vertical area by the ground area - E3 Line drawings showing the interception of direct shortwave radiation by buildings __________________ ___ 4 Line drawings showing the interception of longwave radiation by buildings and vegetation (adopted from Myrup and Morgan, 1972) ____- < 4 Flow chart of the Qutcalt and Carlson URBD model (1975) 4. Map showing the path of the third flight across Baltimore, Md., along which multispectral scanner data (including thermal) were obtained by the Environmental Research Institute of Michigan (ERIM) aircraft on May 11, 1972 _ ass 8 Graph showing site characteristics for seven land use types 10 Graphs showing simulated surface temperatures in Baltimore, Md., for various land use categories based on hourly and averaged daily values on May 11, 1972 wor 192 Graph showing effect on increased air temperature on the predicted surface temperature _______________ 16 TABLES Page Recorded meteorological data __--- ___ E8 Surface weather observations at Baltimore-Washington International Airport, May 11, 1972 .________.__- 8 Land use characteristics ______----- 9 Simulation of solar radiation flux for medium density residential land use, May 11, 1972, Baltimore, Md ___ 11 Simulation of micro-meteorological conditions for medium density residential land use, May 11, 1972, Baltimore, Md RP Eel 11 Significance test for differences between URBD model using observed hourly data and mean daily data 88 INPUE = awe Siz 13 Control run values of the URBD model for use in sensitivity testing sek =-, A8 Sensitivity of the URBD model to changes in substrate diffusivity (x) nes a= -.. 18 Sensitivity of the URBD model to changes in the height of obstructions (Re) _________________-_-------- 14 Sensitivity of the URBD model to changes in the surface relative humidity fraction (SRHF) _________- 14 Sensitivity of the URBD model to changes in silhouette ratio (SILRAT) _______________________________ 14 Sensitivity of the URBD model to changes in albedo (ALB) 14 Rank (R) of absolute value of T;,; -T, at 1800 for each land use type -. 14 Sensitivity of the URBD model to decreased wind SPe@d (w) ________________oooocccccccccccccccecc~---- 15 Sensitivity of the URBD model to increased wind speed (a) ...__._.__L_.._..... 15 Comparison of observed and simulated surface temperature, 1400 EST, May 11, 1972 (in degrees Celsius) %« 15 Simulation of global incoming solar radiation a 15 IH iv CONTENTS LIST OF ABBREVIATIONS AND SYMBOLS ALB average shortwave albedo for all surfaces. BAL,, the sum of R,,, S, H, and LE based on the second surface temperature estimate (t-1). BAL,, - the sum of R,, S, H, and LE based on the first surface temperature estimate (t-2). BE AM direct shortwave radiation flux, in watts per square meter. CBD Central Business District land use category. CITY subroutine in the URBD model which computes the soil damping depth (z,,). COMM commercial area land use category. Cp specific heat of air at constant pressure. CPATH solar geometry input to the URBD model. E, saturation vapor pressure. EPHE M subroutine in the URBD model which computes the solar declination. EXT shortwave radiation flux on a horizontal surface outside the Earth's atmosphere. F artificial heat flux due to combustion. H kinetic or sensible heat flux into the atmosphere. HDR high density residential land use category. HEM diffuse shortwave radiation flux. J Joule. L latent heat of vaporization. LDR low density residential land use category. LE latent energy used in evapotranspiration. MDR medium density residential land use category. PARKS park land area land use category. Qair specific humidity of the air. surf specific humidity at the ground surface. READIN geographical and meteorological input data to the URBD model. RH relative humidity. Ri bulk Richardson number. Ru, | longwave radiation emitted by the Earth's surface. R,J - longwave radiation emitted by the Earth's atmos- phere. R, radiation balance at the terrestrial surface (net radiation). R.,f - solar shortwave radiation reflected by the Earth's surface. R.,J - solar shortwave radiation incident upon the Earth's surface. RUNTRI subroutine in the URBD model used to deter- mine the condition of the soil temperature profile. S net heat flux to buildings, roads, and substrate. SECANT algorithm used in the URBD model to compute a new surface temperature estimate, in order to stabilize the soil temperature profile. SHDRAT shadow ratio. SIG Stefan-Boltzman constant. SILRAT SKY silhouette ratio. area of a ground observer's sky hemisphere, in percent. SOLURB subroutine in the URBD model which computes the shortwave radiation input. SRHF surface relative humidity fraction. SUN backscattered radiation flux on a horizontal surface at the Earth's surface. Tar air temperature, in degrees Celsius. To control run temperature of the URBD model for each land use type. TERRAIN total shortwave radiation flux averaged over the urban surface. mean diurnal air temperature, in degrees Celsius. maximum temperature, in degrees Celsius. minimum temperature, in degrees Celsius. model computed surface temperature, based on manually derived high- and low-surface tempera- ture estimates. TRANS transportation land use category. TSKY radiation temperature of the sky hemisphere. a girs surface temperature, in degrees Celsius. T;; second surface temperature estimate, used in the URBD model to stabilize the soil temperature profile. first surface temperature estimate, used in the | URBD model to stabilize the soil temperature profile. T2 temperature at the first node below the ground sur- face, in degrees Celsius. le UBAL subroutine in the URBD model which computes R,, S, H, and LE. UPRIG subroutine in the URBD model which computes the amount of radiation on a vertical surface. VERT total shortwave radiation on a vertical surface, in watts per square meter. bairp slope, or b value, computed in the Student's t-test, of the albedo parameter. slope or b value, computed in the Student's t- test, of the silhouette ratio parameter. Osrp - slope, or b value, computed in the Student's t-test, of the surface relative humidity fraction parameter. b., - slope, or b value, computed in the Student's t-test, of the surface roughness parameter. b, slope of a specified regression line. A, obstacle height, in centimeters. k von Karman's constant (0.4). p atmospheric pressure, in millibars. r coefficient of correlation. t the period of a specified temperature fluctuation. W watt. Pa height above, or depth below, surface. #4 atmospheric damping height, in centimeters. £q soil damping depth, in centimeters. *, surface roughness length, in centimeters. Az the change in height above, or depth below, surface, in centimeters. k thermal diffusivity, in square centimeters per second. A thermal conductivity, in Joules per centimeter per second per degree Celsius. wind speed at height z, in centimeters per second. air density, in grams per cubic centimeter. "© E THE INFLUENCES OF LAND USE AND LAND COVER IN CLIMATE ANALYSIS TESTING AN URBAN CLIMATE SIMULATOR By Gorpon M. GrEENE * ABSTRACT Geographic data, generally in the form of terrain informa- tion, are a primary input to urban climate simulation models. Five urban terrain parameters: silhouette ratio, substrate dif- fusivity, obstruction height, surface relative humidity frac- tion, and albedo account for most of the geographic input. In a test of the sensitivity of model-predicted surface tem- peratures to simulated changes in the five urban terrain parameters, the influences of these parameters on the urban thermal regime of Baltimore, Md., were analyzed. Results indicated that changes on the order of 30 percent, in individ- ual urban terrain parameters, have relatively little effect on surface temperatures. When more than one of these param- eters are considered together, however, changes within the +80 percent range become more significant. The major bene- fit of the urban climate model in this analysis is not as a predictive tool, but as a method whereby relationships be- tween input components are examined. INTRODUCTION The use of the energy-balance relationship to ex- plain the partitioning of shortwave and longwave radiation at the Earth's surface can be deceptively simple. Stated in an equation: Net radiation=heat conducted into the ground-|-turbulent transfer of heat by (1) convection and evaporation. Traditionally, the ideal setting for such a model of reality is an infinite, smooth plain with unvarying wind speed, air temperature, relative humidity, and solar radiation flux. In contrast, consider this ideal against the reality of a few square blocks near the center of a large