• Avissar, R., , and D. Werth, 2005: Global hydroclimatological teleconnections resulting from tropical deforestation. J. Hydrometeor., 6 , 134145.

    • Search Google Scholar
    • Export Citation
  • Bonan, G. B., 1997: Effects of land use on the climate of the United States. Climatic Change, 37 , 449486.

  • Brovkin, V., , A. Ganopolski, , M. Claussen, , C. Kubatzki, , and V. Petoukhov, 1999: Modelling climate response to historical land cover change. Global Ecol. Biogeogr., 8 , 509517.

    • Search Google Scholar
    • Export Citation
  • Chase, T. N., , R. A. Pielke, , T. G. F. Kittel, , R. R. Nemani, , and S. W. Running, 2000: Simulated impacts of historical land cover changes on global climate in northern winter. Climate Dyn., 16 , 93105.

    • Search Google Scholar
    • Export Citation
  • Chen, F., 2005: Variability in global land surface energy budgets during 1987–1988 simulated by an off-line land surface model. Climate Dyn., 24 , 667684.

    • Search Google Scholar
    • Export Citation
  • Chen, F., , and J. Dudhia, 2001: Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129 , 569585.

    • Search Google Scholar
    • Export Citation
  • Chen, F., and Coauthors, 1996: Modelling of land surface evaporation by four schemes and comparison with FIFE observations. J. Geophys. Res., 101 , 72517268.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , K. J. Westrick, , and C. F. Mass, 1999: Evaluation of MM5 and Eta-10 precipitation forecasts over the Pacific Northwest during the cool season. Wea. Forecasting, 14 , 137154.

    • Search Google Scholar
    • Export Citation
  • Copeland, J. H., , R. A. Pielke, , and T. G. F. Kittel, 1996: Potential climatic impacts of vegetation change: A regional modeling study. J. Geophys. Res., 101 , 74097418.

    • Search Google Scholar
    • Export Citation
  • Ek, M. B., , K. E. Mitchell, , Y. Lin, , E. Rogers, , P. Grunmann, , V. Koren, , G. Gayno, , and J. D. Tarpley, 2003: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational Mesoscale Eta Model. J. Geophys. Res., 108 .8851, doi:10.1029/2002JD003296.

    • Search Google Scholar
    • Export Citation
  • Gandu, A. W., , J. C. P. Cohen, , and J. R. S. de Souza, 2004: Simulation of deforestation in eastern Amazonia using a high-resolution model. Theor. Appl. Climatol., 78 , 123135.

    • Search Google Scholar
    • Export Citation
  • Ganopolski, A., , S. Rahmstorf, , V. Petoukhov, , and M. Claussen, 1998: Simulation of modern and glacial climates with a coupled global model of intermediate complexity. Nature, 391 , 351356.

    • Search Google Scholar
    • Export Citation
  • Giorgi, F., , M. R. Marinucci, , and G. T. Bates, 1993: Development of a second-generation regional climate model (RegCM2). Part I: Boundary layer and radiative transfer processes. Mon. Wea. Rev., 121 , 27942813.

    • Search Google Scholar
    • Export Citation
  • Grell, G. A., , J. Dudhia, , and D. R. Stauffer, 1995: A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Rep. NCAR/TN-398+STR, 122 pp.

  • Gutman, G., , and A. Ignatov, 1998: The derivation of the green vegetation fraction from NOAA/AVHRR for use in numerical weather prediction models. Int. J. Remote Sens., 19 , 15331543.

    • Search Google Scholar
    • Export Citation
  • Hahmann, A. N., , and R. E. Dickinson, 1997: RCCM2–BATS model over tropical South America: Applications to tropical deforestation. J. Climate, 10 , 19441964.

    • Search Google Scholar
    • Export Citation
  • Heck, P., , D. Lüthi, , H. Wernli, , and C. Schär, 2001: Climate impacts of European-scale anthropogenic vegetation changes: A sensitivity study using a regional climate model. J. Geophys. Res., 106 , 78177835.

    • Search Google Scholar
    • Export Citation
  • Henderson-Sellers, A., , R. E. Dickinson, , T. B. Durbidge, , P. J. Kennedy, , K. McGuffie, , and A. J. Pitman, 1993: Tropical deforestation: Modeling local- to regional-scale climate change. J. Geophys. Res., 98 , 72897315.

    • Search Google Scholar
    • Export Citation
  • Hong, S-Y., , and H-L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev., 124 , 23222339.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43 , 170181.

  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437471.

  • Kanae, S., , T. Oki, , and K. Musiake, 2001: Impact of deforestation on regional precipitation over the Indochina Peninsula. J. Hydrometeor., 2 , 5170.

    • Search Google Scholar
    • Export Citation
  • Koren, V., , J. Schaake, , K. Mitchell, , Q-Y. Duan, , F. Chen, , and J. M. Baker, 1999: A parameterization of snowpack and frozen ground intended for NCEP weather and climate models. J. Geophys. Res., 104 , 1956919585.

    • Search Google Scholar
    • Export Citation
  • Lean, J., , and P. R. Rowntree, 1997: Understanding the sensitivity of a GCM simulation of Amazonian deforestation to the specification of vegetation and soil characteristics. J. Climate, 10 , 12161235.

    • Search Google Scholar
    • Export Citation
  • Leathers, D., , M. Malin, , D. Kluver, , G. Henderson, , and T. Bogart, 2007: Hydroclimatic variability across the Susquehanna River basin, USA, since the 17th century. Int. J. Climatol., in press, doi:10.1002/joc.1668.

    • Search Google Scholar
    • Export Citation
  • Narapusetty, B., , and N. Mölders, 2005: Evaluation of snow depth and soil temperature predicted by the Hydro–Thermodynamic Soil–Vegetation Scheme (HTSVS) coupled with the fifth-generation Pennsylvania State University–NCAR Mesoscale Model. J. Appl. Meteor., 44 , 18271843.

    • Search Google Scholar
    • Export Citation
  • Narisma, G. T., , and A. J. Pitman, 2003: The impact of 200 years of land cover change on the Australian near-surface climate. J. Hydrometeor., 4 , 424436.

    • Search Google Scholar
    • Export Citation
  • Nobre, C. A., , P. J. Sellers, , and J. Shukla, 1991: Amazonian deforestation and regional climate change. J. Climate, 4 , 957988.

  • Pan, H-L., , and L. Mahrt, 1987: Interaction between soil hydrology and boundary-layer development. Bound.-Layer Meteor., 38 , 185202.

  • Pan, Z. T., , E. Takle, , M. Segal, , and R. Arritt, 1999: Simulation of potential impacts of man-made land use changes on U.S. summer climate under various synoptic regimes. J. Geophys. Res., 104 , 65156528.

    • Search Google Scholar
    • Export Citation
  • Petoukhov, V., , A. Ganopolski, , V. Brovkin, , M. Claussen, , A. Eliseev, , C. Kubatzki, , and S. Rahmstorf, 2000: CLIMBER-2: A climate system model of intermediate complexity. Part I: Model description and performance for present climate. Climate Dyn., 16 , 117.

    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., , R. L. Walko, , L. T. Steyaert, , P. L. Vidale, , G. E. Liston, , W. A. Lyons, , and T. N. Chase, 1999: The influence of anthropogenic landscape changes on weather in south Florida. Mon. Wea. Rev., 127 , 16631673.

    • Search Google Scholar
    • Export Citation
  • Schultz, P., 1995: An explicit cloud physics parameterization for operational numerical weather prediction. Mon. Wea. Rev., 123 , 33313343.

    • Search Google Scholar
    • Export Citation
  • Semazzi, F. H. M., , and Y. Song, 2001: A GCM study of climate change induced by deforestation in Africa. Climate Res., 17 , 169182.

  • Su, H., , S. S. Chen, , and C. Bretherton, 1999: Three-dimensional week-long simulations of TOGA COARE convective systems using the MM5 mesoscale model. J. Atmos. Sci., 56 , 23262344.

    • Search Google Scholar
    • Export Citation
  • Werth, D., , and R. Avissar, 2002: The local and global effects of Amazon deforestation. J. Geophys. Res., 107 .8087, doi:10.1029/2001JD000717.

    • Search Google Scholar
    • Export Citation
  • Werth, D., , and R. Avissar, 2005: The local and global effects of African deforestation. Geophys. Res. Lett., 32 .L12704, doi:10.1029/2005GL022969.

    • Search Google Scholar
    • Export Citation
  • Xiu, A. A., , and J. E. Pleim, 2001: Development of a land surface model. Part I: Application in a mesoscale meteorological model. J. Appl. Meteor., 40 , 192209.

    • Search Google Scholar
    • Export Citation
  • Zheng, X., , and E. A. B. Eltahir, 1998: The role of vegetation in the dynamics of West African monsoons. J. Climate, 11 , 20782096.

  • View in gallery

    A reference map showing the location of the two domains in the MM5 simulation and the area in which the land cover forcing was applied.

