GLACE The Global Land-Atmosphere Coupling Experiment. 2. Model characteristics and comparison

1 GLACE: The Global Land-Atmosphere Coupling Experiment Model characteristics and comparison Zhichang Guo1, Paul A Dirmeyer1, Randal D Koster2, Gordon Bonan3, Edmond Chan4, Peter Cox5, C.T.Tony Gordon6, Shinjiro Kanae7, Eva Kowalczyk8, David Lawrence9, Ping Liu10, Cheng-Hsuan Lu11, Sergey Malyshev12, Bryant McAvaney13, J.L McGregor6, Ken Mitchell11, David Mocko10, Taikan Oki14, Keith W Oleson3, Andrew Pitman15, Y.C Sud2, Christopher M Taylor16, Diana 10 Verseghy4, Ratko Vasic17, Yongkang Xue17, and Tomohito Yamada14 11 12 18 October 2022 14 Center for Ocean-Land-Atmosphere Studies, Calverton, MD, 20705, 15 USA 16 NASA Goddard Space Flight Center, Greenbelt, MD, 20771, USA 17 National Center for Atmospheric Research, Boulder, CO 80307, USA 184 Meteorological Service of Canada, Toronto, Ontario M3H4 5T4, Canada 19 Hadley Center for Climate Prediction and Research, Exeter EX1 3PB, 20 UK 21 Geophysical Fluid Dynamics Laboratory, Princeton, NJ 08542, USA 22 Research Institute for Humanity and Nature, Kyoto 602-0878, Japan 23 CSIRO Atmospheric Research, Aspendale, Victoria 3195, Australia 24 University of Reading, Reading, Berkshire RG66 BB, UK 25 10 Science Applications International Corporation, Beltsville, MD 20705, 26 USA 11 27 National Center for Environmental Prediction, Camps Springs, MD 28 20746, USA 12 29 Princeton University, Princeton, NJ 08544, USA 13 30 Bureau of Meteorology Research Centre, Melbourne, Victoria 3001, 31 Australia 14 32 University of Tokyo, Tokyo 153-8505, Japan 15 33 Macquarie University, North Ryde, New South Wales 2109, Australia 34 16 Centre for Ecology and Hydrology, Wallingford, Oxfordshire OX10 35 8BB, UK 17 36 University of California, Los Angeles, CA 90095, USA 1 2 Abstract 3The twelve weather and climate models participating in the Global Land4Atmosphere Coupling Experiment (GLACE) show both a wide variation in the 5strength of land-atmosphere coupling and some intriguing commonalities In 6this paper, we address the causes of variations in coupling strength – both 7the geographic variations within a given model and the model-to-model 8differences The ability of soil moisture to affect precipitation is examined in 9two stages, namely, the ability of the soil moisture to affect evaporation, and 10the ability of evaporation to affect precipitation Most of the differences 11between the models and within a given model are found to be associated 12with the first stage – an evaporation rate that varies strongly and 13consistently with soil moisture tends to lead to a higher coupling strength 14The first stage differences reflect 15parameterization and model climate identifiable differences in Intermodel differences in the 16evaporation-precipitation connection, however, also play a key role model 11 Introduction Interaction between the land and atmosphere plays an important role 3in the evolution of weather and the generation of precipitation Soil moisture 4may be the most important state variable in this regard Much research has 5been conducted on the effects of soil wetness variability on weather and 6climate, encompassing various observational studies (e.g., Namais 1960; 7Betts et al 1996; Findell and Eltahir 2003) and theoretical treatments (e.g., 8Entekhabi et al 1992, Eltahir 1998) These studies notwithstanding, the 9strength of land-atmosphere interaction is tremendously difficult to measure 10and evaluate Consider, for example, attempts to quantify the impact of soil 11moisture on precipitation through joint observations of both Precipitation 12may be larger when soil moisture is larger, but this may tell us nothing, for 13the other direction of causality – the wetting of the soil by precipitation – 14almost certainly dominates the observed correlation Global-scale or even 15regional-scale estimates of land-atmosphere coupling strength simply not 16exist 17 This difficulty motivates the use of numerical climate models to 18address the land-atmosphere feedback question With such models, 19idealized