city. Streets are lined with multistory buildings, one side warmed by the early morning sun, the other side retaining the nighttime coolness. One block holds a small park with asphalt tennis courts, a large fountain used for wading, and a number of *The research reported herein was funded in part by the National Aeronautics and Space Administration and U.S. Geological Survey as one phase of the Central Atlantic Regional Ecological Test Site (CARETS) Project. *Department of Geography, University of Michigan, Ann Arbor, Mich. large old trees. A small shopping area is separated from the park by a four-lane street, busy with com- mutters. The shops are mostly two stories, some with freshly tarred roofs. A study of the energy balance relationship of such an area requires an accounting of the tremendous variety in structures, their ma- terials, shapes, sizes, and orientations. Myrup (1969) and Outcalt (1972a, b) were among the first to quantify some of these variables, frequently referred to as urban terrain parameters, and incor- porate them into urban simulation models. The purpose of this report is to examine the influences of five urban terrain parameters: sil- houette ratio, substrate diffusivity, obstruction height, surface relative humidity fraction, and albedo, on the urban thermal regime. The analysis is based on a test of the sensitivity of model- predicted surface temperatures to simulated changes in the five parameters. The simulation analysis was made in accordance with conditions existing in Baltimore, Md., on May 11, 1972. Land use and land cover data, from which values for the terrain parameters could be obtained, as well as thermal infrared images, against which the model could be verified, were available for that date in Baltimore. REVIEW OF URBAN CLIMATE MODELING Historically, the first major step in the study of urban climate relations was the recognition that large cities often have higher air temperatures in the city center than at the suburban fringes. This phenomenon was termed the "heat-island" effect and has been well described in the literature (Born- stein, 1968; Terjung, 1970). The temperature differences between urban and rural areas result from numerous factors. Thermal characteristics of steel, concrete, and asphalt pro- mote higher heat absorption and storage. A city center has a large proportion of vertical surfaces intercepting shortwave and longwave radiation. The atmospheric composition may be modified by more dust and aerosols over a city (Tuller, 1973). Rain in cities is usually drained away soon after falling, removing the possibility of net solar radiation being E 1 E 2 used for evaporation (Hage, 1975). Finally, human activities in cities produce heat from a large number of sources. As urban climate investigations became more sophisticated, increasing emphasis was placed on explaining differences in microclimates within the city proper (Terjung and others, 1970). Those factors which distinguish an urban area from the countryside are not spatially homogeneous and may, therefore, be studied as a function of land use (Outcalt, 1972a). Alexander and others (1976), suggest that the concept of an urban heat island may exist only as a function of faulty research methodology. For example, the use of street-level spot observations taken from a car may smooth out microclimate diversity within a city, if the immediate influence of a street on surface tempera- tures is stronger than that of the surrounding en- vironment. In effect, the overall "texture" of an urban surface is so varied that the concept of a representative site for different land uses is mean- ingless. The only realistic approach, therefore, is to use area-averaging parameters (Myrup and Morgan, 1972; Marotz and Coiner, 1973). The increasing use of remotely sensed thermal data provides the synoptic view of urban climates that early investigators were attempting to create. The ability to analyze the microclimate at a specific point and height above the ground is sacrificed, however, since the remotely sensed thermal image is a two-dimensional projection of a three-dimension- al surface. Even so, the temperature field shown for a remotely sensed area is a unique value caused by the complex interaction of solar radiation with the particular geometric and thermal. properties of that area. As demonstrated by Outcalt (19722), remotely sensed data can be used to categorize land use and land cover, and the interaction of land use and land cover with climatic variables. One method for studying this interaction is through computer modeling, which simulates the events assumed to be taking place. Models applied to urban climates are generally classified as at- tempts to study either mesoscale or microscale in- teractions (Schneider and Dickinson, 1974). Meso- scale models usually operate in two dimensions and are used to investigate the nature of the urban boundary layer. Emphasis is placed on those fac- tors which alter the turbulent transfer processes for heat, moisture, and momentum over an urban area, similar to a micrometeorologist studying the changes in turbulence caused by the transition from THE INFLUENCES OF LAND USE AND LAND COVER IN CLIMATE ANALYSIS a grassy field to an asphalt runway (see, for ex- ample, Gutman and Torrance, 1975; Leahey and Friend, 1971; McElroy, 1973; and Bornstein, 1975). The approach used in this study relies on a micro- scale one-dimensional energy balance model pio- neered by Myrup (1969). His first model has since been expanded for application to Ann Arbor, Mich., (Outcalt, 1972a) and Baltimore, Md. (Alexander and others, 1976). Myrup's original work also has been expanded and applied in a detailed study to Sacramento, Calif. (Myrup and Morgan, 1972). The use of the model is helpful in studying urban climates because of the complexity of the city at- mospheric interaction. Another benefit of this simu- lation model is that it allows one to link relation- ships which are only empirically justifiable to those processes which are well understood. The model provides the framework for arranging the informa- tion one has available as a system. ENERGY BALANCE SIMULATION IN URBAN ENVIRONMENTS The underlying principle of the surface energy balance in equation 1 is the conservation of energy. In a steady-state condition, the amount of energy gained must equal the amount of energy lost. Change is thought of as an instantaneous change to a new steady state. For urban applications, equation 1 can be rewritten as: R.+F=LE+H+S, (2) where: R, = radiation balance at the terrestrial surface (net radiation). F = artificial heat flux due to combus- tion LE = latent energy used in evapotranspi- ration H = kinetic or sensible heat flux into the atmosphere S = net heat flux to buildings, roads, and substrate. Following convention, the energy flux for any term is considered positive if directed toward the surface, negative if directed away. Although many authors have used F as a component of the energy relationship (for example, Gutman and Torrance, 1975; Myrup, 1969), it was not included in this model since there are not enough data to give a meaningful artificial heat flux value to the different land uses. It is not clear if this is a significant omis- TESTING AN URBAN CLIMATE SIMULATOR sion. Yu and Wagner (1975) estimated F as 140 W/m - for New York City on a clear winter day with an air temperature of 0° C. A more reasonable estimate for May would be the estimate of heat flux due to vehicular activity of 7 W/m-, although air conditioning should also be considered. Both these heat fluxes are much smaller than the mean midday flux of longwave and shortwave radiation of S70 W/m -. By specifying the boundary conditions, one can compute the energy flux contributed by each of the four energy balance components from the fol- lowing assumptions and procedures. While each component can be expressed as a function of sur- face temperature, it is not possible to analytically solve equation 2 for the unique equilibrium surface temperature which balances the equation. One can, however, use either an analog model or a numerical iteration scheme to find the surface temperature. The appendix to the Baltimore study (Alexander and others, 1976) contains a good summary of the derivations for each of the energy flux components. Rather than repeating those equations, this section presents the components in their finite-difference form for use in a digital computer model and ex- plains the modifications used to adapt the energy balance relation to a city. In addition, this section examines the response of the energy flux com- ponents in the urban terrain parameters. The first component, net radiation (R,), can be broken down into its shortwave and longwave forms: Rn+ (Rcwl = lwf )+(lel “let ) where R,,] is the solar shortwave radia- tion incident upon the Earth's surface, R,, t is the solar shortwave radia- tion reflected by the Earth's surface, Rl is the longwave radiation emitted by the Earth's atmosphere, and R.,,t is the longwave radiation emitted by the Earth's surface. (8) Looking at the shortwave equation : Rak =[ (1 - ALB)Y® (1 - SHDRAT) x (BEAM + HEM)]+(SHDRATx HEM) +(SILRATXxVERT) (4) where: ALB=average shortwave albedo for all surfaces E3 SHDRAT=shadow ratio (discussed below) BEAM=direct shortwave radiation flux HEM=diffuse shortwave radiation flux SILRAT=sgilhouette ratio (discussed be- low) VERT=total shortwave radiation on a ver- tical surface. The definition of the urban terrain parameter SILRAT is illustrated in figure 1. The role SILRAT EXPLANATION Ground area Vertical area vertical area ground area . FIGURE 1.