  • View in gallery

    The difference in (left) August and (right) February ensemble means between the Clear-cut and Prehistoric scenarios (taken as Clear-cut minus Prehistoric) in (a),(b) the latent heat flux (W m−2), (c), (d) net surface shortwave radiation (W m−2), and (e), (f) 2-m temperature (°C).

  • View in gallery

    (a) The difference in August ensemble means between the Clear-cut and Prehistoric scenarios in the sensible heat flux (W m−2), (b) the ratio of the August ensemble means (taken as Clear-cut divided by Prehistoric) in the Bowen ratio (unitless), and (c) the difference between the August ensemble means in the height of the atmospheric boundary layer (m). The line contours in (b) show the August ensemble-mean Bowen ratio in the Prehistoric scenario; contours are drawn every 0.05 from 0.20 to 0.35, with labels at 0.25 and 0.30.

  • View in gallery

    The difference in ensemble-mean, area-mean (over the land cover–forcing area) quantities between the August Clear-cut and Prehistoric ensembles, separated into 3-h increments, for (a) T2m, ground temperature, ground heat flux, latent heat flux, sensible heat flux, and net surface shortwave radiation. (b) Difference in temperature with height. Net shortwave radiation is defined as positive downward; all other fluxes are defined as positive from the surface into the atmosphere. The height axis in (b) has been exaggerated to better show changes near the surface; tickmarks correspond to model levels.

  • View in gallery

    Similar to Fig. 2, but only for August and for mixing ratio (colors; g kg−1) in (a) a zonal cross section taken at 41.5°N and (b) a meridional cross section taken at 77°W. Black line contours show the difference in (a) zonal and (b) meridional winds (m s−1), with westerly and southerly winds defined as positive; purple line contours show the difference in vertical velocity (cm s−1), with upward motion defined as positive. Line contours are drawn every 0.1 m s−1 or 0.1 cm s−1; the zero contour is not shown; negative contours are dashed.

  • View in gallery

    As in Fig. 2, but only for August and in (a) 2-m mixing ratio (g kg−1), (b) total precipitation (cm), (c) σ = 0.91–0.97 vertically integrated moisture divergence (10−7 s−1), and (d) vegetative-canopy moisture (cm). The line contours in (c) show the difference in the σ = 0.91–0.97 vertically integrated wind.

  • View in gallery

    Time series of the difference between the Clear-cut and Prehistoric August ensemble-mean (a) volumetric water content (m3 m−3) and (b) soil temperature (°C) for each soil level in the Noah LSM. In both (a) and (b), the legend labels give the midpoints of each soil layer.

  • View in gallery

    Estimates of the intraensemble variability of the difference between the Clear-cut and Prehistoric ensembles [as in Eq. (1); shading] and the corresponding coefficient of variation [as in Eq. (2); black contours] for (a) 2-m temperatures (°C), (b) latent heat flux (W m−2), and (c) ABL height (m). Contour levels for the coefficient of variation go from 0 to 0.60 by 0.10, with an additional contour at 0.15. The coefficient of variation was calculated only where the std dev exceeded the white contour level.

  • View in gallery

    Mean anomalies from three National Climatic Data Center Climate Reference Network stations within the inner MM5 domain in February temperatures (filled bar), August temperatures (unfilled bar), February precipitation (diagonal upslope), and August precipitation (diagonal downslope) for 2000–04. Anomalies were calculated from the 1971–2000 climatological means.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 20 20 4
PDF Downloads 9 9 3

Mesoscale Simulations of the Land Surface Effects of Historical Logging in a Moist Continental Climate Regime

View More View Less
  • 1 Department of Geography and Center for Climatic Research, University of Delaware, Newark, Delaware
© Get Permissions
Full access

Abstract

An enhanced knowledge of the feedbacks from land surface changes on regional climates is of great importance in the attribution of climate change. To explore the effects of deforestation on a midlatitude climate regime, two sets of two five-member ensembles of 28-day simulations were conducted using the fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5) coupled to the “Noah” land surface model. The four ensembles represented conditions in summer (August) and winter (February) across the northern mid-Atlantic United States before and after extensive late-nineteenth-century logging of hardwood forests in central and northern Pennsylvania. Prelogging ensembles prescribed a vegetative cover of an evergreen needleleaf forest; postlogging ensembles prescribed sparse vegetation and bare soil to simulate clear-cut deforestation. The results of the MM5 experiments showed a decided seasonality in the response of the land surface–atmosphere system to deforestation, with much stronger effects arising in summer. In August, deforestation caused a repartitioning of the surface energy budget, beginning with a decrease in the latent heat flux of more than 60 W m−2 across the land cover–forcing area, representing almost one-half of the latent heat flux under prelogging land cover. Concomitant with this decrease in evapotranspiration, mean 2-m air temperatures warmed by at least 1.5°C. Increases in sensible heat flux led to a 150-m mean increase in the height of the atmospheric boundary layer over the deforested area. Low-level atmospheric mixing ratios and total precipitation decreased under clear-cut conditions. Mean soil moisture increased in all model levels to 150 cm because of a decrease in vegetative uptake of water, except at the 5-cm level at which such decreases were effectively balanced by greater soil evaporation and less precipitation. A strong diurnal variation in the response to deforestation of ground and lower-atmosphere temperatures and heat fluxes was also identified for the summer season. The February simulations showed the effects of deforestation during low-insolation months to be small and variable. The strong response of the summer land surface–atmosphere system to deforestation shown here suggests that land cover changes can appreciably affect regional climates. Thus, the role of human-induced and naturally occurring land cover variability should not be ignored in the attribution of climate change.

* Current affiliation: Walker Institute for Climate System Research and Department of Meteorology, University of Reading, Reading, United Kingdom

Corresponding author address: N. P. Klingaman, Department of Meteorology, University of Reading, Earley Gate, P.O. Box 243, Reading, Berkshire, RG6 6BB, United Kingdom. Email: n.p.klingaman@rdg.ac.uk

Abstract

An enhanced knowledge of the feedbacks from land surface changes on regional climates is of great importance in the attribution of climate change. To explore the effects of deforestation on a midlatitude climate regime, two sets of two five-member ensembles of 28-day simulations were conducted using the fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5) coupled to the “Noah” land surface model. The four ensembles represented conditions in summer (August) and winter (February) across the northern mid-Atlantic United States before and after extensive late-nineteenth-century logging of hardwood forests in central and northern Pennsylvania. Prelogging ensembles prescribed a vegetative cover of an evergreen needleleaf forest; postlogging ensembles prescribed sparse vegetation and bare soil to simulate clear-cut deforestation. The results of the MM5 experiments showed a decided seasonality in the response of the land surface–atmosphere system to deforestation, with much stronger effects arising in summer. In August, deforestation caused a repartitioning of the surface energy budget, beginning with a decrease in the latent heat flux of more than 60 W m−2 across the land cover–forcing area, representing almost one-half of the latent heat flux under prelogging land cover. Concomitant with this decrease in evapotranspiration, mean 2-m air temperatures warmed by at least 1.5°C. Increases in sensible heat flux led to a 150-m mean increase in the height of the atmospheric boundary layer over the deforested area. Low-level atmospheric mixing ratios and total precipitation decreased under clear-cut conditions. Mean soil moisture increased in all model levels to 150 cm because of a decrease in vegetative uptake of water, except at the 5-cm level at which such decreases were effectively balanced by greater soil evaporation and less precipitation. A strong diurnal variation in the response to deforestation of ground and lower-atmosphere temperatures and heat fluxes was also identified for the summer season. The February simulations showed the effects of deforestation during low-insolation months to be small and variable. The strong response of the summer land surface–atmosphere system to deforestation shown here suggests that land cover changes can appreciably affect regional climates. Thus, the role of human-induced and naturally occurring land cover variability should not be ignored in the attribution of climate change.

* Current affiliation: Walker Institute for Climate System Research and Department of Meteorology, University of Reading, Reading, United Kingdom

Corresponding author address: N. P. Klingaman, Department of Meteorology, University of Reading, Earley Gate, P.O. Box 243, Reading, Berkshire, RG6 6BB, United Kingdom. Email: n.p.klingaman@rdg.ac.uk

1. Introduction

Located in the mid-Atlantic region of the United States, the Susquehanna River basin (SRB) encompasses more than 70 000 km2 in New York, Pennsylvania, and Maryland. The Susquehanna River is the major tributary to the Chesapeake Bay, the largest estuary system in the United States. Hence, changes in the flow, sediment load, and water quality of the Susquehanna River play a crucial role in the health of the Chesapeake Bay. Intense clear-cut logging in central Pennsylvania and southern New York during the mid-to-late nineteenth century caused a dramatic change in land cover and vegetation type in the SRB as evergreen forests were reduced to sparse shrubs and bare soil. Today, 60% of the SRB has become secondary-growth deciduous forest, and the remaining area contains agricultural and urban environments that are home to more than four million inhabitants.