experiments can be crafted and sensitivities carefully examined A 20few recent examples include the studies of Dirmeyer (2001), Koster and 21Suarez (2001), Schlosser and Milly (2002), and Douville (2003) Modeling studies, of course, are far from perfect The ability of land 2states to affect atmospheric states in atmospheric general circulation models 3(AGCMs) is not explicitly prescribed or parameterized, but is rather a net 4result of complex interactions between numerous process parameterizations 5in the model As a result, land-atmosphere interaction varies from model to 6model, and this model dependence affects AGCM-based interpretations of 7land use impacts on climate, soil moisture impacts on precipitation 8predictability, and so forth (Koster et al 2002) The broad usage of GCMs for 9such research and the need for an appropriate interpretation of model results 10makes necessary a comprehensive evaluation of land-atmosphere interaction 11across a broad range of models The Global Land-Atmosphere Coupling 12Experiment (GLACE) was designed with this in mind 13 In GLACE, twelve AGCMs perform the same highly-controlled numerical 14experiment, an experiment designed to characterize quantitatively the 15general features of land-atmosphere interaction In GLACE, three 16- 16member ensembles of 3-month simulations are performed: an ensemble in 17which the land states of the different members vary independently (W); an 18ensemble in which the same geographically- and temporally-varying land 19states are prescribed for each member (R), and an ensemble in which only 20the subsurface soil moisture values are prescribed for each member (S) By 21quantifying the inter-ensemble similarity of precipitation time series within 22each ensemble and then comparing this similarity between ensembles, we 23can isolate the impact of the land surface on precipitation – we can quantify 1the degree to which the atmosphere responds consistently to anomalies in 2land states (hereafter referred to as the “land-atmosphere coupling 3strength”) The companion paper (Koster et al., this issue) describes the 4experiment and analysis approach in detail and provides an overview of the 5model comparison Note that the focus on subsurface moisture (ensemble S above) is of 7special interest It is well accepted that the variability of soil moisture is 8much slower than that of atmospheric states (Dirmeyer 1995) Hope for 9improving the accuracy of seasonal forecasts lies partly within the “memory” 10provided by soil moisture By quantifying the impact of subsurface soil 11moisture on precipitation, GLACE helps evaluate a model’s ability to make 12use of this memory in seasonal forecasts 13 Koster et al (this issue) and Koster et al (2004) highlight “hot spots” of 14land-atmosphere coupling regions of strong coupling between soil moisture 15and precipitation that are common to many of the AGCMs What causes such 16commonalities, and how they relate to climatological and hydrological 17regime? Which aspects of land surface and atmospheric parameterization 18cause the large model-to-model differences of coupling strength among the 19AGCMs? How are the signals that exist in the land surface states transmitted 20to and manifested in the atmosphere states? 21 Such critical questions, which arise naturally from a survey of the 22GLACE results and lie at the heart of our understanding of land-atmosphere 1feedback, are addressed in the present paper First, we address in section 2the aforementioned commonalities of coupling strength patterns 3Comparison among GCMs in section provides an analysis of model-to4model differences in the coupling strength in the path of soil moisture’s 5impacts on precipitation In section 4, we explore the link between model-to6model differences in the coupling strength and differences in atmospheric 7and land surface parameterizations Further discussions and summary are 8presented in section 92 Commonalities in coupling strength 10 The multi-model synthesis used in the companion paper (Koster et al., 11this issue) proves to be an effective way to