-SILRAT, where SILRAT = plays as a roughness factor is shown in figure 2. The average height of obstructions for a particular area, or for a particular land use, is represented by A,. As building density increases (fig. 24 to B), the total amount of incident shortwave radiation per unit ground area also increases. Figure 2C is in- cluded to show that past a certain critical value of SILRAT, the amount of incident shortwave radiation will decrease as the density increases. The increase in shadowed area with larger zenith angles is roughly approximated by: SHDRAT=[1-cos (zenith angle)] ®. (5) E 4 : A N \\\\\\\\\ C FIGURE 2.-The interception of direct shortwave radiation by buildings. The net longwave radiation is computed as: Riy4 =[ (1-SILRATYX8SIGX® TSKY +] +[ (SILRAT-1)xSIG ® T sug*l, (6) where SIG is the Stefan-Boltzman constant and TSKY is the apparent radiant temperature of the sky hemisphere. Emissivity is assumed to be 1. In this case, SILRAT is used as a value roughly equivalent to a variable called the view factor, de- fined as that portion of radiation leaving a surface that is intercepted by another surface. A repre- sentative sky hemisphere, as "seen" from a point radiation source, is illustrated in figure 34. This radiation source is located at point a in figure 3B. Note, however, that the value of the view factor could change somewhat depending on the position of a in figure 3B, while SILRAT would remain con- stant. The approximation is used only in the absence of actual values for the view factor. The two terms that contribute to turbulent heat transfer are each computed using analogous rela- tionships. Sensible heat flux to the air can be cal- culated as: H= _KEouCr 3 y % R3" |x (Ia-Iwny. C (In(z4/2z0)) THE INFLUENCES OF LAND USE AND LAND COVER IN CLIMATE ANALYSIS NE‘vaporative heat loss is similar: 2 1p = _K PEE _ 3 x |(1-18 x Ri)" | % (Qai,~me),(8) (In(za/2z0)) where: k = von Karman's constant (0.4) p = atmospheric pressure w = wind speed at height z C, = specific heat of air at constant pressure %a = atmospheric damping height (dis- cussed below) z, = surface roughness length (discussed below) Ri = bulk Richardson number Tu, = air temperature Tu; = surface temperature Qu, = specific humidity of the air Qu; = specific humidity at the ground surface L = latent heat of vaporization. In general, wind speeds are slower near the ground than they are away from the surface due to the drag exerted on moving air by the ground surface. When temperatures are nearly constant with height, wind speed increases linearly with the natural logarithm of the height (Sellers, 1965). If one assumes some layer of air close to the ground, across which heat is transferred solely by molecular processes, it is possible to imagine a non-zero height, z,, at which the wind speed is zero. This height is called the roughness length. Presumably, as the surface texture gets rougher, the lower surface of the boundary layer rises due to turbulence, and the height at which the wind speed is zero also rises. The height above which the effect of the surface roughness is no longer felt, the atmospheric damping height, is represented by Za. The logarithmic relationship between 2;, z,, and u has been derived from observations over uniform flat surfaces to a height of about 20 m (Sellers, 1965). Under these conditions, z, can be derived as the y-intercept of the log-law relation. Since the assumption of a uniform surface is clearly violated in urban areas, Lettau suggested an approximation of z, as a function of the average height of obstruc- tions and the silhouette ratio (Lettau, 1969) : 2,=0.5x h, x SILRAT (9) He also notes that the parameter SILRAT loses # ~* TESTING AN URBAN CLIMATE SIMULATOR E 5 validity as it approaches unity. Like SILRAT, the use of z, can be abused in extreme conditions. Look:- ing again at figure 2, it is clear that, as the build- ings crowd closer together, there will be a point when the roof tops function as the new ground sur- face and z, goes back toward 0. The second factor in equations 7 and 8 corrects for the unstable air conditions occurring through- out most of a diurnal cycle. The bulk Richardson number, Ri, in this adiabatic correction function is computed by Sellers (1965) as: RI: (980/ Tail) X [(Tair Tsurf)/ lrl(2d/Z/ZO)] (10) The urban terrain parameter SRHF, the surface relative humidity fraction, is used to compute the C3 << NORTH B FiguUurE 3.-The interception of longwave radiation by build- ings and vegetation (adopted from Myrup and Morgan, 1972). A, Sky hemisphere view from point a; view factor-(1-SKY)/total area. B, Site view. specific humidity at the surface: .622p X Ep X SRHF (11) p-Ep where E, is the saturation vapor pressure. Myrup (1969) used the surface relative humidity fraction, SRHF, instead of surface relative humidity because of the heterogeneous nature of a given urban area. Since the surface relative humidity is considered to be 100 percent when the soil contains enough water, Myrup calculated SRHF by finding the relative proportion of an area covered by freely transpiring vegetation. To calculate the soil heat flux, the model solves S=r/Ax (T2 - Tur); (12) where A is the thermal conductivity and Az is the change in depth. In this model, eight nodes are used to compute the soil temperature profile. The surface node is T,,,, and T2 is the ground tempera- ture at the first node below the surface, set at 5 cm in this study. The program uses an implicit solu- tion to a thermal diffusion equation to update the soil temperature profile based on a method sug- gested by Outcalt and Carlson (1975). Thermal conductivity is not an input parameter due to the diverse surfaces and materials in any one land use. Instead, thermal diffusivity, «, is calcu- lated as a function of SRHF by scaling « between the diffusivity of dry concrete (0.02 ecm" sec") and wet soil (0.005 ecm* sec). «=(SRHF x 0.005)+[(1-SRHF) 0.02]. (18) The assumption is that as the SRHF becomes smaller, the percentage of surface covered by con- crete becomes larger, and the value of « approaches the « of concrete (Outcalt, 1972a). Thermal con- ductivity is found as the product of « and an aver- age heat capacity of 0.21 J em (Sellers, 1965). Equations 2 through 13 may now be used to esti- mate the effects of changes in the urban terrain parameters, silhouette ratio, height of obstructions, surface relative humidity fraction, shortwave albe- do, and thermal diffusivity on the surface tempera- ture. The variable SILRAT is used in the computation of R,, H, and LE. As SILRAT increases, Ry} - Rt becomes less negative, making R, larger. In equation 4, an increase in SILRAT also increases the amount of energy intercepted on vertical sur- E 6 THE INFLUENCES OF LAND USE AND LAND COVER IN CLIMATE ANALYSIS faces, further increasing R,. If all other variables are held constant, T.,, will increase. SILRAT also affects H and LE on the opposite side of the equal sign in equation 2 because it is used to compute z,. An increase in z, increases the turbulent exchange coefficient, the first factor in equation 7. The value of the second factor, the adia- batic correction for unstable air, also will increase but at a much slower rate. The overall effect of an increased z, will be to magnify the direction of flux as determined by u;. For example, if T.,; is larger than Tau, an increased z, will cause H to become more nega- ive. With all else held constant, a stronger H away from the surface will cause T,,,, to drop. Similarly, an increased z, increases the magnitude of LE, with the sign depending on the relation Qai-Qur;. It would be possible to have the signs of and Toa-T.; such