A better understanding of the feedbacks from land surface changes on local climate is of great importance in the attribution of climate change. Because of a lack of existing research and a scarcity of local climate records prior to 1900, few estimates exist of the impact of the acute deforestation in the SRB on its climate and hydrology. Leathers et al. (2007) investigated the hydroclimatological history of the SRB and explored the role of land cover changes on the regional climate. They found that the SRB did not respond strongly to any of the major hemispheric-scale forcing mechanisms thought to affect the climate of eastern North America. Instead, they hypothesized that regional forcing mechanisms, including land surface changes, were likely responsible for much of the SRB climate variability found in the observational record.

Although the impact of historical land cover change on the SRB climate system is unknown, existing research—focused primarily on the tropics—has indicated two changes frequently associated with deforestation: an increase in the surface albedo and a decrease in the surface roughness length (Lean and Rowntree 1997; Kanae et al. 2001). A higher surface albedo implies greater surface reflectance and hence a decrease in surface net radiation. Several studies have noted that this decrease in surface energy has reduced surface evapotranspiration (e.g., Hahmann and Dickinson 1997; Lean and Rowntree 1997). One would ordinarily expect a decrease in evapotranspiration to result in a local decrease in precipitation, but this inference is complicated by the deforestation-induced decrease in surface roughness length, which may increase mass and moisture convergence by strengthening the low-level winds (Kanae et al. 2001). The sign of the resultant feedback on moisture availability depends on whether evapotranspiration or moisture convergence dominated the response to deforestation. If the decrease in evapotranspiration reduced local moisture availability by more than the reduction in the surface roughness length increased moisture availability (via moisture convergence), then the subsequent reduction in precipitation would, in turn, cause a decrease in evapotranspiration. If allowed to continue unabated, such a feedback could produce local-scale drought conditions. On the other hand, if the increase in moisture convergence dominated the reduction in evapotranspiration, then the resultant negative feedback on moisture availability could mitigate the effects of local deforestation. Although this issue has been examined in the tropics (e.g., Henderson-Sellers et al. 1993; Kanae et al. 2001) and subtropics (e.g., Zheng and Eltahir 1998; Pielke et al. 1999), the question is far from resolved, particularly in the midlatitudes.

Studies using general circulation models (GCMs) have shown that the effects of deforestation on midlatitude climate vary with land cover type and regional climatology. Bonan (1997) used the National Center for Atmospheric Research (NCAR) Community Climate Model, version 2, (CCM2) coupled to the NCAR land surface model (LSM) to investigate deforestation in the continental United States. When modern vegetation replaced prehistoric forests, temperatures cooled across the country by as much as 2°C. Brovkin et al. (1999) simulated the global deforestation of the last millennium in the Climate and Biosphere Group model, version 2, (CLIMBER-2; Ganopolski et al. 1998; Petoukhov et al. 2000) and found that mean temperatures decreased by up to 0.5°C in the Northern Hemisphere midlatitudes and 0.35°C globally. This signal was most pronounced in regions where cropland replaced forests. Chase et al. (2000) found a slight increase in global-mean temperatures resulting from global deforestation and land use change. North of 30°N, the average land surface temperature increased by 0.29°C when prehistoric forests were replaced with modern land cover types.

Other GCM studies have assessed the effects of deforestation in tropical regions. Semazzi and Song (2001) ran the Community Climate Model, version 3, (CCM3) coupled to the NCAR LSM and found that replacing equatorial African tropical rain forests with savanna grasslands led to a 2.5–5.0°C warming during July–September. Werth and Avissar (2002) employed the National Aeronautics and Space Administration (NASA) Goddard Institute for Space Studies GCM and determined that replacing Amazonian rain forest with mixed shrubs and grasslands produced a basinwide 300 mm yr−1 decrease in precipitation. These results are consistent with those of Nobre et al. (1991), who substituted degraded pastureland for Amazonian rain forest. Werth and Avissar (2005) and Avissar and Werth (2005) also found significantly drier conditions across Africa and Southeast Asia as a result of deforestation.

Several mesoscale-model studies have evaluated the impact of complete deforestation on the regional climates of Amazonia (in particular, in the eastern rain forests), Australia, the United States, and the Mediterranean Sea. Gandu et al. (2004) found a 1°C warming during the eastern Amazon summer dry season when pastureland replaced rain forest in the Colorado State University Regional Atmospheric Modeling System (RAMS). Narisma and Pitman (2003) used the fifth-generation Pennsylvania State University–NCAR Mesoscale Model (MM5) and indicated that removing the natural (1788) vegetation in Australia in favor of modern (1988) vegetation caused a 1°C (0.5°C) warming in January (July). The warming was caused by a decrease in the surface latent heat flux that itself was caused by removing the vegetated canopy. Likewise, Copeland et al. (1996) employed RAMS across the United States and demonstrated that removing the prehistoric land cover from the eastern United States likely caused a warmer (by 0.22°C), wetter summer associated with a slight decrease (increase) in the latent (sensible) heat flux to the atmosphere. Pan et al. (1999) showed that substituting crops for native grasses and deciduous and needleleaf forests for evergreen in the United States in the second-generation Regional Climate Model (RegCM2; Giorgi et al. 1993) created a warmer (cooler) and drier climate in the western (central) United States. Heck et al. (2001) suggested that the presettlement land cover in the Mediterranean was associated with cooler, wetter springs and warmer, drier summers relative to the present day.

Land surface models capable of being coupled to mesoscale models have progressed rapidly over the last several years, simulating surface and subsurface energy and moisture fluxes in response to atmospheric forcings for a wide range of land cover types at high spatial resolutions. Land surface heterogeneities, including soil texture, moisture, and vegetation type differences, can have large effects on aspects of atmospheric circulation, including cumulus cloud formation and distribution (Xiu and Pleim 2001). The National Centers for Environmental Prediction (NCEP)–Oregon State University (OSU)–Air Force–Hydrologic Research Laboratory (Noah) LSM presents some important enhancements, including explicit simulation of soil moisture in four soil layers, three pathways for evapotranspiration (i.e., vegetative transpiration, soil evaporation, and evaporation of canopy-intercepted water), and new parameterizations for stomatal conductance and surface moisture (Chen et al. 1996; Koren et al. 1999; Chen and Dudhia 2001; Xiu and Pleim 2001; Ek et al. 2003). These new parameterizations allow for a more accurate representation of the interactions between the land surface and atmospheric boundary layer (ABL; Ek et al. 2003).

There is little agreement in the literature on the effect of large-scale deforestation on a midlatitude regional climate, even when the same region (e.g., the central United States) has been examined. Few studies have employed sophisticated LSMs that can achieve a reasonable degree of accuracy in moisture and heat fluxes at high spatial resolution. As a consequence, little is known about the effect of deforestation on quantities other than near-surface temperature and moisture. No strong conclusions have been reached on the seasonal variation of the land–atmosphere system’s response to deforestation; even less attention has been paid to variability on diurnal time scales. To address these deficiencies in previous research, this study makes use of the Noah LSM coupled to MM5 (version 3.6.2). The immediate goal of this research is to explore the effects of historical deforestation on the hydroclimatology of the SRB using the MM5–Noah coupled system. The design of our experiments will enable us to examine both seasonal and diurnal variations in the response to deforestation; ensembles of simulations will increase the robustness of our results. Simulations will be conducted with land surface conditions prescribed to approximate those that existed before and immediately after the deforestation of the mid-nineteenth century. Today, the region is a mixture of deciduous forest, farmland, and urban land cover types; our simulations will use idealized mid-nineteenth-century pre- and postlogging land cover as a case study for accelerated, extensive, anthropogenic deforestation. The results of this case study are not intended to be directly relevant to the current climate of central Pennsylvania because the region is no longer deforested. Rather, we expect that the results of our simulations will apply to similar deforestation scenarios in other midlatitude regions. In addition, the results of this study will help us to understand the long-term hydroclimatic variability of the SRB.

2. Experiment design

a. The coupled MM5–Noah system

MM5 is a limited-area, nonhydrostatic atmospheric model that has been widely used for mesoscale forecasting (e.g., Colle et al. 1999; Narapusetty and Mölders 2005) and hindcasting (e.g., Su et al. 1999). The model can horizontally interpolate land surface and atmospheric input data to operate on a variable horizontal grid as fine as the order of a kilometer. MM5 experiments often augment large-scale atmospheric initial and lateral boundary conditions with surface and upper-air observations to give greater accuracy at mesoscale resolutions; these additional observations are integrated using a Cressman-type analysis scheme. MM5 provides for “nesting” of model domains so that a domain with high spatial resolution may be inserted into a domain with coarse spatial resolution to provide greater detail and to resolve small-scale physical processes. MM5 uses a terrain-following, sigma-coordinate system in the vertical direction. [Further details on MM5 can be found in Grell et al. (1995)].