identify robust (across models) 12regions of significant soil moisture impact on precipitation and near-surface 13air temperature – the commonalities in geographic pattern synthesized from 14the approach are less subject to the quirks or deficiencies of any individual 15model We can apply the same multi-model analysis procedure here to the 16other model variables As in the companion paper (see Section of Part 1), 17we first disaggregate interpolate variables from each model to the same fine 18grid, one with a resolution of 0.5º × 0.5º, and then we average the results 19with equal weights 20 As explained in the companion paper, the variable Ω v measures the 21degree to which the sixteen time series for the variable v generated by the 22different ensemble members are similar, or coherent Thus, Ω v(S)-Ωv(W) or 1Ωv(R)-Ωv(W) are measures of the control of land states on the atmospheric 2variable v As in the companion paper, we computed Ω v and the standard 3deviation σv for each model across 224 aggregated 6-day totals (16 4ensemble members times 14 intervals in each simulation time-series) The upper left panel of Fig.ure shows the mean of ΩP(S) – ΩP(W) for 6precipitation across the 12 models, i.e., the model-average impact of 7subsurface soil moisture on precipitation This figure essentially repeats the 8contents of the top panel of Figure 10 from the companion paper Notice that 9the larger soil moisture impacts on precipitation generally occur in the 10transition zones between humid and arid climates, such as the central Great 11Plains of North America, the Sahel in Africa, and the northern and western 12margins of the Asian monsoon regions 13 How can we characterize the evaporation signal that best serves as a 14link between soil moisture anomalies and precipitation – that best explains 15the geographical variations of ΩP(S) - ΩP(W) shown in the figure? In Figure 2, 16we argue that such an evaporation signal (as a proxy for the full surface 17energy balance) must have two characteristics: it must respond coherently 18to soil moisture variations, and it must show wide temporal variations The 19four panels show idealized evaporation time-series for 16 parallel ensemble 20members under four situations: (i) a low coherence in the evaporation time 21series [i.e., a low value of ΩE(S) – ΩE(W)] and a low variability of evaporation 22[i.e., a low value of σE(W)]., (ii) a low coherence but a high variability of 1evaporation, (iii) a high coherence yet a low variability of evaporation, and 2(iv) a high coherence and a high variability of evaporation Clearly, cases (i) 3and (ii) cannot lead to a robust precipitation response (across ensemble 4members) to soil moisture, given that evaporation is the key link between 5the two, and evaporation itself has no coherent response to soil moisture A 6coherent evaporation response, however, does not by itself guarantee a 7coherent precipitation response For case (iii), the evaporation response to 8soil moisture is robust, but the atmosphere would not see a strong signal at 9the surface due to the low evaporation variability Only the fourth situation 10provides a signal for the atmosphere that is both coherent and strong 11 We argue that for soil moisture to affect evaporation, both Ω E (S) – 12ΩE(W) and σE(W) must be suitably high In other words, the product (Ω E(S)13ΩE(W)) × σE(W) must be high We use this product throughout this paper to 14characterize the strength of the evaporation signal for land-atmosphere 15feedback (We assume here that σE(W) and σE(S) are similar; analysis of the 16model data confirms this.) It is noted that this product is neither the most 17optimal diagnostic, nor immune to impacts of precipitation on evaporation 18As shown in the following sections, however, this product proves to be a 19good measure skill to characterize the model behavior in land-atmosphere 20coupling strength The upper right panel of Fig shows the global 21distribution of ΩE(S)-ΩE(W) (again, averaged across the models), and the 22lower left panel shows that for σE(W) Neither diagnostic by itself explains