that an increased 2, would magnify LE away from the surface but in- crease H toward the surface. Equation 9 shows that A,, the height of obstruc- tions, is linearly related to z,. One would expect an increase in A, to have the same effect on H and LE as a comparable increase in SILRAT if R, was held constant. Note that an increased A, would not create an increased vertical surface area since SILRAT is input separately. The effects of changes in ALB and « are the most transparent of all the parameters. An increase in ALB causes R.,4 -R.,f to decrease and the value of T.; must, therefore, decrease to compensate. Equation 12 demonstrates that an increased a would increase the magnitude of the soil heat flux. For example, at midday T,,,, is much higher than T2. An increased A would increase the heat flux into the ground and cause T,,,, to drop if all else were constant. The thermal profile would also adjust more quickly to changes at the surface since the speed of diffusion has increased. The parameter SRHF affects both LE and S. As equation 183 shows, an increase in SRHF de- creases the value of «k which lowers the magnitude of S. The value of Q,,; in equation 11 is also deter- mined by SRHF being affected by it linearly. The relative magnitude of Q,,, then determines whether the flux of evaporation increases or decreases. For example, an increase in SRHF would force LE to become more negative if Q,,, was larger than Qu, forcing T.,, to decrease. Of the five urban terrain parameters, only changes in ALB cause predictable changes in sur- face temperature. In all other cases, the direction of the component flux is a function of the sign of the temperature or specific humidity difference. In the case of SILRAT, h,, and SRHF, more than one component is adjusting to the change. THE MODEL The model used for this study is based on a surface-energy budget simulator developed by Out- calt and Carlson (1975). Their report describes in detail the structure of the program. This section generally outlines the model used, (URBD), and describes changes made to accommodate the urban influences on the energy-balance components as dis- cussed in the last section. If a city is considered as a mosaic of textures and materials, an urban model must differentially respond to the same meteorological conditions based on certain site characteristics. Outcalt's original study of land use effects (19722) analyzes the responses of four land use types, each defined by the specification of three urban terrain parameters, surface relative-humidity fraction, obstruction height, and silhouette ratio. For this study, albedo was added in order to ac- centuate the differences between heavily and lightly vegetated areas. In addition, roughness length was input instead of obstruction height since the data used to operate the model were provided in this form. This substitution is merely for convenience and does not affect the simulated temperature val- ues since z, and A, are linearly related as shown in equation 9. Other input variables include latitude of the site and the time of year, which are used to calculate the solar geometry. The density of dust particles in the air, and amount of precipitable water to calcu- late the amount of shortwave radiation transmitted through the atmosphere, are also specified as input variables. Air temperature, atmospheric pressure, relative humidity, and wind speed are read into the model as the meteorological boundary conditions. As the model is currently written, these last four vari- ables are specified for each time step the model is run, every hour in this case. The model in this form is referred to as the observed hourly input model. With slight modification in the input pro- cedure, however, the model can also run on mean daily values for those variables using a wide range of time intervals. This form of the model is called the daily mean value input model. The overall flow of the program is illustrated in figure 4. Notice that the model cycles through the wa TESTING AN URBAN CLIMATE SIMULATOR daily (24 hourly) data twice in order to stabilize the soil temperature profile before solving for the equilibrium surface temperature. Fed by a high guess and a low guess for the surface temperature, the SECANT algorithm computes a new guess with the following relationship: BALt—1_BALz—2 Thew=Tr- (14) where T;, refers to the previous guess of the tem- perature and T;, refers to the guess prior to T.;... BAL is the current sum of R,, S, H, and LE. The routine converges to BAL=0+0.7TW/m* usually within six iterations. The subroutines used in the Outcalt and Carlson URBD model (1975) were not designed for city use, which necessitated some restructuring and modification. For example, the shadow generator, SHDRAT, has been included in subroutine CPATH, the solar geometrician. Subroutine SOLURB, which calculates shortwave radiation, does not allow the albedo to vary as a function of the angle of inci- dence. Such a refinement would not be justified since the albedo for any land use is an average for such diverse materials as roofs, vegetation, and concrete. Subroutine RUNTRI has also been simplified by disallowing variable porosity and thermal conduc- tivity in the substrate. Instead, thermal diffusivity is assumed to take on values that can only range between the diffusivity of dry concrete (roughly 0.02 cm" sec") and that of wet soil (roughly 0.005 cm* sec). It is necessary in this model to establish a lower boundary for the soil heat flux in order to simulate the diffusion of heat through the substrate at each time step. The damping depth, 24, is calculated in subroutine CITY using the Terzaghi relationship (Outcalt and Carlson, 1975) : zgq=(12 X t X K)l/2, (15) where t represents the period of the temperature fluctuation one is concerned with. In this case, one is dealing with diurnal variation so t = 4.82 x 10* sec (12 hours). Using the extremes of « as end- points, the range of z, is between 51 cm and 102 ecm. APPLICATION OF THE MODEL To test the response of the model to differing urban terrain characteristics, data were obtained E 7 from a project conducted by the U.S. Geological Survey (USGS) (Alexander and others, 1976). One of the principal goals of the USGS investiga- tion was to evaluate how well thermal scanner data, collected by spacecraft, could be used for studying urban climates. The data for this model analysis come from an early experiment in the USGS study. On May 11, 1972, the Environmental Research Institute of Mich- igan (ERIM) directed three aircraft flights over Baltimore, Md., at an altitude of approximately 1,500 m. The third ERIM run, flown between the hours of 1345 and 1415 e.d.t., provides the observed data used for the verification of the URBD model. The flight path is shown in figure 5, running across Baltimore from the northwest to the southeast. CALL READIN geographic and meteorologic END data urban terrain parameters WRITE: A,, S, H, LE and equilibrium surface temperature CALL CMY find substrate diffusivity soil damping depth A YES CALL EPHEM Second. day find solar declination of calculation ? CALL RUNTRI Initialize soil NO condition soil profile temperatures temperature profile 3 YES CALL CPATH find solar geometry for each time step CALL UBAL CALL SOLURB No find shortwave radiation input ( eos surface temperature CALL SECANT provide new guess for surface temperature FUNCTION UPRIG find amount of radiation on vertical surface CALL UBAL use low guess for surface temperature, find A,,, S, H. LE CALL UBAL use high guess for surface temperature, find A,,, S, H, LE FIGURE 4.-Flow chart of the Outcalt and Carlson URBD model (1975). E 8 76:30’ 76:45’ 76°37'30" I o Timonium 39° 15" X 0 _ 2 4 6 _ 8 KILOMETERS \\ 0 2 4 MILES Baltimore-Washington International Airport -/ 1 1 FicurE 5.-Path of the third flight across Baltimore, Md., along which multispectral scanner data (including ther- mal) were obtained by the ERIM aircraft on May 11, 1972. Weather records for May 11, 1972, show that the day was clear and cool with a steady breeze from the west. Temperature ranges for May 9-11, 1972, listed in table 1, support the observation that a cold front passed through the area on May 10, 1972, bringing a fairly homogeneous air mass over the Washington-Baltimore area (Alexander and others, 1976). Under these conditions, hourly meteorologi- cal observations from Baltimore-Washington Inter- national Airport, located eight miles southwest of Baltimore in the direction of Washington, are repre- sentative of the region. These hourly values of air temperature, wind speed, air pressure, and relative humidity are listed in table 2. The principal advantage of the third May 11, 1972, flight, over the two flown earlier that day, is that it scanned an area with a high diversity of land use types. The effects of seven land uses on surface temperature were simulated in this study to give an idea of the possible range of microcli- mates within a city. These land uses are: low- density residential (LDR), medium-density resi- dential (MDR), high-density residential (HDR), THE INFLUENCES OF LAND USE AND LAND COVER IN CLIMATE ANALYSIS central business district (CBD), commercial (COMM), transportation (TRANS), and urban parks (PARKS). The urban terrain characteristics of these land use types are presented in table 3. Values of these parameters were determined early TABLE 1.