For our experiments, we have coupled the Noah LSM (Chen and Dudhia 2001; Ek et al. 2003; Chen 2005) to MM5. The Noah LSM is based on the OSU LSM (Pan and Mahrt 1987). The Noah model operates on four vertical soil layers spanning the depths 0–10, 10–40, 40–100, and 100–200 cm. Within each soil layer, the model predicts soil temperature and volumetric water content. The model parameterizes a root depth that varies with vegetation type and which the model uses to calculate the amount of water the vegetation can withdraw from each soil layer. Vegetative transpiration is calculated from the leaf area index (LAI), a monthly climatological green-vegetation fraction, and a minimum stomatal resistance Rsmin. The green-vegetation fraction is defined to be the fraction of each grid box within which midday downward solar radiation is intercepted by the photosynthetically active green canopy (Chen and Dudhia 2001). This fraction is ordinarily obtained from a 5-yr climatological dataset of Advanced Very High Resolution Radiometer measurements at 0.15° × 0.15° resolution (Gutman and Ignatov 1998); in our experiment we prescribed values for the green-vegetation fraction (see section 2b). Because the normalized difference vegetation index (NDVI) method used to obtain the green-vegetation fraction does not contain sufficient degrees of freedom to calculate the fraction and the LAI simultaneously (Gutman and Ignatov 1998), the Noah LSM holds the LAI constant at 4.0 for all land cover types; the root depth and Rsmin vary with land cover type.

b. Ensemble design and model setup

To examine the impact of deforestation, we conducted two sets of two five-member MM5 ensembles, with separate five-member ensembles for winter (February) and summer (August) under pre- and postlogging land cover. Each set of two ensembles prescribed a different land cover type inside a rectangular box that approximates the area in which intensive logging took place in the nineteenth century. This area (hereinafter, the “land cover–forcing region”) encompasses central and northern Pennsylvania and a small portion of southern New York (Fig. 1). One set of ensembles prescribed an evergreen needleleaf forest similar to the prelogging vegetation, and the other one prescribed sparse vegetation to represent the land surface conditions after the region had been clear-cut. The text will refer to these sets of simulations as the Prehistoric and Clear-cut scenarios, respectively. The Prehistoric (Clear-cut) land cover has a low (high) albedo, a high (low) roughness length, moderate (low) soil moisture availability, and a low (high) thermal inertia (Table 1). The present-day dominant land cover—a broadleaf deciduous forest—has moderate values for all parameters and is included in Table 1 for reference. Because the current land cover is heterogeneous in type, all values in Table 1 are area averages over the land cover–forcing region. All values were taken from the 25-category U.S. Geological Survey (USGS) dataset available from NCAR with MM5 and were prescribed for the entire land cover–forcing region. This 4-km-resolution version of the USGS dataset was provided to the Noah LSM through the MM5 “TERRAIN” preprocessing program. We elected to use this dataset over the default USGS vegetation parameters included with the LSM because the former dataset contains separate values for the winter and summer seasons, whereas the default values in the LSM are annual means. Given that investigating the seasonal variation of the land surface response to deforestation is one of the key objectives of this study, it is critical to use a land cover dataset that allows the land surface parameters to vary between the two seasons considered.

Table 1 gives the monthly green-vegetation fraction for each scenario. The value for the present land cover is an area average of the Gutman and Ignatov (1998) dataset over the land cover–forcing region. Values for the Prehistoric and Clear-cut scenarios were specified within TERRAIN and were taken from the Gutman and Ignatov (1998) dataset at locations of the same land cover type and similar latitude to the land cover–forcing region. Specifying the green-vegetation fraction introduces a source of uncertainty into the experiment, but this was judged to be preferable to leaving the fraction at its present-day value. Outside of the land cover–forcing region, the model used the Gutman and Ignatov (1998) climatological value. Table 1 also gives the annual-mean values of root depth and Rsmin, taken from the Noah LSM’s lookup table. Note that the value of Rsmin is undefined for the Clear-cut scenario because the LSM does not consider the “barren or sparsely vegetated” land cover type to have a vegetated canopy.

Within each set of two five-member ensembles, one ensemble used lateral boundary conditions from August and the other used lateral boundary conditions from February. Boundary conditions were obtained for 2000–04; each year forced one member within each ensemble. Each integration ran for 28 days from the first of the month. Simulations were conducted under summer and winter conditions to investigate the possibility that the seasonal cycle may play a role in the effects of logging on the regional climate. NCEP–NCAR 40-Year Reanalysis (Kalnay et al. 1996) data provided the large-scale atmospheric forcing. The reanalysis data were combined with NCEP upper-air and surface observations (e.g., radiosondes, aircraft, buoys) in the MM5 preprocessing programs. All simulations used a two-domain nest covering the mid-Atlantic and New England regions of the United States (Fig. 1). The outer domain (hereinafter domain 1) had a spatial resolution of 45 km, and the inner domain (hereinafter domain 2) had a spatial resolution of 15 km to provide finer detail in the area immediately surrounding the land cover forcing. Both domains were run with nonhydrostatic dynamics. Domain 1 ran with a time step of 2 min and was forced at the boundaries with the combination of reanalysis and observations; domain 2 took its boundary conditions from the output from domain 1. All modeled quantities were output every 3 h to allow for examinations of diurnal variability in the response to deforestation (section 3b). All simulations employed the Noah LSM, the Schultz moisture scheme (Schultz 1995), the Hong–Pan planetary boundary layer scheme (Hong and Pan 1996), and the Kain–Fritsch convective parameterization (Kain 2004), which includes both shallow and deep convection. These schemes and this parameterization were chosen for their inclusion of critical physical processes, their accuracy at fine spatial resolutions, and their computational efficiency.

3. Response of the land–atmosphere system to deforestation

Where differences between scenarios are shown as vertical cross sections, we have retained the native terrain-following sigma-coordinate system of MM5 (denoted by σ) to show differences at constant height above the orography (rather than converting to levels of constant pressure) because of variations in orography across our land cover–forcing region. At the surface, σ has a value of 1; it has a value of 0 at the top of the model (100 hPa). The region has a mean elevation of 442 m, which corresponds to a mean reference surface pressure in MM5 of approximately 945 hPa. The highest values of reference surface pressure are in the south (962 hPa), and the lowest values are in the north and west (926 hPa). Where σ values are used in the text, we will provide approximate pressure levels based on the mean orography of the region.

a. Mean response of atmospheric temperatures and fluxes

Means were calculated across each five-member ensemble: August and February for the Clear-cut and Prehistoric land cover scenarios. The difference between the Clear-cut and Prehistoric ensemble means quantifies the effect of regional deforestation in this midlatitude climate regime.

Our results show a marked seasonality in the land surface response to replacing the prehistoric evergreen needleleaf forest with sparse vegetation. The dominant effect of the land cover change was to reduce the summertime latent heat flux from the land surface into the atmosphere by more than 60 W m−2 (Fig. 2a; latent heat flux is defined as positive into the atmosphere). This decrease in evapotranspiration in the Clear-cut scenario followed from replacing readily transpiring evergreen trees with bare soil and sparse vegetation. Table 2 shows ensemble-mean, time-mean values, area-averaged over the land cover–forcing region for the Prehistoric scenario, the Clear-cut scenario, and the difference between them (taken as Clear-cut minus Prehistoric) for key quantities examined in this study. The decrease in evapotranspiration in the August Clear-cut scenario represents nearly one-half of the latent heat flux in the August Prehistoric simulation (Table 2). This change induced a large imbalance in the summertime surface energy budget. Far lower insolation in winter resulted in a lower latent heat flux during February in the Prehistoric scenario (Table 2), which led to a much smaller difference in evapotranspiration between the Clear-cut and Prehistoric scenarios in that season (Fig. 2b). Simply put, removing the prehistoric forest had little effect on wintertime evapotranspiration because there was, climatologically, very little wintertime evapotranspiration to affect.

In both seasons, removing the vegetated canopy caused a decrease in net surface shortwave radiation (calculated as incident minus reflected shortwave radiation; Figs. 2c,d). This decrease was greater in August than in February: in the summer simulations net surface shortwave radiation decreased by nearly 30 W m−2 in some areas of the land cover–forcing region. Removing the vegetated canopy increased both components of the net surface shortwave in August and February, with the increase in reflectance greater than the increase in insolation (Table 2). This demonstrates that the increase in the surface albedo (approximately 0.15; Table 1) that results from replacing trees with bare soil dominated the increase in penetration of solar radiation from removing the canopy. In August, the decrease in net surface shortwave radiation fueled a small portion of the decrease in the latent heat flux; most of the decrease was due to a reduction in evapotranspiration. In February, most of the decline in surface shortwave was redistributed evenly between decreases in the latent and sensible heat fluxes (Table 2).

