SEARCH

SEARCH BY CITATION

Keywords:

  • CMIP5;
  • soil moisture;
  • sensible heat;
  • LCL;
  • sensitivity;
  • climate change

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Metrics
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Data from the Coupled Model Intercomparison Project Phase 5 for historical and future climate scenarios are examined for changes in the energy cycle component of land surface feedback on the atmosphere, namely, through the linkages from soil moisture to sensible heat flux to the height of the lifting condensation level marking the cloud base. Climate models project heightened sensitivity in both of these segments of the feedback pathway over most of the globe. This is in agreement with studies showing similar increases in land-atmosphere feedback through the water cycle, despite different physical processes, and may contribute to prevalent droughts and floods found in most climate change forecasts.

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Metrics
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Dirmeyer et al. [2013a] showed that the feedback from land to atmosphere via the water cycle evident in Coupled Model Intercomparison Project Phase 5 (CMIP5) model simulations of twentieth century climate are consistent with previous studies [e.g., Dirmeyer, 2000; Douville, 2003; Koster et al., 2004, 2006; Guo et al., 2011]. More importantly, future climate pathways suggest an expansion of regions of strong land-atmosphere coupling and a lengthening of the seasonal periods of strong coupling Dirmeyer et al. [2013b]. However, climate change has multidimensional effects in climate space [Diffenbaugh and Giorgi, 2012].

Soil moisture affects the partitioning of net radiation between sensible and latent heat flux to the atmosphere. Through the water cycle, changes in evapotranspiration and its variability influence atmospheric humidity, stability, cloud formation, and precipitation. There also exists a thermally driven land-atmosphere coupling pathway through the energy cycle [e.g., Betts et al., 1996; Findell and Eltahir, 2003a]. This pathway is the subject of this paper. Surface sensible heat flux drives growth of the boundary layer, enabling the formation of clouds by mixing moist air upward to the height where temperature and pressure allow condensation to occur [Betts, 2004]. These two pathways are preferentially expressed in different regions [Findell and Eltahir, 2003b]. Temperature and humidity profiles, available shortwave and longwave radiation at the surface, and the land state determine which pathway is more critical. “Hot spots” of land-atmosphere coupling arise where complete coupling pathways are in effect [Guo et al., 2006].

Each pathway has two segments [Dirmeyer, 2006; Ferguson et al., 2012]. The land segment is the connection between soil moisture and surface fluxes. The atmosphere segment is where surface fluxes impact air temperature, humidity, and the occurrence and intensity of precipitation. Neither segment is completely independent of the other, as surface fluxes are determined by land and atmosphere conditions. Despite these interdependencies, this framework allows for a convenient partitioning of feedback pathways [e.g., Dirmeyer et al., 2012].

We examine here the thermal pathway, from soil moisture to sensible heat flux to boundary layer growth, in CMIP5 model simulations for recent and future climate scenarios. Simulations of twentieth century climate provide a multimodel estimate of the two segments of the thermal pathway. Future scenarios provide an opportunity to assess how coupled feedback might change over the next century.

Data and Metrics

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Metrics
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Monthly mean data from CMIP5 [Taylor et al., 2012] are used: single simulations from 10 climate models for transient twentieth century (historical) and future climate scenarios of moderate (4.5 W m−2; Representative Concentration Pathway 45 (RCP45)) and strong (8.5 W m−2; RCP85) radiative forcing [van Vuuren et al., 2011]. The models are listed in Table S1 in the supporting information. Data from 1916 to 2005 of historical runs are used, after linear trends are removed independently from time series of each of the 12 months. Trends are removed as they corrupt calculations of correlations and variances used to isolate the interannual fluctuations attributable to land-atmosphere feedback. For the future simulations, the last 90 years from each integration are detrended and used.

We follow the methodology of Betts [2004] to quantify the thermal feedback pathway from land to atmosphere. Surface soil moisture w and surface sensible heat flux H fields are used, and the height of the lifting condensation level above the ground in pressure coordinates PLCL is calculated from near-surface air temperature T, relative humidity r, and surface pressure PS [Georgakakos and Bras, 1984]:

  • display math

PLCL correlates strongly with boundary layer height [Dirmeyer, 2012]. Dew point Td is calculated using the formulation of Murray [1967].

We estimate the sensitivity of sensible heat flux to soil moisture inline image and the sensitivity of PLCL to sensible heat flux inline image using the 90 yearly values for each model at each land grid point for each month. The product of the two represents the sensitivity in the total pathway: inline image. Here upward sensible heat flux is positive. These differentials are estimated as the slope of the linear regression through the detrended monthly values for the 90 years of simulation estimated separately for each month of the year.

Assuming the monthly means from each year of a simulation are independent (found to be valid except for soil moisture in about half the models over some very arid grid boxes), a correlation between the pair of terms in the differentials larger than ±0.208 is significant at the 95% confidence level using Student's t test. Since significant sensitivities can exist where the actual impact is slight due to small variability in the denominator, we calculate sensitivity indices by multiplying each ratio by the interannual standard deviation of the term in the denominator [Dirmeyer, 2011].

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Metrics
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

The multimodel averages for each of the index terms are presented in Figure 1 for June, July, and August (JJA) and for other seasons in the supporting information (Figures S1–S3 in the supporting information). The patterns for the two segments are very similar over the summer hemisphere (spatial correlations of −0.71 in JJA and −0.65 in December, January, and February (DJF)), but less so over the winter hemisphere (−0.55 in JJA and −0.44 in DJF). In the equinox seasons there is fair correspondence across the globe. The models suggest a strong feedback path during JJA over much of southern North America, India, Southeast Asia, and northeastern China, as well as eastern Europe and the Sahel. All but the East Asian regions correspond well to the hot spots found through the hydrologic cycle [e.g., Koster et al., 2006]. Figures S4–S15 present seasonal means of the indices for each model.

image

Figure 1. Multimodel mean of the sensitivity indices for (a) atmosphere segment, (b) land segment, and (c) total feedback path, averaged for JJA in the historical climate scenario. Areas are masked where the multimodel mean of the correlations is not significant. Unit is W m−2 in Figure 1b, pascal otherwise. Striping indicates where at least eight models concur on an increase (vertical) or decrease (horizontal) in RCP8.5.