-Recorded meteorological data [Washington, D.C.: T,,,, occurred at 1640 e.s.t., T,,,, occurred at 0515 e.s.t. Highest observed wind speed since 0800 : 671 em sec-' at 1720 e.s.t. from west. Source: "Washington Post," daily weather information, May 10-12, 1972 ] Tuesday, May 9, 1972 (24 hours ending at 2000 e.s.t.) Tmest°C) Tnain(°C) _ Precip. (cm) Baltimore 28 14 0.02 Washington, D.C. ___ 28 16 18 Wednesday, May 10, 1972 (24 hours ending at 2000 e.s.t.) Tman(°C) Tmln(°C) Precip. (cm) Baltimore: ......._:c.. 12 10 2.11 Washington, D.C. ___. 13 11 1.57 Thursday, May 11, 1972 (24 hours ending at 2000 e.s.t.) Tmss('C):Tmis(*C) ~ Precip. (cm) Baltimore P1 4 0.0 Washington, D.C. ___. 28 4 .0 TABLE 2.-Surface weather observations at Baltimore- Washington International Airport, May 11, 1972 [Source: National Climate Center, National Oceanic and Atmospheric Administration, Asheville, North Carolina 28801] Time Ta‘r w P RH (e.s.t.) (°C) (cm sec-) (mb) (percent) 0055 7.2 360 1021 0.60 0155 7.8 309 1021 .56 0255 5.6 257 1021 60 0355 5.6 257 1021 .60 0455 4.4 309 1021 64 0555 5.0 257 1022 .67 0655 8.3 206 1028 .65 0755 12.8 309 1023 .55 0855 15.0 515 1023 Al 0955 15.6 515 1023 .36 1055 16.7 463 1022 .34 1155 18.3 515 1021 .82 1255 18.9 112 1021 .30 1355 19.4 412 1020 .81 1455 20.6 669 1019 .31 1555 21.1 566 1018 .81 1655 21.1 566 1018 .31 1755 20.6 566 1018 .31 1855 18.3 463 1017 .34 1955 16.1 463 1018 Al 2055 15.6 463 1018 44 2155 14.4 463 1019 AT 2255 12.8 412 1019 .58 2855 12.8 257 1019 .53 Mean 13.9 431 1020 50 TESTING AN URBAN CLIMATE SIMULATOR TABLE 3.-Land use characteristics [Source: Lewis and Outcalt (1976)] Land use ALB SRHF Zo Type (percent) (percent) (cm SILRAT Low density residential (LDR) -:ccn.s_0lc~. 0.18 0.60 78 0.13 Medium density residential (MDR) - .18 .30 86 19 High density residential (HDR) __ .14 .01 110 19 Central business | district (CBD) ___ .15 .05 400 22 Commercial (COMM) .. _ .15 .05 78 .02 Transportation (TRANS) ........_ 15 15 6 .02 Parks - Urban (PARKS) _...__.... .23 .90 80 i174 in the USGS-National Aeronautics and Space Ad- ministration study (Alexander and others, 1976) and were reported in Lewis and Outcalt (1976). Although the category names may connote spe- cific images of land use, these categories should not be interpreted too literally. For instance, the medi- um-density residential category specified by Lewis had a roughness length of 86 cm and a silhouette ratio of 0.19. In contrast, Myrup and Morgan (1972) used respective values of 532 ecm and a 0.50 ratio to describe the same category. The land use categories used in this study are, therefore, general labels assigned for evaluating the influence of the urban terrain parameters. A clearer view of the relationships between the land use categories is provided in figure 6. Low density, medium density, and high-density resi- dential, for example, are chiefly distinguished by differences in surface relative humidity. The com- mercial and central business district categories both have very little evaporative surface, but their height and roughness characteristics are clearly dissimilar. Albedo was added as an urban terrain parameter primarily because it distinguishes natural surfaces from artificial surfaces. Vegetation, in general, tends to reflect much more radiation in the near infrared range than do nonvegetated surfaces (Sel- lers, 1965). This effect would tend to decrease the amount of energy absorbed, decreasing the surface temperature. A simulation run, using the URBD model, for May 11, 1972, was made using the data in tables 1 and 2 in order to test the response of the model to changes in the urban terrain values. In addition, latitude was specified as 89.9° lat., precipitable water was 10.0 mm, and dust particles were given as 1.0 particles ecm~>* (Lewis and Outcalt, 1976). E9 An observed sky radiant temperature, TSKY, of -5° C was used rather than the common formula of TSEKY=T,-22° C. An emissivity of 1.0 was assumed throughout the model. Sample output data for the simulation of the micrometeorological conditions on May 11, 1972, in the medium-density residential land use category are shown in tables 4 and 5. The first five columns on the left in table 4 repeat meteorological boundary conditions input from table 3. The five columns on the right show the results of the shortwave radia- tion calculations. The shortwave radiation flux on a horizontal surface outside the Earth's atmosphere is depicted as HXT. The sum of the BEAM, dif- fused, and backscattered radiation flux on a hori- zontal surface at the Earth's surface is denoted as SUN. The total shortwave radiation flux on a verti- cal surface normal to the solar azimuth angle is represented as VERT. The TERRAIN variable averages SUN and VERT by considering the sil- houette ratio and amount of area in shadow. The energy balance equation components for each time interval (3600 sec in this case) and the equilibrium surface temperature are listed in table 5. At 0700 (solar time), for example, the model calculates that 344 W/m of total shortwave ra- diation fall on a horizontal surface in Baltimore and 759 W/m strike a vertical surface facing the sun. Since the silhouette ratio of vertical to horizontal surfaces for medium-density residential is 0.19, the vertical surface flux does not contribute significant energy to the overall average shortwave radiation flux of 344 W/m. At this hour, the net radiation flux (R,) is 243 W/m" due to the sun- light and rapidly increasing air temperature. Heat is being conducted into the soil (away from the surface) at a rate of -243 W/m while the sensible heat flux to the air is -167 W/m, indicating that convection is removing heat from the surface. Evaporative heat flux (LE) is still positive, 166 W/m, indicating that heat is being added to the surface as water vapor condenses, forming dew. Graphs of the simulated surface temperatures, as calculated from the standard run for each land- use category, are presented in figure 7. Driven by the four meteorological variables listed in table 2 at each time step, the URBD model computes the hourly temperatures as the curves labeled "H." The input variable reading procedure in the model was modified to use the 24-hour temperature aver- age for the four meteorological variables instead of observed hourly values. The curve labeled "D" shows the resulting surface temperature. E 10 THE INFLUENCES OF LAND USE AND LAND COVER IN CLIMATE ANALYSIS Pls e LZ 440 als cep 400 | 4 CC A- E 360 |- < iz #L 8 320|~ < A § 280 | = * el 7. 5 ss PARKS wd A- w 9 200 | é t- LDR 3 |- 3 160 (d . ‘ CC “30 4 e $955 .8 MDR S 120 ~ Cy 3 $€%‘\°: I HDR 3 comm QPC‘Q' Ton. ;: g 80 |- 90“ 4 5 < 3L~~ TRANS 40 7 ) 9 .02 ba D6 --of . 10 ./ 12 _ 14 _ 16. .18 . 20... 22 SILHOUETTE RATIO FIGURE 6.-Site characteristics for seven land use types. ANALYSIS OF THE MODEL The seven standard runs of the URBD model, shown in figure 7, illustrate the differences between surface temperatures calculated with observed- hourly-value input and daily-near-value input. The daily input curves are relatively smooth and sym- metrical around solar noon. The symmetry is not 1. Is there a difference between driving the model surprising, since energy is being added and sub- with mean daily values for air temperature, rela- tracted from the model by net radiation. The daily tive humidity, wind speed, and air pressure, input model also generates curves with shallow When the simulations of surface temperature have been completed, the effectiveness of the URBD model can be evaluated by answering the follow- ing questions: and driving the model at each time step? ranges, compared to the 20° C changes in the hourly 2. How sensitive is the model to changes in the input model. In addition, the maximum surface urban terrain parameters? temperatures of hourly input lag behind the daily 3. How do the simulated surface temperatures com- - input model by two to three hours. pare with observed results? To determine the underlying causes for these TESTING AN URBAN CLIMATE SIMULATOR E 11 TABLE 4.