About KNMI Colloquia and workshops Staff Publications Contact Homepage Research KODAC: KNMI Operational Data Centre Job opportunities Larger Font Smaller Font Standard Font
LoCo/WATCH Workshop 25-27 June 2008

Purpose and set-up
Land-atmosphere interaction is a research area which is rapidly developing. After the well-known GLACE experiments and various diagnostic studies, new research has evolved in modelling and observating the degree of land-atmosphere coupling and feedbacks on local and regional scales.

The theme of hydrological land-atmosphere coupling is also addressed in the recently launched EU project WATCH, and the ongoing work in the West African Monsoon project AMMA.

In order to bring together the present-day key expertise on hydrological land-atmosphere coupling the WATCH project board and the GEWEX/GLASS panel together organized a workshop in De Bilt (The Netherlands), on 25 -27 June 2008. It continues the earlier De Bilt LoCo workshops held in 2002 and 2005. This workshop aims to produce an outline of an overview scientific paper on the subject, for the time being divided into 3 major topics:

  • what kind of diagnostic does express what kind of process?
  • how can we best measure hydrological land-atmosphere coupling?
  • in places with strong coupling: how strong is the local vs remote coupling, and where does it depend on?
  • how much of the coupling is related to land-processes, how much to atmospheric physical processes, how much to dynamical adjustment of circulation?
Spatial patterns
  • how do the spatial patterns of coupling change under greenhouse gas scenarios?
  • do changes in land use (irrigation) change the coupling/recycling of precipitation?
Program and presentations
Introduction and literature review Bart van den Hurkpdf-file
Feedback analyses in WATCH Eleanor Blythpdf-file 1 pdf-file 2
Tools and diagnostics of local land-atmosphere coupling
Longterm change in actual and potential evaporation Jim Shuttleworthpdf-file
In search of land-atmosphere interactions in satellite derived soil moistureRichard de Jeupdf-file 1 pdf-file 2
The KNMI SCM testbedRoel Neggerspdf-file
The Land Information System (LIS)Joe Santanellopdf-file 1 pdf-file 2 pdf-file 3
Characterizing local land-ABL coupling with a single column model Mike Ekpdf-file
Coupling mechanisms derived from observational/modelling studies
Precipitation recycling in Europe Bernie Bisselinkpdf-file
Land-atmosphere coupling in West AfricaChris Taylorpdf-file
Land-atmosphere coupling mechanisms with a potentially large scale impactStefan Hagemannpdf-file
Spatial distribution of coupling strength
Assessing land-atmosphere coupling over the continental US form spaceCraig Fergusonmail for pdf-file
Present day conditionsRandy Kosterpdf-file
Future conditionsSonia Seneviratnepdf-file
Workshop report (draft)
Participants (see photo)
  • Berny Bisselink (VU)
  • Chris Taylor (CEH)
  • Joseph Santanello (NASA/UMD)
  • Christa Peters-Lidard (NASA)
  • Randy Koster (NASA)
  • Sonia Seneviratne (ETH)
  • Richard de Jeu (VU)
  • Cor Jacobs (WUR)
  • Paul Houser (CREW)
  • Roel Neggers (KNMI)
  • Stefan Hagemann (MPI)
  • Andrea Molod (NASA)
  • Jim Shuttleworth (Univ. Arizona)
  • Michael Ek (NCEP)
  • Craig Ferguson (Univ Princeton)
  • Eleanor Blyth (CEH)
  • Helio Camargo Jr (KNMI)
  • Bart van den Hurk (KNMI)
Summary of talks (by Helio Camargo)
1. Introduction and Literature Review – Bart van den Hurk

Land-atmosphere coupling is a subject of growing interest in the scientific community. As measuring the phenomena is not straight forward and must be done indirectly, the use of models with different degrees of complexity have been used to help its diagnostics and the better understanding of the physical mechanisms responsible for it. In general, these models represent the soil and atmospheric boundary layer processes whereas others include some other physical processes like radiation, cloud formation, etc, where soil moisture feedbacks on precipitation or temperature can be studied given some idealized initial states or forcings. The sensitivity of the climate to these conditions showing for example the likelihood of dry or wet conditions to prevail during Summertime in Europe (Entekhaby 1992, D’Andrea et al. 2006), the negative feedback of a humid atmosphere (McNaughton and Spriggs, 1989 ) as well as some diagnostics indices and methods for coupling measurement (e.g. Findell and Elthair 2003, Ek and Hotslag 2004, Santanello 2005) could be studied and better understood, despite of their relatively simplicity compared to ocean coupled or atmospheric global circulation models.

Even though “simpler” models give some insights on how the mechanisms work and suggest methods for assessing coupling strength in a more local scale, it is also important to map globally the regions where precipitation or temperature are sensitive to land processes. In this prospective, the omega coupling coefficient (Koster et al. 2004) shows the “hotspots”, or regions where soil moisture contributes to the variability of the atmospheric state. Different model responses due to differences in land surface or boundary layer parametrization or even due to the representation of other processes which can affect coupling, indicate the necessity of improving the models representation, which is only possible with a better understanding of reality. It claims, then, for the necessity to explore the observation diagnostics of coupling and to understand how well the models can capture the feedback present in nature in the spatial and temporal time scale.