all 1characteristics of the distribution of ΩP(S) – ΩP(W) in Figure 1a Figure 1d 2shows the distribution of the product (Ω E(S)-ΩE(W)) × σE(W) averaged over 3the 12 models The resemblance between the geographical patterns of the 4product and those of ΩP(S) – ΩP(W) is strong (spatial correlation is 0.42) It 5suggests that the coupling between precipitation and soil moisture is largely 6local and confirms that the coupling is strongest in regions having both a 7coherent evapotranspiration (ET) signal and a high ET variability The scatter plots in Figure illustrate further the control of hydrological 9regime on the product (ΩE(S)-ΩE(W)) × σE(W) The lines represent a best fit 10through the mean of the dependant variable in bins of 200 points each A 11roughly linear inverse relationship is seen between the soil wetness and the 12impact of this wetness on ΩE(S)–ΩE(W) The scatter plot explicitly 13demonstrates that ET is more sensitive to land state in dry climates than in 14areas with moderate soil wetness The results are consistent with the finding 15of Dirmeyer et al (2000), that showed the sensitivity of surface fluxes to 16variations in soil moisture generally concentrates at the dry end of the range 17of soil moisture index In contrast, the standard deviation of ET (σE) is not 18large for low soil moisture, simply because of the small values of ET in such 19regions Put together, the product (Ω E(S)–ΩE(W)) × σE(W) has minima for 20very wet and dry soils, and a maximum at intermediate soil moisture values 21(Figure 3c) 10 2Fig Time series of evaporation for different ensemble members under four 3situations: (i) low ΩE with low σE, (ii) low ΩE with high σE, (iii) high ΩE with low σE, 4(iv) high ΩE with high σE (see text for detail) 35 4Fig Scatter plots of ΩE(S)–ΩE(W), σE , and (ΩE(S)–ΩE(W)) × σE against mean 5soil wetness All variables are averaged across the twelve models 36 2Fig Inter-model standard deviation of ΩE(S)−ΩE (W) and ΩP(S)−ΩP(W) among the twelve 3models (top) and the ratio of the mean to the standard deviation (bottom) 37 1 38 1Fig 5: Global distribution of (ΩE(S)–ΩE(W)) × σE for the models participating in GLACE 39 4Fig Areal average of σE * (ΩE(S)−ΩE (W)) vs ΩP(S)−ΩP(W) over global ice-free 5land points and some “hot spot” indicated as boxes enclosed by dash lines in 6Fig for all twelve models 40 2Fig Global average of ΩP(S)−ΩP(W) vs ΩP(R)−ΩP(W) over ice-free land points for 3all twelve models 41 4Fig Areal mean of ΩE(S)−ΩE (W), σE, and σE * (ΩE(S)−ΩE (W)) in the different climate 5regime (The values for UCLA are not shown because soil moisture values for this model were 6not available.) 42 5Fig Areal average of ΩE(S)−ΩE (W) vs σE over global ice-free land points and 6some “hot spot” indicated as boxes enclosed by dash lines in Fig for all 7twelve models 43 Fig 10 Global average of (ΩE(S)−ΩE(W)) × σE vs (ΩE(R)−ΩE(W)) × σE over ice-free 5land points for all twelve models 44 2Fig 11 Global average of (ΩE(S)−ΩE(W)) × σE over ice-free land points versus spatial 3pattern correlation between (ΩE(R)−ΩE(W)) × σE and ΩP(R)−ΩP(W) for all twelve models 45 2Fig 12 Global distribution of [Ω E ( S ) − Ω E (W )] [Ω E ( R ) − Ω E (W )] for the models 46 1participating in GLACE 47 3Fig 13 Global average over ice-free land points of coherence changes from 4control case to S experiments calculated separately from total precipitation, 5convective and large-scale precipitation components for those models which 6reported them separately 48 1 49 ... Abstract 3The twelve weather and climate models participating in the Global Land4Atmosphere Coupling Experiment (GLACE) show both a wide variation in the 5strength of land-atmosphere coupling and some... on the left in Fig The top panels 13show the inter -model standard deviation of Ω(S)−Ω(W) among the 12 models, and the 14bottom panels show the ratio of the mean to the standard deviation The. .. from model to model The GFDL, 14CAM3, and NSIPP models have the strongest land-atmosphere coupling strengths, and 15GFS/OSU, HadAM3, BMRC, and GEOS have the weakest (Table 1) The breakdown of the