-Simulation of solar radiation flux for medium-density residential land use, May 11, 1972, Baltimore, Md. [Latitude=39.9° Month=5 Soil wet fraction=0.30 Soil thermal diffusivity=0.015 Precipitable water (mm)=10.0 Silhouette ratio=0.19 Time increments=3600 Dust particles (¢cm-*)=1.0 Roughness length (cm=86] Solar declination=17.6° Day of the month=11 Albedo=0.19 » (Radiation values, in W/m-) Solar atr u P time (°C) (em sec-) (mb) RH EXT BEAM SUN VERT TERRAIN 0.00 7.2 360 1021 0.60 0 0 o o a 1.00 7.8 309 1021 .56 0 0 0 0 0 2.00 5.6 257 1021 .60 0 0 0 0 0 3.00 5.6 257 1021 60 0 0 0 0 0 4.00 4.4 309 1021 64 0 0 0 0 0 5.00 5.0 257 1022 67 6 0 3 3 3 6.00 6.3 206 1023 .65 265 101 144 556 162 7.00 12.8 809 10283 .55 524 291 344 759 344 9.00 15.0 575 10283 Al 765 491 554 786 587 9.00 15.6 575 1023 .36 972 671 729 722 697 10.00 16.7 463 1022 in 1131 812 872 609 807 11.00 18.3 515 1021 82 1231 901 970 494 865 12.00 18.9 772 1021 ET 1307 932 994 440 883 13.00 19.4 412 1020 '81 1231 902 963 494 866 14.00 20.6 669 1019 31 1131 812 873 610 807 15.00 21.1 566 1018 .81 972 671 781 728 699 16.00 21.1 566 1018 .81 765 492 549 831 588 17.00 20.6 566 1018 .81 524 202 344 761 345 18.00 18.8 463 1017 .84 265 101 145 559 162 19.00 16.1 463 1018 Al 6 0 3 3 3 20.00 15.6 463 1018 44 0 0 0 0 0 21.00 14.4 463 1019 AT 0 0 0 0 0 22.00 12.8 412 10198 53 0 0 0 0 0 model differences, the data were analyzed using a variation of the Students' t-test as described by Outcalt (1972a). The maximum surface tempera- tures, taken from the graphs, were regressed against four of the urban terrain parameters in two mul- tiple-linear regression models: one for the daily mean value input, and one for the hourly observed TABLE 5.-Simulation of micrometeorological conditions for medium-density residential land use, May 11, 1972, Balti- more, Md. (W/m) Solar Time R. 8 H LE Tf CC) 0.00 -84 =47 -69 222 8.4 1.00 -88 -55 -89 178 8.6 2.00 -81 -41 -29 157 6.3 3.00 -81 =41 -29 157 6.3 4.00 -81 -63 -45 188 5.4 5.00 10 -66 -41 185 6.0 6.00 72 -156 -82 167 10.7 7.00 248 -248 -167 166 16.5 8.00 434 -288 -246 50 18.7 9.00 592 -265 -278 -54 19.7 10.00 697 -808 -283 -112 21.3 11.00 755 -292 -207 -165 22.8 12.00 T177 -232 -820 -225 22.5 13.00 758 -810 -264 -179 24.2 14.00 699 -230 -281 -188 24.1 15.00 591 -211 -227 -158 24.4 16.00 434 -159 -171 -105 28.6 17.00 246 -95 -108 -49 22.0 18.00 69 -58 -54 48 19.2 19.00 -87 -40 -87 166 16.7 20.00 -91 -56 -54 201 16.4 21.00 -91 -70 -66 227 15.5 22.00 -90 -89 -47 256 14.1 28.00 -89 -66 -86 185 13.7 value input. The Students' t-test was then used to compare the slope (b value) for a parameter in one model with the slope for the same parameter in the other model. When the slopes are significantly different, it can be inferred that the action of the test terrain parameter in both models helped create the resulting differences in surface temperature. The results of this test, applied to both daily maxi- mum temperatures and temperature ranges, are listed in table 6. The similarity of the maximum temperatures ob- served in figure 7 suggests that there were no signif- icant model differences caused by the terrain param- eters, and the t-test confirms this. All eight test statistics for (H-D)/S.E.(H) and (H-D)/S.E.(D) fall below the critical region. It is surprising, though, that the large differences in range cannot be attributed to any of the variables tested. Despite diurnal range differences of up to 20° C, the test statistics were even smaller than in the test on maximum surface temperature. The next step was to examine the sensitivity of the hourly input model to changes in the terrain parameters. Sensitivity testing of a model can be approached several ways. At one extreme, the model can be subjected to large perturbations in parameter values and the temperature responses can then be examined. At the other extreme, parameter values can be changed by the potential experimental measurement error (Myrup and Morgan, 1972). E 12 THE INFLUENCES OF LAND USE AND LAND COVER IN CLIMATE ANALYSIS TEMPERATURE, IN DEGREES CELSIUS TEMPERATURE, IN DEGREES CELSIUS | I | 2.00 1 | | | | | o.oo 6.00 1200 18.00 24.00 0.00 6.00 1200 18.00 24.00 0.00 6.00 1200 1800 24.00 0.00 6.00 1200 18.00 24.00 SOLAR TIME SOLAR TIME SOLAR TIME SOLAR TIME D E F G FIGURE 7.-Simulated surface temperatures in Baltimore, Md., for various land use categories based on hourly and aver- aged daily values on May 11, 1972. H indicates meteorological variables input hourly; D indicates meteorological vari- ables input daily. A, Low-density residential. B, Medium-density residential. C, High-density residential. D, Central business district. E, Commercial land use. F, Transportation land use. G, Parks-urban land use. In climate simulation, it is often meaningless to have large perturbations in just one variable since such a change is normally correlated with changes in a number of other variables. In this study, a middle ground between the extremes was sought by. increasing and decreasing terrain parameter values by 830 percent. The control run temperatures for 0500 and 1300 hours, using the land use input data described previously, are shown in table 7. Each of the param- eters, ALB, h,, SILRAT, «, and SRHF were then altered by 80 percent, while all other variables were kept equal to their control run values. The results of these tests are listed in tables 8-12 and summarized in table 13. Very little change in surface temperatures can be discerned from the data at 1300. The overall average deviation from the control temperature is +0.53° C. The maximum deviation, -2.7° C, occurred when SRHFx1.3 was applied to the category PARKS. Over this range of perturbation, all five parameters acted linearly, causing temperatures to vary as much above the control values as below. Compare, for example, the columns labeled "T.,,,- To", at 1800 in table 10. A 30 percent decrease in SRHF creates a temperature change with the same magnitude, but opposite direction, as a 80 percent increase in SRHF. Overall, changes in SRHF caused the largest temperature changes. Looking at the tables, it is clear that changes in SRHF had the greatest effect on those land use types with high SRHF values. Conversely, « had the most influence on those land use types with low SRHF values. The results at 0500 are not as easily explained. In almost all tests, both increased and decreased TESTING AN URBAN CLIMATE SIMULATOR TABLE 6.-Significance test for differences between URBD model using observed hourly data and mean daily data as input bare bsenr big bsura: Maximum surface temperature Hourly values (H) -_. -16.0 -12.6 0.002 2.8 Daily values (D) -_ -19.4 -16.1 -.007 2 Standard error, S.E. (H) . 48.7 4.1 009 6.5 Standard error, S.E. (D) .. 92.4 8.6 009 13.8 (H-D)/S.E. (H) -__. .08 .9 2.25 82 (H-D)/S.E. (D) -__- .04 A 1.0 15 Diurnal temperature range Hourly values (H) -_ -14.8 -8.0 _- -0.004 _ -1.7 Daily values (D) -___ -29.6 -1.9 -.01 -5.8 Standard error, S.E. (H) -. 58.7 5.0 .005 8.0 Standard error, S.E. (D) _. 92.8 8.6 009 13.8 (H-D)/S.E. (H) ___. 28 -1.22 1.2 Ab (H-D)/S.E. (D) ___ 16 - T1 .67 26 parameter values caused a slight increase in temper- ature. The only exceptions were those land use types which had an extreme value for the parameter being tested. Although the model appears to be ignoring the magnitude of the change in each parameter, it is more likely that the change does not have a linear effect on the energy-balance com- ponents. By 0500, the surface temperature had a number of hours to come to equilibrium with T,,, and T2 in the absence of additional radiation. Since the signs of the values Tar - Ta and T2 - Tay control the direction of flux, the components could be particularly sensitive. The results of the test for differences between the hourly input model and the daily input model, and the sensitivity tests suggest that changes in individual urban terrain parameters have very little impact on the URBD model. The wide diurnal range of surface temperatures observed in figure 7, E 13 TABLE 7.-Control run values of the URBD model for use in sensitivity testing. [T, is the control run temperature] T, (°C) Land use 0500 1300 LDR . erent a da' lns ae aes 4.6 20.8 MDR 6.0 24.2 HD R _.. : 4 Ew sa 2) ae i te th ca nen i oe nut is in he se on se man 7.5 29.0 CPD - »« ~ s 9.1 27.9 COMM 1 Ll -. 