2. Feedback Analysis in WATCH – Eleanor Blyth

The Water and Global Change (WATCH) project goal is to bring together the hydrological, water resources and climate communities to analyze, quantify and predict the components of the current and future global water cycles and related water resources states; evaluate their uncertainties and clarify the overall vulnerability of global water resources related to the main societal and economic sectors. More specifically related to the theme of this workshop, WATCH is interested in assessing until what extent and/or where, the changes in feedbacks in many scales affects or will affect water resources and water reservoirs and/or how it can affect the frequency of floods or droughts.

In this prospective, there are important feedbacks that affect water resources in different time scales. Changes in land use (vegetation changes) for instance, affect water reservoirs more as a long term change. Relatively shorter terms include droughts which can be enhanced by land-atmosphere feedback, where dryer soils can reduce local precipitation positive feedback. In the wet side of the positive feedback, a more humid boundary layer over wetter soil conditions can be favorable for local convection (Douglas and Clark 2000) and influence clouds formation. Snow cover and CO2 feedbacks also influence water reservoir contents and are direct linked to the warm projections in future scenarios.

The challenge of trying to map and quantify the behavior of these feedbacks and how they affect water reservoir contents relies in trying to assess and understand their relative impact in present conditions. The use of both observational and modeling studies, provide reliable information for future conditions to be used for advisory purposes for hydrology applications. The migration of climate zones in future climate is an example where the change in the coupling strength regime may affect water cycle and subsequently water reservoirs.

3. Long-term change in actual and potential evaporation - Jim Shuttleworth

Pan evaporation has been showing systematic reduction in area average throughout the years over some regions in the globe. However, the association of such reduction to area averaged evaporation is controversial by the fact that these changes may be result of the reduction of surface radiation due to higher aerosol concentrations or due to the fact that the area average evaporation is increasing, leading to higher atmospheric humidity and then reducing pan evaporation.

Although it can not be globally applied due to spatial variation in its parameters, an adapted Penman-Monteith equation was proposed for the better understanding of the aforementioned controversy. The rate of change between actual and open water evaporation with varying surface resistance for a dry case was high, with the actual evaporation increasing while pan evaporation decreases. For the humid case, the changes are small and difficult to identify.

4. Precipitation Recycling in Europe – Bernie Bisselink

The precipitation recycling model (Dominguez et al. 2006) was applied to calculate the local recycling ratio over Europe, defined as the ratio of precipitation in a grid cell originated by the evaporation of a given region and the total precipitation in a cell.

In monthly time scales, high recycling ratios were found for central Europe in dry summers mainly due to weak advection in the region (small large-scale influence) and high evaporation rates. For the Balkans, high recycling ratios were encountered in wet summers due to the low advection component and convective precipitation systems. For the 2003 heat wave in Europe case study, the contribution of continental sources of moisture was dominant, with evaporation contributing a lot for local precipitation. In daily time scales, high recycling ratios are associated with low advection of moisture fluxes (as for monthly time scale) and are generally high when contribution of evaporation is relevant.

5. In search of land-atmosphere interactions in satellite derived soil moisture - Richard de Jeu

Long time series and a good spatial coverage of soil moisture data are very important information for land-atmosphere coupling diagnostics. A consistent 30-year surface soil moisture analysis product was created. The product is validated against observed data and analyzed in terms of the role of soil moisture in land-atmosphere interaction.

In general the developed product agrees well with observations. Autocorrelation correlation lengths described the temporal soil moisture characteristics. Statistical analysis showed that dominant soil moisture characteristics using the TRMM dataset were comparable with the El Nino index. The sign of the 28-year trend in soil moisture in Australia could be captured and the soil moisture bimodality in some regions of the world compared very well with the hotspots (Koster et al. 2004). It is important do highlight that soil moisture bimodality does not necessarily mean coupling. It also shows up when the precipitation regime is bimodal.

Being aware of the limitation of the blended product, it shows to be a very important tool to have a temporal and spatial coverage of soil moisture, a variable for which in situ measurement is not broadly available. For more information on the data, visit: www.geo.vu.nl/~jeur/lprm

6. Assessing land-atmosphere interactions over the continental US from space – Craig Ferguson

As a pioneer attempt to assess land-atmosphere interactions remotely, relationships between remote sensing coupling variables and soil moisture were used. These variables (SW radiation, cloud albedo, cloud forcing, surface air temperature, and mixing ratio) compare well with observed data and the results of remote sensing soil moisture coupling is consistent with Bett’s 2007 approach.

Comparisons the with Findell and Eltahir framework (Findell and Elthair 2003) showed, in general, indications of the positive feedback in the eastern part of the US and of the atmospheric controlled region in the central-northwestern portion of the country. The transition and the negative feedback regions were not reproduced, as well as the hostpots identified by Koster 2004.