Download figure to PowerPoint

The changes for JJA in a changing climate are presented in Figure 2. Patterns are quite similar for the two cases, with changes often being somewhat stronger for the RCP8.5 case. Changes to the atmosphere segment (Figures 2a and 2d) are overwhelmingly positive, suggesting increasing sensitivity of cloud base to variations to sensible heat flux for most locations.

image

Figure 2. As in Figure 1 for the changes (a–c) from historical to RCP4.5 and (d–f) from historical to RCP8.5. Only nine models contribute to Figures 2a–2c (see note in Table S1). Unit is W m−2 in Figures 2b and 2e, pascal otherwise.

Download figure to PowerPoint

Changes in the land segment (Figures 2b and 2e) are less systematic. The stronger feedback sensitivities (negative values for this term) are largely colocated with hot spots from Figure 1b. For these areas, sensible heat flux will vary more with fluctuations in soil moisture. The total index (Figures 2c and 2f) indicates that atmospheric convection will be more strongly tied to soil moisture in a warming climate, consistent with the evapotranspiration-humidity analysis of Dirmeyer et al. [2013a]. Results for other seasons are consistent (Figures S16–S18).

Figures S4–S15 show that sometimes one or two models may dominate the multimodel mean, occasionally with the opposite sign to most other models. This problem can magnify when calculating differences. There is no objective way to filter outliers, but we can instead assess the sign of the change predicted by each model and tally these projections. Dirmeyer et al. [2013a, 2013b] apply this approach to CMIP5 results, using the null hypothesis that a change of either sign is equally likely. This approach is applied in Figure 1; the striping indicates areas where at least 8 of 10 models agree on the sign of change from historical to RCP8.5 cases.

As in Figure 2d, the overall dominance of positive changes, indicating an increase in the atmospheric segment sensitivity, is clear in Figure 1a. However, the regions of greatest agreement are much more widespread than only the areas of largest mean change and tend to indicate increased coverage of areas of strong feedback, rather than intensification within core areas. The same is true for the indices of the land segment (Figure 1b) and the total feedback pathway (Figure 1c). Considering, for example, North America, there is near-unanimous consensus for a stronger land segment along the Gulf of Mexico and in an arc from the central Rockies to the Great Lakes. However, the simple mean in Figure 2e gives no hint of this uncertainty, showing strong changes over the eastern two thirds of the conterminous United States. There are many other examples where model consensus and multimodel mean are inconsistent. Finally, the changes in land and atmosphere segments have different global patterns, and the land segment contributes most of the change in the total feedback pathway in JJA (based on spatial correlation among maps of model consensus), but the atmospheric segment contributes more in other seasons.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Metrics
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Dirmeyer et al. [2013a, 2013b] investigated the projected changes to indices of land-atmosphere coupling via moisture from CMIP5 simulations. Here the thermal land surface controls on the atmosphere are examined. This is the other component of enthalpy as identified by Santanello et al. [2009, 2011], examined by Tawfik and Dirmeyer [2014], and likely responsible for the climate change responses described by Mueller and Seneviratne [2012]. Separate sensitivity indices have been derived for the land and atmosphere segments, as well as a combined index.

The two separate indices are usually in good agreement over monsoon regions and the summer hemisphere, in general, but often poor in the winter hemisphere even in the subtropics. Soil moisture controls on boundary layer growth exist to some degree over all but cold or very dry regions, with the strongest feedback largely colocated with previously defined hot spots [e.g., Koster et al., 2006]. One notable divergence is that strong coupling is seen over southern China and Indonesia where no hot spot is apparent in the water cycle [cf. Dirmeyer et al., 2013a].

Model responses to both climate change scenarios are similar, with the RCP8.5 response typically slightly stronger. Models suggest global strengthening of land and atmosphere segments of the feedback pathway in the energy cycle, particularly around the margins of hot spots. A similar result has been found for the water cycle pathways [Dirmeyer et al., 2013a]. Such a concomitant response is not inevitable, as the two pathways involve very different physical mechanisms and predominate in different regions [Findell et al., 2011; Ferguson and Wood, 2011]. Yet models forecast that soil moisture will become a stronger factor in thermally driven boundary layer development in a warming climate, reinforcing the enhanced coupling in the water cycle. This mechanism could help explain the projected model trends toward more hydrologic extremes in a warming climate [Meehl et al., 2007]. With increased sensitivity, land use and cover changes could have more impact on climate than they do today.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Metrics
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We thank the CMIP climate modeling groups for making their model output available through the U.S. Department of Energy's Program for Climate Model Diagnosis and Intercomparison (PCMDI). Analyses for this study were performed by the coauthors as part of a graduate class project in Land-Climate Interactions at George Mason University, Fairfax, Virginia, USA. The research was supported by the National Science Foundation (ATM-0830068), the National Oceanic and Atmospheric Administration (NA09OAR4310058), and the National Aeronautics and Space Administration (NNX09AN50G).

The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Metrics
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Data and Metrics
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
README.txtplain text document1KReadme
CLIM714_B_Sup_v5.docxWord 2007 document6309KTable S1 and Figures S1–S18

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.