2 a a cem all an a me ne ol ae on mile e we we se aie 6.8 27.5 TRANS | co . so uce co o t s e ie as h os r e m he hl in mn aaa in or 5.0 27.1 PARKE 12 .on 8.1 17.4 discussed earlier, suggests that wind speed and air temperature may be quite influential. Since hourly values for both wind speed and air temperature more than doubled over the course of May 11, 1972, such dynamic atmospheric conditions repre- sent a realistic test of the model. The response of the model to large changes in wind speed is shown in tables 14 and 15. By exam- ining the temperature change between 1x 0.25 and wx0.50 and comparing it to the change between wxX1.5 and 1x2.0, one can conclude that the model is more sensitive to changes in wind speed at low velocities. Land use types with little evaporative surface area, such as HDR, CDB, and COMM, how- ever, showed very little response to changes in speed. The test of the model response to changes in air temperature was combined with the verification of the model so the two will be discussed simultane- ously. Observed radiant surface temperatures at 1400 e.s.t. for Baltimore on May 11, 1972, are shown in table 16. These values were corrected temperatures obtained from the ERIM thermal scanner data (Lewis, Outcalt, and Pease, 1975). As shown in figure 8, the model consistently underpredicts. Since the average deviation from the line of per- fect prediction is -14° C, 14° C was added to each hourly air temperature, and the model was rerun. By adding 14° C, the average hourly temperature TABLE 8.-Sensitivity of the URBD model to changes in substrate diffusivity (x) [Temperatures in degrees Celsius] k X13 Land use 0500 1300 0500 1300 Tuaf Tur~Ic Tsr~Ic Tua Tur-~Tc Tnaf Tug -Ic LDR i. sn 2 non eon an mer nl o ae mnie males 5.0 0.4 20.9 0.1 5.0 0.4 20.7 =-O.1 MDR «.... 2... ue ae ce en 6.6 6 24.8 .6 6.2 2 28.7 -.6 HDR -s... . - cn ac le eee ake me mad Hae 8.4 .9 80.8 1.8 7.8 > C 27.6 -1.4 CBD 22 000004 url memes 9.8 7T 28.8 .9 8.8 =.8 27.1 =,8 COMM -.. : .. . .. «sn ulan hane al selene 7.6 .8 29.0 1.5 6.8 0 26.3 -L1i2 TRANS: ..on onic coas aarens 5.7 AA. 28.9 1.8 5.4 4 25.7 x14 PARKE 8.3 ] 17.8 =i 8.6 5 17.5 1 E 14 THE INFLUENCES OF LAND USE AND LAND COVER IN CLIMATE ANALYSIS TABLE 9.-Sensitivity of the URBD model to changes in the height of obstructions (hi) [Temperatures in degrees Celsius] h, x0.7 b, X1.3 Land use 0500 1300 0500 1300 Tuer Tug~Tc Tow Tug~Ic Tow: Tug-Ic r Tug-~Ic LDR .. . 4.9 0.3 21.4 0.6 5.0 0.4 20.3 -0.5 MDR. -==... 6.2 2 24.6 4 6.5 .5 28.8 =.4 HD o . » en een erea wana a ae ince 7.4 1 28.9 8.0 .5 29.0 .0 CBD 8.7 mad 27.9 0 9.6 .5 27.8 v1 COMM ceca 6.9 1 27.5 0 7.4 .6 27.4 het . TRANS -_ sola.... ect ccc rel 5.4 4 27.1 0 5.6 6 26.9 -.8 PARKS sec c 3.5 4 18.1 7 8.5 4 16.8 =.G6 TABLE 10.-Sensitivity of the URBD model to changes in the surface relative humidity fraction (SHRF) [Temperatures in degrees Celsius] f SRHF 0.7 SRHF x1.3 Land use 0500 1300 0500 1300 Tug Tsar/T Tc Tnaf Twr e Tug-Ic Tsar Tag -Ic LDR .< omnes 5.7 1.1 22.8 2.0 4.1 -0.5 18.9 - L.Q MDR 6.7 yi 25.4 1.2 6.0 .0 23.0 =~1.9 HDR! oan e deve nnn ana Gauss 7.8 .8 29.0 .0 7.7 Aj 28.9 -O.1 CBD - sel- ul 9.4 .8 28.3 4 9.1 .0 27.4 COMM .= .A... See 7.2 4A 271.1 4 ".A .8 271.2 =~.G8 TRANS ... a_ ices. 5.6 .6 27.2 L 5.5 5 26.7 -.4 PARKE c 4.9 1.8 20.1 2.7 1.8 -1,8 14.7 ~2.1 TABLE 11.-Sensitivity of the URBD model to changes in silhouette ratio (SILRAT) [Temperatures in degrees Celsius] SILRAT 30.7 SILRAT x 1.3 Land use 0500 1300 0500 1300 o Tug-Tc e Twr 9 Tur -Ic Tap Tug-Ic ss antes 4.9 0.3 21.3 0.5 5.0 0.4 20.4 -0.4 MDR LL 6.1 AA 24.4 4 6.5 .5 24.0 -.8 HDR 7.8 e 28.7 -.8 8.1 .6 29.2 2 CBD 8.7 - 27.8 9.7 .6 271.9 .0 COMM. .n sc ise an- 6.8 .0 27.4 -.1 7.4 .6 27.5 .0 TRANS .se 5.4 A 27.0 5.6 .6 26.9 mg +1 PARKS ses sec 8.5 4 18.0 .6 8.5 A 17.0 ~.4 TABLE 12.-Sensitivity of the URBD model to changes in albedo (ALB) [Temperatures in degrees Celsius] ALB X 0.7 ALB X1.3 Land use 0500 1300 0500 1300 Tua Tag-ITc Tug Tnaf Tug~Ic Tug -Ic LDR Gove... nee cae 5.0 0.4 21.1 0.8 5.0 0.4 20.5 -0.8 MDR ~";.3 c.. uR L 6.3 BJ 24.5 .8 6.3 .8 23.9 HDR .Y 7.8 3 29.2 2 7.8 .8 28.7 -.8 CBD _ _s 2 n salsa iim aniem 9.8 2 28.0 1 9.3 2 27.7 -.2 COMM "...no Tik 3 27.7 «2 7.1 .8 21.2 -.8 TRANS: - cellule ll ele a 5.5 5 27.4 .8 5.5 5 26.6 -.5 PARS... » cen. cnn o ran an nae aed 8.5 4 17.8 4 8.5 A 17.0 -.4 TABLE 13.-Rank (R) of absolute value of T suy-Tp at 1300 for each land use type [Temperature value is the average of parameterx0.7 and parameterX1.31 LDR MDR HDR CBD COMM TRANS PARKS Parameter °C (R) °C (R) °C (R) °C (R) °C (R) °C (R) °C (R) SRHF .. 1.95 (1) 1.20 (1) 0.05 (3) 0.45 (2) 0.25 (2) 0.25 (3) 2.170 (1) Ro .55 (2) A0 (8) .05 (8) 00 (5) .05 (8) 10 (5) .65 (2) SILRAT 45 (8) 20 (5) 25 (2) 05 (4) 05 (3) 15 :(4) 50 (8) ALB .80 (4) 80 (4) 25 (2) 15 (8) 20 (2) 40 (2) A0 (4) K| tinier 10 (5) 55 (2) 1.60 (1) 85: (1) 1.35 (1) 1.60 (1) 10 (5) TESTING AN URBAN CLIMATE SIMULATOR E 15 TABLE 14. -Sensitivity of the URBD model to decreased wind speed (n) [Temperatures in degrees Celsius] ux 925 ux 9.50 Land use 0500 1300 0500 1300 Tug Tug-Ic Tnup Taq -Ic r Tug-Ic r Tug -It LDR... mass 4.6 0.0 25.4 4.6 4.8 0.2 28.1 2.8 MDRAL 1.3.30. ..I IEICE 5.4 =.6 26.9 2.7 5.8 -= 2 25.6 1.4 HDR 6.0 =L1.6 28.8 ~.2 6.8 28.8 e CBD _ .... ooo 7.4 28.2 .8 8.4 =-.T 28.0 A1 coOMmMM 5.7 =L.A 27.5 .0 6.8 =.0 27.5 .0 TRANS! ».. 4.8 =.2 27.8 aT. 5.1 1 27.5 A PARS. .nn2 ce 8.6 5 28.1 5.7 8.5 A 20.0 2.6 TABLE 15.-Sensitivity of the URBD model to increased wind speed (1) [Temperatures in degrees Celsius] ”X16 “x10 Land use 0500 1300 0500 1300 Trap Taag-Tc Tie Tug-Tc Teg List A \_} aTe Tug=Tc TDR o new anns ans 5.1 0.5 19.6 -T8 5.1 0.5 18.9 1.9 MDR! .... 220 coll - nen ese 6.7 +7: 28.4 =,8 6.9 .9 22.9 -L.8 HDR -so o clo clan e name an nine ne care's 8.4 .9 29.0 .0 8.9 1.4 29.0 .0 CBD | nene ene nee ein ee orer 9.7 .6 27.7 -.8 10.1 1.0 27.1 COMM ! ~.- 5 - c clam n arine oin nin case 7.8 1.0 27.4 =1 8.2 1.4 27 4 -I TRANS. .» ~ n ne car - mk 5.9 .9 26.6 -.0 6.2 1.2 26.4 *T PARKS... 8.4 .8 16.1 ~L.8 3.4 .8 15.3 ~B1 TABLE 16. -Comparison of observed and simulated surface tempera- ture, 1400 EST, May 11, 1972 (in degrees celsius) [Control run: »=0.95, by =0.62, I air+14°C: r=0.95, bi =1.09 (Lewis and others, 1975)] Turf Land use Observed Simulated Simulated (Conprol) (T + 14°C TDRs cs cern nls cee n cans ss 82.9 20.8 31.4 MDR 40.0 24.2 87.1 HDR _._ Jo Le 47.9 29.0 47.4 cOMM .__.___-__.____. 40.0 27.5 44.2 TRANS 40.0 27.1 40.0 PARKS 28.0 17.4 26.7 was doubled. The resulting surface temperatures average 55 percent higher than the control tempera- tures (see table 16). This second set of predicted temperatures is much closer to the observed values than those of the control run. If surface temperature was a linear function of air temperature, the slope of the regres- sion line for the control data would be expected to be the same as the slope for the T.,; +14° C data. These two slopes are 0.62 for the control run and 1.09 for the increased air temperature. Unfortu- nately, there are not sufficient degrees of freedom to test the hypothesis that the two slopes-are equal. A secondary check on the model is provided by global radiation data collected at Timonium, Md., on May 11, 1972, (Lewis, Outcalt, and Pease, 1975). The location of Timonium, about 12 miles north of TABLE 17.-Simulation of global incoming solar radiation {Measured on the parking lot at the State fairgrounds, Timonium, Md., on May 11, 1972 (Lewis and others, 1975)] Radiation (W/m-) Time (solar) Measured Simulated 0600 84 141 0700 286 338 0800 481 541 0900 725 722 1000 809 865 1100 941 955 1200 997 986 1300 955 955 1400 851 865 1500 662 724 Baltimore, is depicted in figure 5. The observed data shown in table 17 compares quite closely to simulated global radiation, particularly between 0900 and 1400. The maximum deviation between those hours is 6.9 percent. CONCLUSION The application of an energy-balance simulation model to urban area analyses requires recognition of the tremendous variety of surfaces, heights, ori- entations, densities, and materials found in a city. Although sensitivity analysis demonstrates that changes on the order of 30 percent, in individual urban terrain parameters, have relatively little ef- fect on surface temperatures, the effects can be E 16 51 ® 2 - g 27 |- {HDR (4) w 1 comm, E X § "[ o - |- TRANS o 39 |- g o/ «MOR € iF & F3 35 |- xQ‘ & $1 (s. $6 LDR = 31- 0OK a E aL £ HDR w comm g 27 |- XPARKS 8 { TRANS CC |- e- Rags _ MDR a [ LOR 5 a 19 |- © -| _ PARKS A. 15 J- | l {..} | | [.. | | | | [- 1 | l 15 19 23 27 31 35 39 43 47 51 OBSERVED SURFACE TEMPERATURE, IN DEGREES CELSIUS FIGURE 8.