Concomitant with the decrease in evapotranspiration, mean August 2-m temperatures (denoted as T2m) warmed in excess of 1.5°C across the entire land cover–forcing region in the Clear-cut scenario (Fig. 2e). Time-mean warming was limited in the vertical: differences greater than 0.25°C between the August Prehistoric and Clear-cut ensemble means reached only to the σ = 0.9 level (approximately 860 hPa; not shown). There was, however, some diurnal variation in warming with height, which will be discussed in section 3b. No such surface warming was observed for February (Fig. 2f). One February Clear-cut ensemble member showed cooler T2m of approximately 1°C, resulting in a decrease in the ensemble mean (Table 2), whereas the other four members exhibited near-zero change. The August Clear-cut scenario also showed an enhanced flux of sensible heat into the atmosphere (Fig. 3a; sensible heat flux is defined as positive into the atmosphere). Driven by the increase in the sensible heat flux and the decrease in evapotranspiration, the Bowen ratio in the August Clear-cut scenario increased more than threefold from the Prehistoric-scenario value of approximately 0.30 (Fig. 3b). A small decrease (nearly 6 W m−2 in the area mean; Table 2) in the mean sensible heat flux occurred in the February Clear-cut scenario, which reversed the sign of the sensible heat flux from the Prehistoric ensemble. As with T2m, this decrease in the sensible heat flux was driven by one ensemble member; others showed almost no change.

The top of the ABL typically determines the maximum height that turbulent eddies can reach. The vertical penetration of these eddies is critical for the generation of convective processes; a deeper ABL is commonly associated with greater convective instability. In the August Clear-cut scenario, the ABL deepened by more than 100 m across most of the land cover–forcing region (Fig. 3c); however, no such change occurred in February. The August deepening is almost certainly the result of increased low-level buoyancy from the warmer T2m and the increase in the flux of sensible heat into the atmosphere in the Clear-cut scenario. From the depth of the ABL alone, one would expect an increase in convection and potentially precipitation over the forcing region. The difference in precipitation between the Clear-cut and Prehistoric ensembles will be discussed in section 3c.

As net surface shortwave radiation decreased in the August Clear-cut scenario, we concluded that the increases in T2m and the sensible heat flux in that scenario were a direct result of the collocated and coincident decrease in evapotranspiration. The reduced evapotranspiration led to a repartitioning of the surface energy budget: a portion of the excess energy was transmitted into the atmosphere as sensible heat, warming the atmosphere to the σ = 0.9 level and raising the ABL height, while the remaining energy warmed the surface and subsurface. The partitioning of the excess heat energy between sensible and ground heat fluxes will be explored further in section 3b. These effects were far more pronounced in August than February because of greater summertime insolation, which provides a climatologically stronger latent heat flux.

b. Diurnal variation of the response of temperatures and fluxes

To investigate the diurnal variation of surface and near-surface temperature and fluxes, we computed the area-mean difference between the ensembles at each 3-h MM5 output time step [beginning at midnight UTC, or 2000 local time (LT)], then took the mean of the differences over the month for each three-hourly period over the day. In Fig. 4a then, the bars labeled “20:00” give the mean difference over the month at 2000 LT.

During the morning and afternoon, the latent heat flux is greatly reduced; the decrease in evapotranspiration exceeds 200 W m−2 in the early afternoon (1400 LT; Fig. 4a). Only about 40 W m−2 of this reduction can be accounted for by a decrease in shortwave radiation at the surface, which is defined as positive downward to the surface in Fig. 4a. Of the anomalous additional energy available at the surface in the Clear-cut scenario, approximately 40% entered the lower atmosphere as sensible heat flux; the increase in sensible heat flux reached more than 60 W m−2 at 1400 LT. An additional 50% of the anomalous surface energy was transmitted to the soil column by the ground heat flux; the increase exceeded 70 W m−2 at 1100 and 1400 LT. In the Prehistoric ensemble, the ensemble-mean, area-mean ground heat flux at 1400 LT—the peak value during the day—was 25 W m−2. In the Clear-cut ensemble this quantity increased by more than 300% to 101 W m−2; increases of similar proportion occurred through the morning and afternoon. The difference in ground temperature between the Clear-cut and Prehistoric land cover scenarios was nearly 4°C in the afternoon, when the decrease in latent heat was at its peak. (An increase in outgoing longwave radiation from the warmer surface balanced the remaining 10% of the anomalous energy from the reduction in evapotranspiration.) The soil anomalously warmed at 0–10 cm during the afternoon as well, as heat from the ground was conducted down into the soil column; area-mean afternoon 0–10-cm soil temperatures were more than 6°C warmer in the Clear-cut scenario and increased almost 3°C in the daily mean (Table 2). Changes in soil temperature will be discussed in greater detail in section 3c.

Daytime T2m, however, did not warm to the same extent as the ground temperatures in the Clear-cut scenario; at 1400 LT, values of T2m were less than 0.5°C warmer in the Clear-cut scenario than in the Prehistoric scenario. This led to strong anomalous vertical thermal gradients in the Clear-cut scenario of more than 1.5°C m−1 (Fig. 4b). We hypothesize that this was caused by daytime mixing of the near-surface with the lower troposphere, which warmed by as much as 0.5°C during the morning and early afternoon. In the Clear-cut scenario, additional sensible heat flux and the lack of a vegetated canopy would facilitate increased exchange of heat with the overlying atmosphere, which could then be mixed up through the boundary layer. Given the increase in lower-tropospheric temperatures in the Clear-cut scenario to σ = 0.87 (approximately 820 hPa) through the morning and early afternoon, this appears to be the case.

At night, the ground heat flux became positive as most of the energy conducted into the upper soil layers during the day was returned to the surface when the surface cooled in the evening (Fig. 4a). This heat flux warmed the ground and the lowest few meters of the atmosphere in the Clear-cut scenario overnight. The anomalous near-surface vertical thermal gradients that were present in the morning and afternoon disappeared during nighttime (Fig. 4b), which is generally characterized by less mixing between the surface and the ABL; at night the strongest anomalous thermal gradients were between 2 m and the lowest model level (σ = 0.995). There was no meaningful change in lower-tropospheric nighttime temperatures in the Clear-cut scenario, and the anomalous sensible heat flux was much smaller during the night.

Although these are area-mean changes, section 3a showed that the differences in most quantities considered were homogeneous across the land cover–forcing region. Spatial means are therefore not expected to have affected these results.

c. Response of precipitation and subsurface temperature and moisture

Lower-troposphere mixing ratios to σ = 0.8 (about 775 hPa) decreased by as much as 0.8 g kg−1 in the August Clear-cut scenario, indicating that at least the lowest kilometer of the atmosphere has been noticeably dried (Fig. 5). Removing the forest canopy dried the near-surface atmosphere by about 5% at the 2-m level (Table 2 and Fig. 6a). The lower-tropospheric moisture decrease was strongest in the center of the forcing region, although decreases in surface evapotranspiration were fairly homogeneous (Fig. 2a). The increase in near-surface temperatures promoted lower-tropospheric buoyancy, causing additional upward motion over the land cover–forcing region in the Clear-cut scenario, and some compensating descent at the outer zonal and meridional boundaries of the region (Fig. 5). The climatological westerlies increased across the region (Fig. 5a) because of the decreased roughness length of the bare soil and sparse vegetation in the Clear-cut scenario (Table 1). Removing the forests also increased the climatological convergent meridional wind pattern, although this increase was smaller than the increase in the zonal winds (Fig. 5b); from a climate perspective, the westerlies dominate any meridional flow in this region.

Despite an increase in ABL height and additional upward vertical motion, the August Clear-cut scenarios showed a marked decrease in total precipitation of 1.5–3.0 cm across the majority of the land cover–forcing region (Fig. 6b). When area-averaged over the forcing region, precipitation was reduced by about 12% in the Clear-cut ensemble over the Prehistoric ensemble (Table 2); in regions to the north and west of the forcing region this reduction exceeded 30%. Many of the largest reductions in precipitation occurred on the edges of the land cover–forcing region, where Fig. 5 showed anomalous descent in the lower troposphere. Upstream areas to the west and south of the forcing region also experienced anomalous low-level vertically integrated moisture divergence (σ = 0.91–0.97, representing three model levels between 870 and 920 hPa; Fig. 6c). In these areas, the abrupt decrease in roughness length at the land cover–forcing boundary in the Clear-cut scenario caused anomalous divergence in the lower portions of the atmosphere through reduced frictional drag. This anomalous divergence combined with the reduction in low-level moisture and anomalous subsidence (Fig. 5) to cause a sharp reduction in precipitation over the month. In a similar way, downstream areas on the eastern and northern boundaries experienced anomalous moisture convergence, which acted to counter the anomalous subsidence and low-level drying to produce precipitation at or exceeding the amounts in the Prehistoric ensemble. In response to the question of energy–moisture feedback processes posed in section 1, however, the decrease in evapotranspiration was the dominant factor affecting precipitation across the land cover–forcing region, not increases in moisture convergence due to a reduced roughness length. This implies that deforestation can initiate a positive feedback process between decreases in surface evapotranspiration and moisture in which a decrease in evapotranspiration leads to a reduction in low-level moisture availability and precipitation, which in turn causes surface drying and further decreases in evapotranspiration. Our simulations were not long enough to allow examination of this positive feedback, however, and therefore its existence remains a hypothesis.