-Effect of increased air temperature on the pre- dicted surface temperature. Control run values repre- sented by a small circle and T... + 14°C values by an x. multiplied when more than one parameter are considered. The seven different land use types, speci- fied by the urban terrain variables, had varied re- sponses to the same meteorological boundary con- ditions. This result suggests that further work in analyz- ing the URBD model could start by using each of the three-dimensional cells of figure 6 as a unique "land use." One could then test the surface-temper- ature response to changes in three linked parameters rather than just one. The comparison of the hourly observed-value in- put model with the daily mean-value input model demonstrates the advantage of time dependent val- ues for wind speed and air temperature. The signif- icant differences suggest that the effects of changes in air relative humidity should also be tested. Al- though the use of an average air temperature did not appear to seriously affect the surface tempera- tures around midday, the consequences appear to be more serious at night. In the absence of any shortwave radiation, air with a temperature much warmer than the ground temperature provides a major source of heat to the system. The sensitivity of the model to air temperature and wind speed calls into question the use of uni- form meteorological boundary conditions for all THE INFLUENCES OF LAND USE AND LAND COVER IN CLIMATE ANALYSIS land use types. Characteristically, temperatures throughout a city are not uniform 6 feet off the ground. Since this value exerts such strong control on the sensible heat flux to the air and evaporative heat flux, its specification should be carefully studied. While the predictive value of the URBD model was deemphasized in this study, the goal of accu- rately simulating urban meteorological processes provides the impetus for modeling urban climates. In some ways, the fortuitous results of the first at- tempt to verify an urban simulator (Outcalt, 1972a and 1972b) may have misled investigators. The fact that Outcalt's simulated temperatures were within 1.5° C of observed temperatures at Ann Arbor, Mich., belies the complex energy exchange processes occurring in a city. A unique surface temperature occurs for a given set of energy fluxes, but an observed surface temperature can result from a numerous set of urban terrain parameters from which the four components of the energy balance equation are calculated. It should also be pointed out, though, that Baltimore is a larger, more com- plex site than Ann Arbor. In addition, the potential climatic effects of Chesapeake Bay are ignored by the one-dimensional URBD model. In this case, the real benefit of an urban climate simulator lies not in the predicted values but in the ability to examine relationships between com- ponents. An understanding of the sensitivity of the model, in turn, can be used to improve simulation techniques and field verification. REFERENCES CITED Alexander, R. A., Lewis, J. E., Lins, H. F., Jenner, C. B., Outcalt, S. I., and Pease, R. W., 1976, Application of Skylab data to land use and climatological analysis: Final Report SKLAB/EREP Investigation No. 469, 210 p. Bornstein, R. D., 1968, Observations of the urban heat island effect in New York City: Journal of Applied Meteorology, v. 7, no. 4, p. 575-582. 1975, The two-dimensional URBNET urban boundary layer model: Journal of Applied Meteorology, v. 14, no. 8, 1459-1477. Gutman, D. P. and Torrance, K. E., 1975, Response of the urban boundary layer to heat addition and surface roughness: Boundary Layer Meteorology: v. 9. no. 2, p. 217-233. Hage, K. D., 1975, Urban-rural humidity differences: Journal of Applied Meteorology, v. 14, no. 7, 1277-1283. Leahey, D. M. and Friend, J. P., 1971, A model for predicting the depth of the mixing layer over an urban heat island with application to New York City: Journal of Applied Meteorology, v. 10, no. 6, p. 1162-1173. Lettau, H., 1969, Note on aerodynamic roughness parameter estimation on the basis of roughness element description: TESTING AN URBAN CLIMATE SIMULATOR Journal of Applied Meteorology, v. 8, p. 828-882. Lewis, J. E., and Outcalt, S. I., 1976, Verifying an urban surface climate simulation model: paper submitted to be read at the International Geographical Union held in Moscow, July-August, 1976. Lewis, J. E., Outcalt, S. I., and Pease, P. W., 1975, urban surface thermal response associated with land use: un- published paper presented at the Symposium on Me- teorology as Related to Urban and Regional Land-Use Planning, Asheville, North Carolina, November 1975. Marotz, G. A. and Coiner, J. C., 1973, Acquisition and charac- teristics of surface material data for urban climatological studies: Journal of Applied Meteorology, v. 12, no. 6, p. 919-923. McElroy, J. L., 1973, A numerical study of the nocturnal heat island over a medium-sized mid-latitude city (Columbus, Ohio): Boundary Layer Meteorology, v. 3, p. 442-453. Myrup, L. O., 1969, A numerical model of the urban heat island: Journal of Applied Meteorology, v. 8, no. 6, p. 908-918. Myrup, L. O., and Morgan, D. L., 1972, A numerical model of the urban atmosphere: Contributions in Atmospheric Science, v. 1, no. 4, University of California at Davis, 287 p. Outcalt, S. I., 1972a, A reconnaissance experiment in mapping and modeling the effect of land use on urban thermal E17 regimes: Journal of Applied Meteorology, v. 11, no. 8, p. 1869-1373. 1972b, A synthetic analysis of seasonal influences in the effects of land use on the urban thermal regime: Archiv fiir Meteorologie, Geophysik und Bioklimatologie, ser. A, v. 20, p. 253-260. Outcalt, S. I., and Carlson, J., 1975, A coupled soil thermal regime surface energy budget simulator: Conference on Soil Water Problems in Cold Regions, Calgary, Canada, proceedings, 211 p. f Schneider, S. H. and Dickinson, R. E., 1974, Climate model- ing: Review of geophysics and space physics, v. 12, no. 3, p. 447-493. Sellers, W. D., 1965, Physical Climatology: Chicago, Univer- sity of Chicago Press, 272 p. Terjung, W. H., 1970, Urban energy balance climatology: Geographical Review, v. 60, no. 1, p. 81-53. Terjung, W. H. and Collaborators, 1970, The energy-balance climatology of a city-man system, Annals Association of American Geographers, v. 60, no. 3, p. 466-492. Tuller, S. E., 1978, Microclimatic variations in a downtown urban environment: Geografiska Annaler, v. 55A, no. 3, p. 123-136. The Washington Post, Daily weather data, May 10-12, 1972. Yu, T. W. and Wagner, N. K., 1975, A numerical study of the nocturnal urban boundary layer: Boundary Layer Me- teorology, v. 9, no. 2, p. 143-162. * U.S. GOVERNMENT PRINTING OFFICE: 1980 O- 311-344/131 DEPARTMENT OF THE INTERIOR U.S.GEOLOGICAL SURVEY 66°52°30" 106000 METERS r q 50000 METERS |) 48000 METERS 18°15 46000 METERS p> 106000 METERS 66°5230" Base from U.S. Geological Survey 1:20,000; Bayaney, 1957; Monte Guilarte, 1960. HLHON olL3NOYAN HLHON 3nHL APPROXIMATE MEAN DECLINATION, 1984 114000 METERS - 4730" Geologic map of the south part of the Bayaney Quadrangle modified from Nelson and Tobisch 1968, and the north part of the Monte Guilarte Quadrangle from R.D. Krusheusky and A.F. Curet (unpublished data 1978). Ouad- rangle boundary is at lat. 18°15" N. BHT RH d Ter me eni CONTOUR INTERVAL 10 METERS NATIONAL GEODETIC VERTICAL DATUM OF 1929 66°45¢ *! 50000 | METERS 48000 METERS 18915" 46000 METERS 66945¢ INTERIOR-GEOLOGICAL SURVEY, RESTON, VIRGINIA-1984-GB4690 GEOLOGY OF THE TANAMA AND HELECHO PORPHYRY COPPER DEPOSITS AND VICINITY PROFESSIONAL PAPER 1327 PLATE 1 EXPLANATION Lares Limestone (Oligocene) San Sebastian Formation (Oligocene)-Sandstone, siltstone, shale, and basal conglomerate Quartz diorite porphyry (Eocene)-Divided into: Mineralized and hydrothermally altered rocks Breccia containing porphyry and metavolcanic clasts Tf Felsic tuff, breccia, and tuffaceous sandstone (Eocene)-Consists of the Matilde Formation, the Milagros Formation, and the upper part of the Anon Formation mopp mo J/i'l Kg 2,5 Granitic rocks (Cretaceous)-Mainly quartz diorite belonging to the Utuado Batholith Atul g . & 0,8. 08 < ["s flab: ~,| Basaltic lava, breccia, lapilli tuff, and tuffaceous sandstone (Cretaceous)-These s & rocks are thermally metamorphosed to hornblende hornfels over most of the mapped area. As mapped, includes rocks assigned to the Maricao Basalt and the Robles Formation -- Contact-Dashed where inferred --- Fault-Dashed where inferred e Strike and dip of beds -o- Strike of mineralized fractures DH 40 O Drill holes, Copper Creek area, with hole number K Quarry Rectangles around deposits indicate locations of succeeding figures: Laundry Creek, figure 29; Tanama, figures 11, 30, 31, 33, 34; and Helecho, figures 26, 32, and 35. Coordinates in meters, Puerto Rico coordinate system PUERTO RICO