The Noah LSM hydrological scheme accounts for two processes essential for evaluating the response of the soil column to deforestation: the distribution of through-falling precipitation (i.e., precipitation not intercepted by the canopy) to the soil, and the uptake of water from each soil layer by vegetation (section 2a; Chen and Dudhia 2001; Ek et al. 2003). The latter of these depends upon the parameterized root depth; the depth is one soil layer (10 cm) for barren and sparsely vegetated ground (Clear-cut) and four layers (200 cm) for evergreen needleleaf vegetation (Prehistoric; Table 1). Therefore, root penetration was greatly reduced in the Clear-cut scenario relative to the Prehistoric, resulting in decreases in transpiration and—in particular, at depth—vegetative uptake of soil water.

Time-mean soil volumetric water content increased in all layers except the 0–10-cm layer in the August Clear-cut scenario (Fig. 7a and Table 2; soil volumetric water content is expressed as the decimal fraction of soil volume occupied by water). This increase initially occurred in the deepest layers of the soil, layers from which the vegetation in the Clear-cut (Prehistoric) scenario cannot (can) withdraw water. As the deepest model layers became more moist and the available water capacity of each layer decreased, the shallower model layers progressively became more moist. The top soil layer anomalously dried early in the month (Fig. 7a), likely from an increase in evaporation from the exposed soil surface and a decrease in precipitation (Fig. 6b). Although removing the canopy reduced interception loss and canopy water storage (Fig. 6d) and increased the fraction of incident precipitation reaching the soil surface, the overall decrease in precipitation was the dominant term and thus reduced the total input of water to the soil column. It was not until the deeper soil layers moistened more considerably in the latter part of the month that the near-surface soil moisture increased because of a reduction in downward moisture transport. By the end of the August simulations, there was essentially no gradient in volumetric water content with depth in the Clear-cut scenario, as compared with an ensemble-mean gradient of 0.015 m3 m−4 for the Prehistoric scenario.

In the uppermost (0–10-cm) soil layer, soil temperatures were tightly coupled to ground temperatures. Section 3b showed that approximately 50% of the anomalous surface energy in the Clear-cut scenario entered the soil column during the day, represented by an anomalously downward ground heat flux. This storage of heat raised near-surface soil temperatures by more than 6°C during the afternoon hours (Figs. 4a and 7b). Diurnal variations in temperature can also be seen at 10–40 cm (Fig. 4b). The diurnal variations were smaller at this depth, however, because of the high heat capacity of soil and the nighttime transmission to the surface of much of the additional heat stored in the near-surface layers during the day. Nonetheless, some of the additional heat at the 0–10-cm layer was transmitted through the depth of the soil column because the mean anomalous upward ground heat flux at night (45.7 W m−2) is less than the mean anomalous daytime downward ground heat flux (59.4 W m−2). Because the 0–10-cm soil layer shows no anomalous warming or cooling at the end of the night (500 LT; Fig. 4a), this imbalance in soil heat storage indicates that some of the heat that entered the soil column during the day was in fact being distributed through the depth of the column. This explains the continual warming of the 40–100- and 100–200-cm columns during the month; no diurnal variation can be seen at these layers because they are coupled to the surface on longer time scales than the upper layers (Fig. 7b). By the end of August, 100–200-cm soil temperatures have warmed by approximately 1.5°C in the Clear-cut scenario, an increase of almost 10% over the Prehistoric scenario (Table 2).

February soil-water differences were much smaller than those in August, with a decrease in 0–10-cm volumetric water content of approximately 0.005 m3 m−3 by the end of the month (not shown). Contrary to the August scenarios, differences in February soil moisture appear to be forced solely from the surface because of an increase in snow depth in the Clear-cut scenario. This increase was caused by an increase in snowfall and a reduction in canopy interception, which led to a slight increase in snow water equivalent (SWE) of almost 1 mm (Table 2). The difference in 0–10-cm soil moisture was negatively correlated to the difference in the area-mean SWE (r2 = 0.27, p < 0.01). This suggests that additional snow in the Clear-cut scenario acted to reduce the water input to the top layer of soil while the anomalous snowpack grew and that the deeper snowpack may have delayed the input of water to the soil from snowmelt. The soil moisture at deeper layers was unchanged in February, likely because the vegetation in the Prehistoric scenario removes little water from the deep soil during winter. Only very small differences were found in February soil temperature (Table 2).

4. Robustness of ensemble results

As with all ensembles of simulations, the significance of the ensemble-mean differences presented in section 3 depends upon the consistency with which the differences between individual ensemble members reflect the ensemble-mean differences. To measure the agreement among the individual members, we computed the difference between each matched pair of August Clear-cut and Prehistoric simulations. The standard deviation of these differences with respect to the difference in ensemble means gives an approximation of the intraensemble variability. In equation form, we calculated
i1558-8432-47-8-2166-e1
where S(X) approximates the intraensemble variability of a variable X; Ci(X) and Pi(X) represent the time-mean value of X from a member i of the Clear-cut and Prehistoric ensemble, respectively; and an overline represents the time-mean, ensemble-mean value of X. Figure 8 shows S in conjunction with the coefficient of variation
i1558-8432-47-8-2166-e2
where V(X) is the coefficient of variation for a variable X.

For the August differences, S is very low relative to the ensemble mean for T2m, latent heat flux, and ABL height. Note that V is less than 0.20 across most of the land cover–forcing area for all three variables; it is slightly lower for T2m and latent heat flux than for ABL height. Higher values of V are found close to the boundaries of the land cover–forcing area because the ensemble-mean differences decrease more quickly at the boundaries than S. Similarly low values of S and V were found in the interior of the land cover–forcing area for the other August quantities (e.g., sensible heat flux and precipitation) considered in section 3. This suggests that the signal shown in the ensemble mean is robust across all ensemble members, at least away from the land cover forcing boundaries.

When S and V were calculated for February, we found that, although the values of S were far lower than for August, the values of V were 2–3 times those of August for T2m and latent heat flux. Such a result reinforces the statement in section 3a that one ensemble member showed a response to deforestation in February but that the response was either damped or counterbalanced by the remaining ensemble members to give an ensemble-mean difference of near zero across the region (Fig. 2 and Table 2). In either case, there was at most a small, inconsistent difference between the Clear-cut and Prehistoric scenarios in February, which stands in marked contrast to the robust results obtained for August.

We forced MM5 with NCEP–NCAR reanalysis and NCEP upper-air data from 2000–04 as lateral boundary conditions. These years were chosen in part because they represent a wide variety of climatic conditions across the SRB. To assess this variability, we have taken data from three National Oceanic and Atmospheric Administration Automated Surface Observing System stations in the SRB. The stations are located in Williamsport (KIPT), Scranton/Wilkes-Barre (KAVP), and Harrisburg (KMDT), Pennsylvania. (Data are available from http://www.nws.noaa.gov/asos/.) February and August spatial- and temporal-mean temperature and precipitation anomalies from these stations indicate that our selected years include both warm and cool months and both wet and dry months (Fig. 9). There is a slight bias toward dry winters and a larger bias toward wet summers, but in general the anomalies are centered around the 1971–2000 climatological values. When considered with the low S and V described above, the variety of our boundary forcings lends further significance to our ensemble mean. MM5 produced a consistent response to deforestation despite inconsistent boundary forcing. In other words, the near-surface warming and drying described in section 3 appear to hold whether the regional climate was warmer or cooler, or wetter or drier, in any particular year in our ensemble.

5. Summary and conclusions

To explore the effects of deforestation on a midlatitude climate regime, four ensembles of five 28-day simulations were conducted using the MM5 mesoscale atmospheric model coupled to the Noah land surface model. The ensembles represented land surface conditions in summer (August) and winter (February) across northern Pennsylvania before and after extensive logging took place there in the late nineteenth century.

The results of the simulations showed a distinct seasonality in the response of the climate system to deforestation, with much stronger effects arising during the summer. Model results suggested the following impacts of deforestation in the August simulations, with values given as ensemble-mean, time-mean, area means (as in Table 2) unless otherwise noted:

  1. Latent heat flux was nearly halved by the deforestation, leading to a daytime excess of heat energy available at the surface. Approximately 40% of this excess was transmitted to the lower atmosphere as sensible heat flux; a further 50% was transmitted into the soil column by the ground heat flux, warming daily mean ground and 0–10-cm soil temperatures by nearly 3°C.
  2. Deforestation increased mean 2-m air temperatures by at least 1.5°C over most of the land cover–forcing area. This was despite an increase in the surface albedo and a subsequent decrease in net surface shortwave radiation.
  3. August atmospheric boundary layer height increased by 150 m over the forcing area with clear-cut land cover conditions, providing a deeper boundary layer through which to mix the additional heat energy from the increase in sensible heat flux.
  4. Model results suggested strong diurnal variation in the response of the near-surface atmosphere to deforestation. Late-morning and afternoon latent heat flux was greatly decreased by 140–220 W m−2 in the Clear-cut scenario, whereas sensible heat flux increased by 25–70 W m−2. Although ground and soil temperatures anomalously warmed during the day, the difference in 2-m temperatures was relatively small, owing to increased mixing of the additional heat energy through the lower atmosphere. Two-meter temperatures increased most during the nighttime hours, possibly because of reduced mixing with the boundary layer above and a return of a large portion of the heat energy stored in the soil during the day. It is important to note, however, that another boundary layer parameterization may include a different representation of the daytime-to-nighttime ABL regime transition that could alter these 2-m temperature estimates.
  5. Low-level mixing ratios and total precipitation amounts decreased by about 5% and 12%, respectively, because of deforestation. The decrease in precipitation indicated that the reduction in evapotranspiration, not the increase in moisture convergence, dominated the precipitation response to deforestation. Such a result implies a positive feedback process between decreases in evapotranspiration and precipitation that could cause additional surface drying and warming.
  6. A reduction in vegetative transpiration and root depth in the Clear-cut scenario led to anomalously higher values of August soil moisture at all model levels except the near-surface (0–10-cm) level. The sparse vegetation in the Clear-cut scenario withdrew far less water from the soil column at depth, causing increases in volumetric water content at lower levels first, followed by at levels nearer the surface. Near the end of the month, initial decreases in soil moisture in the 0–10-cm level due to increased evaporation from the soil surface and reduced precipitation were compensated by reduction in downward moisture transport to the wetter levels below.
  7. August soil temperatures anomalously increased throughout the month at all levels, owing to subsurface storage of a portion of the excess heat energy at the surface.

In summary, these results suggest that deforestation in a moist continental climate regime leads to an increase in 2-m air temperature, a drier low-level atmosphere, decreased precipitation, and warmed soil temperatures during the high-insolation months. Soil moisture was found to increase under these conditions because of a large decrease in transpiration with the loss of the canopy. Simulations showed the effects of deforestation during low-insolation months to be small and variable.

These results confirm and extend previous studies on midlatitude deforestation (Chase et al. 2000; Narisma and Pitman 2003). Moreover, the current study investigated a spatial scale of deforestation that has not been previously explored within a moist continental climate regime. The strong response of the land surface–atmosphere system to deforestation found in this study—and the variability of such a response on seasonal and diurnal time scales—suggests that land surface changes can lead to appreciable effects on regional climates. Thus, attribution of changes in the climate system should not neglect the effects of human-induced and naturally occurring land cover variability.

Acknowledgments

NPK was supported by grants from the Eugene duPont Distinguished Scholarship at the University of Delaware and from the Marshall Aid Commemoration Commission at the University of Reading. KRB was supported by a grant from the NASA Space Grant Foundation through the Delaware Space Grant Consortium. The authors thank our two anonymous reviewers for their helpful comments and suggestions on a previous version of this manuscript. We also appreciate constructive discussions with Prof. Brian Hanson, Dr. Delphis Levia, Ms. Gina Henderson, Ms. Tianna Bogart, Ms. Melissa Malin, and Ms. Daria Kluver at the University of Delaware and Ms. Victoria Sinclair at the University of Reading.

REFERENCES

  • Avissar, R., , and D. Werth, 2005: Global hydroclimatological teleconnections resulting from tropical deforestation. J. Hydrometeor., 6 , 134145.

    • Search Google Scholar
    • Export Citation
  • Bonan, G. B., 1997: Effects of land use on the climate of the United States. Climatic Change, 37 , 449486.

  • Brovkin, V., , A. Ganopolski, , M. Claussen, , C. Kubatzki, , and V. Petoukhov, 1999: Modelling climate response to historical land cover change. Global Ecol. Biogeogr., 8 , 509517.

    • Search Google Scholar
    • Export Citation
  • Chase, T. N., , R. A. Pielke, , T. G. F. Kittel, , R. R. Nemani, , and S. W. Running, 2000: Simulated impacts of historical land cover changes on global climate in northern winter. Climate Dyn., 16 , 93105.

    • Search Google Scholar
    • Export Citation
  • Chen, F., 2005: Variability in global land surface energy budgets during 1987–1988 simulated by an off-line land surface model. Climate Dyn., 24 , 667684.

    • Search Google Scholar
    • Export Citation
  • Chen, F., , and J. Dudhia, 2001: Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129 , 569585.

    • Search Google Scholar
    • Export Citation
  • Chen, F., and Coauthors, 1996: Modelling of land surface evaporation by four schemes and comparison with FIFE observations. J. Geophys. Res., 101 , 72517268.

    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , K. J. Westrick, , and C. F. Mass, 1999: Evaluation of MM5 and Eta-10 precipitation forecasts over the Pacific Northwest during the cool season. Wea. Forecasting, 14 , 137154.

    • Search Google Scholar
    • Export Citation
  • Copeland, J. H., , R. A. Pielke, , and T. G. F. Kittel, 1996: Potential climatic impacts of vegetation change: A regional modeling study. J. Geophys. Res., 101 , 74097418.

    • Search Google Scholar
    • Export Citation
  • Ek, M. B., , K. E. Mitchell, , Y. Lin, , E. Rogers, , P. Grunmann, , V. Koren, , G. Gayno, , and J. D. Tarpley, 2003: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational Mesoscale Eta Model. J. Geophys. Res., 108 .8851, doi:10.1029/2002JD003296.

    • Search Google Scholar
    • Export Citation
  • Gandu, A. W., , J. C. P. Cohen, , and J. R. S. de Souza, 2004: Simulation of deforestation in eastern Amazonia using a high-resolution model. Theor. Appl. Climatol., 78 , 123135.

    • Search Google Scholar
    • Export Citation
  • Ganopolski, A., , S. Rahmstorf, , V. Petoukhov, , and M. Claussen, 1998: Simulation of modern and glacial climates with a coupled global model of intermediate complexity. Nature, 391 , 351356.

    • Search Google Scholar
    • Export Citation
  • Giorgi, F., , M. R. Marinucci, , and G. T. Bates, 1993: Development of a second-generation regional climate model (RegCM2). Part I: Boundary layer and radiative transfer processes. Mon. Wea. Rev., 121 , 27942813.

    • Search Google Scholar
    • Export Citation
  • Grell, G. A., , J. Dudhia, , and D. R. Stauffer, 1995: A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Rep. NCAR/TN-398+STR, 122 pp.

  • Gutman, G., , and A. Ignatov, 1998: The derivation of the green vegetation fraction from NOAA/AVHRR for use in numerical weather prediction models. Int. J. Remote Sens., 19 , 15331543.

    • Search Google Scholar
    • Export Citation
  • Hahmann, A. N., , and R. E. Dickinson, 1997: RCCM2–BATS model over tropical South America: Applications to tropical deforestation. J. Climate, 10 , 19441964.

    • Search Google Scholar
    • Export Citation
  • Heck, P., , D. Lüthi, , H. Wernli, , and C. Schär, 2001: Climate impacts of European-scale anthropogenic vegetation changes: A sensitivity study using a regional climate model. J. Geophys. Res., 106 , 78177835.

    • Search Google Scholar
    • Export Citation
  • Henderson-Sellers, A., , R. E. Dickinson, , T. B. Durbidge, , P. J. Kennedy, , K. McGuffie, , and A. J. Pitman, 1993: Tropical deforestation: Modeling local- to regional-scale climate change. J. Geophys. Res., 98 , 72897315.

    • Search Google Scholar
    • Export Citation
  • Hong, S-Y., , and H-L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev., 124 , 23222339.

    • Search Google Scholar
    • Export Citation
  • Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43 , 170181.

  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437471.

  • Kanae, S., , T. Oki, , and K. Musiake, 2001: Impact of deforestation on regional precipitation over the Indochina Peninsula. J. Hydrometeor., 2 , 5170.

    • Search Google Scholar
    • Export Citation
  • Koren, V., , J. Schaake, , K. Mitchell, , Q-Y. Duan, , F. Chen, , and J. M. Baker, 1999: A parameterization of snowpack and frozen ground intended for NCEP weather and climate models. J. Geophys. Res., 104 , 1956919585.

    • Search Google Scholar
    • Export Citation
  • Lean, J., , and P. R. Rowntree, 1997: Understanding the sensitivity of a GCM simulation of Amazonian deforestation to the specification of vegetation and soil characteristics. J. Climate, 10 , 12161235.

    • Search Google Scholar
    • Export Citation
  • Leathers, D., , M. Malin, , D. Kluver, , G. Henderson, , and T. Bogart, 2007: Hydroclimatic variability across the Susquehanna River basin, USA, since the 17th century. Int. J. Climatol., in press, doi:10.1002/joc.1668.

    • Search Google Scholar
    • Export Citation
  • Narapusetty, B., , and N. Mölders, 2005: Evaluation of snow depth and soil temperature predicted by the Hydro–Thermodynamic Soil–Vegetation Scheme (HTSVS) coupled with the fifth-generation Pennsylvania State University–NCAR Mesoscale Model. J. Appl. Meteor., 44 , 18271843.

    • Search Google Scholar
    • Export Citation
  • Narisma, G. T., , and A. J. Pitman, 2003: The impact of 200 years of land cover change on the Australian near-surface climate. J. Hydrometeor., 4 , 424436.

    • Search Google Scholar
    • Export Citation
  • Nobre, C. A., , P. J. Sellers, , and J. Shukla, 1991: Amazonian deforestation and regional climate change. J. Climate, 4 , 957988.

  • Pan, H-L., , and L. Mahrt, 1987: Interaction between soil hydrology and boundary-layer development. Bound.-Layer Meteor., 38 , 185202.

  • Pan, Z. T., , E. Takle, , M. Segal, , and R. Arritt, 1999: Simulation of potential impacts of man-made land use changes on U.S. summer climate under various synoptic regimes. J. Geophys. Res., 104 , 65156528.

    • Search Google Scholar
    • Export Citation
  • Petoukhov, V., , A. Ganopolski, , V. Brovkin, , M. Claussen, , A. Eliseev, , C. Kubatzki, , and S. Rahmstorf, 2000: CLIMBER-2: A climate system model of intermediate complexity. Part I: Model description and performance for present climate. Climate Dyn., 16 , 117.

    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., , R. L. Walko, , L. T. Steyaert, , P. L. Vidale, , G. E. Liston, , W. A. Lyons, , and T. N. Chase, 1999: The influence of anthropogenic landscape changes on weather in south Florida. Mon. Wea. Rev., 127 , 16631673.

    • Search Google Scholar
    • Export Citation
  • Schultz, P., 1995: An explicit cloud physics parameterization for operational numerical weather prediction. Mon. Wea. Rev., 123 , 33313343.

    • Search Google Scholar
    • Export Citation
  • Semazzi, F. H. M., , and Y. Song, 2001: A GCM study of climate change induced by deforestation in Africa. Climate Res., 17 , 169182.

  • Su, H., , S. S. Chen, , and C. Bretherton, 1999: Three-dimensional week-long simulations of TOGA COARE convective systems using the MM5 mesoscale model. J. Atmos. Sci., 56 , 23262344.

    • Search Google Scholar
    • Export Citation
  • Werth, D., , and R. Avissar, 2002: The local and global effects of Amazon deforestation. J. Geophys. Res., 107 .8087, doi:10.1029/2001JD000717.

    • Search Google Scholar
    • Export Citation
  • Werth, D., , and R. Avissar, 2005: The local and global effects of African deforestation. Geophys. Res. Lett., 32 .L12704, doi:10.1029/2005GL022969.

    • Search Google Scholar
    • Export Citation
  • Xiu, A. A., , and J. E. Pleim, 2001: Development of a land surface model. Part I: Application in a mesoscale meteorological model. J. Appl. Meteor., 40 , 192209.

    • Search Google Scholar
    • Export Citation
  • Zheng, X., , and E. A. B. Eltahir, 1998: The role of vegetation in the dynamics of West African monsoons. J. Climate, 11 , 20782096.

Fig. 1.
Fig. 1.

A reference map showing the location of the two domains in the MM5 simulation and the area in which the land cover forcing was applied.

Citation: Journal of Applied Meteorology and Climatology 47, 8; 10.1175/2008JAMC1765.1

Fig. 2.
Fig. 2.

The difference in (left) August and (right) February ensemble means between the Clear-cut and Prehistoric scenarios (taken as Clear-cut minus Prehistoric) in (a),(b) the latent heat flux (W m−2), (c), (d) net surface shortwave radiation (W m−2), and (e), (f) 2-m temperature (°C).

Citation: Journal of Applied Meteorology and Climatology 47, 8; 10.1175/2008JAMC1765.1

Fig. 3.
Fig. 3.

(a) The difference in August ensemble means between the Clear-cut and Prehistoric scenarios in the sensible heat flux (W m−2), (b) the ratio of the August ensemble means (taken as Clear-cut divided by Prehistoric) in the Bowen ratio (unitless), and (c) the difference between the August ensemble means in the height of the atmospheric boundary layer (m). The line contours in (b) show the August ensemble-mean Bowen ratio in the Prehistoric scenario; contours are drawn every 0.05 from 0.20 to 0.35, with labels at 0.25 and 0.30.

Citation: Journal of Applied Meteorology and Climatology 47, 8; 10.1175/2008JAMC1765.1

Fig. 4.
Fig. 4.

The difference in ensemble-mean, area-mean (over the land cover–forcing area) quantities between the August Clear-cut and Prehistoric ensembles, separated into 3-h increments, for (a) T2m, ground temperature, ground heat flux, latent heat flux, sensible heat flux, and net surface shortwave radiation. (b) Difference in temperature with height. Net shortwave radiation is defined as positive downward; all other fluxes are defined as positive from the surface into the atmosphere. The height axis in (b) has been exaggerated to better show changes near the surface; tickmarks correspond to model levels.

Citation: Journal of Applied Meteorology and Climatology 47, 8; 10.1175/2008JAMC1765.1

Fig. 5.
Fig. 5.

Similar to Fig. 2, but only for August and for mixing ratio (colors; g kg−1) in (a) a zonal cross section taken at 41.5°N and (b) a meridional cross section taken at 77°W. Black line contours show the difference in (a) zonal and (b) meridional winds (m s−1), with westerly and southerly winds defined as positive; purple line contours show the difference in vertical velocity (cm s−1), with upward motion defined as positive. Line contours are drawn every 0.1 m s−1 or 0.1 cm s−1; the zero contour is not shown; negative contours are dashed.

Citation: Journal of Applied Meteorology and Climatology 47, 8; 10.1175/2008JAMC1765.1

Fig. 6.
Fig. 6.

As in Fig. 2, but only for August and in (a) 2-m mixing ratio (g kg−1), (b) total precipitation (cm), (c) σ = 0.91–0.97 vertically integrated moisture divergence (10−7 s−1), and (d) vegetative-canopy moisture (cm). The line contours in (c) show the difference in the σ = 0.91–0.97 vertically integrated wind.

Citation: Journal of Applied Meteorology and Climatology 47, 8; 10.1175/2008JAMC1765.1

Fig. 7.
Fig. 7.

Time series of the difference between the Clear-cut and Prehistoric August ensemble-mean (a) volumetric water content (m3 m−3) and (b) soil temperature (°C) for each soil level in the Noah LSM. In both (a) and (b), the legend labels give the midpoints of each soil layer.

Citation: Journal of Applied Meteorology and Climatology 47, 8; 10.1175/2008JAMC1765.1

Fig. 8.
Fig. 8.

Estimates of the intraensemble variability of the difference between the Clear-cut and Prehistoric ensembles [as in Eq. (1); shading] and the corresponding coefficient of variation [as in Eq. (2); black contours] for (a) 2-m temperatures (°C), (b) latent heat flux (W m−2), and (c) ABL height (m). Contour levels for the coefficient of variation go from 0 to 0.60 by 0.10, with an additional contour at 0.15. The coefficient of variation was calculated only where the std dev exceeded the white contour level.

Citation: Journal of Applied Meteorology and Climatology 47, 8; 10.1175/2008JAMC1765.1

Fig. 9.
Fig. 9.

Mean anomalies from three National Climatic Data Center Climate Reference Network stations within the inner MM5 domain in February temperatures (filled bar), August temperatures (unfilled bar), February precipitation (diagonal upslope), and August precipitation (diagonal downslope) for 2000–04. Anomalies were calculated from the 1971–2000 climatological means.

Citation: Journal of Applied Meteorology and Climatology 47, 8; 10.1175/2008JAMC1765.1

Table 1.

Summary characteristics of the land cover types applied to the forcing area in each simulation: albedo (α; unitless), roughness length (z0; cm), surface soil moisture availability (M0; % of field capacity), thermal inertia (k; J cm−2 K−1 s−1/2), green-vegetation fraction (GVF; unitless), root depth (D; cm), and minimum stomatal resistance (Rsmin; s m−1). An em dash indicates that the value is not defined.

Table 1.
Table 2.

Ensemble-mean, time-mean, area-mean quantities in the land cover–forcing area for the Prehistoric and Clear-cut ensembles and the difference between them (Clear-cut minus Prehistoric). Latent, sensible, and ground heat fluxes are all defined as positive from the surface to the atmosphere.

Table 2.
Save