Although some of the remote sensing based data compared well with observational data for the studied variables, a better understanding of the spatial pattern suggested by the analysis helps in identifying weaknesses and strengths of the method. The use of a GCM output for a “good model” in Koster 2004 analysis was suggested for the method’s application and comparison with the hotspot regions

7. Characterizing local land ABL coupling with a single column model – Mike Ek

The use of a PBL relative humidity tendency was proposed to explore the influence of soil moisture in the development of the atmospheric boundary layer through a range of soil moisture values with different configurations of profiles. The derived equation allowed studying the effect of soil moisture (via evaporative fraction) and atmospheric stability (via the non-evaporative terms) in the evolution of the relative humidity in the ABL.

A case study for Cabauw in The Netherlands showed that soil moisture’s effect is to increase the relative humidity in the top of the ABL (being more favorable for clouds formation) when the stability is relatively strong. On the other hand, when the stability is relatively low, high relative humidity is found over lower soil moisture values, enhancing the role of the atmosphere in making this region more favorable to clouds formation.

The developed equation showed to be a method to assess land-atmosphere coupling, encouraging the application of the same methodology for other regions in the world and similar studies using GCM outputs to test and diagnose land-atmosphere coupling where there are no observed soundings.

8. The KNMI SCM testbed – Roel Neggers

The KNMI testbed is a Single Column Model system allowing “intercomparison” of parametrizations. The models are driven by the ECMWF reanalysis and are compared to observations at Cabauw to routinely evaluate the parametrizations to be used in global models.

Advantages of the testbed include new data input for the statistics of cases in a daily basis due to its daily integrations, the relative low computing time and the portability of testing different parametrizations. Other important feature of the system is the almost online (daily) evaluation also in terms of scores, allowing the quantification of the model performance. More specifically for land surface interactions, the testebed allows freedom turn on/off parametrizations for sensitivity studies.

For more details on the different kinds of settings, access: http://www.knmi.nl/~neggers/KPT/

9. The Land Information System (LIS) – Joe Santanello

The Land Information System (LIS) consists of an “ensemble” of Boundary Layer (BL) and Land Surface (LS) models nested in the high resolution mesoscale Weather Research and Forecast (WRF) model to perform sensitivity studies.

An experiment was conducted to test the sensitivity of dry, normal and wet surfaces in the evolution of the fluxes in the BL. In order to show land-atmosphere processes in an integrated way to diagnose coupling, the fluxes and states are represented in mixing diagrams. In these diagrams, vectors representing the diurnal change in the evolution of the fluxes of heat and moisture from the LS and the BL (entrainment) are drawn allowing the quantification and evaluation of the processes involved in this interaction. For this experiment, the evaporation rates increased with the surface moisture content, while the entrainment flux in the top of the BL decreased when going from the dry to the wet state, being more dominant in dry surfaces. This approach can be applied also for observational data and can be extended by using an extra vector in the case of an advection component in the system.

10. Land-atmosphere coupling in West Africa – Chris Taylor

Rainfall variability over West Africa can not be totally reproduced only due to ocean SST variations. Simulations showed that both together, ocean and land are necessary to explain rainfall variability, in other words, the observed anomalies could not be reproduced without considering land conditions (Zeng et al. 1999). This, plus the factor of the indication of the hotspots found in GLACE, suggesting coupling in the region, motivated series of studies concerning on the influence of the soil moisture in the fluxes and in the states variables in the Sahelian region.

Results showed that the beginning of the convection is most likely to occur in dry soils, neighbored by wet soils condition. Being large enough, this gradient induces a breeze like circulation, with convergence in the dry-wet transition. After this “triggering” process, wet surfaces are necessary to maintain the meso-scale convective systems development (“fueling”).

The net influence of soil moisture on triggering and fueling convection in the regional coupling are not yet understood and are still under investigation. West Africa region can be used as a nice “real experiment” to understand land surface coupling and the mechanisms responsible for the phenomena in different time and spatial scales.

11. Land-atmosphere coupling mechanisms with a potentially large scale impact – Stefan Hagemann

Different types of land-atmosphere in different future scenarios and different models resolutions (global and regional) are analyzed.

Concerning on the soil moisture-precipitation feedback, ECHAM GCM tend to overestimate the intensity of drier summers and wetter winters when compared to the RCM outputs over the Danube catchment. Due to the better representation of the spatial soil moisture variability in present day climate, RCM are more reliable to draw conclusions about future scenarios. Other important feedbacks like the snow-albedo feedback indicating snow and se ice retreating, the permafrost melting-greenhouse gases positive feedback, wetlands increasing area leading to more methane production, CO2-soil moisture negative feedback are uncertain and still a topic under study.

The mentioned land-atmosphere coupling mechanisms may lead to large scale changes in hydrology. In order to understand better the evolution of the present to future climate, the representation of an interactive vegetation and carbon cycle needs to be included in the models. It is recommended the use of RCM for some feedbacks assessment due to its better spatial representation of the feedback processes. Although there were differences in the feedback sign when going from GCM to RCM resolution with the representations of the same physic processes, the change is not the same for different systems.

12. Spatial Distribution of Coupling Strength: Present day conditions – Randy Koster

The land-atmosphere coupling is not straight forward to deduce from observations. Due to this factor, Global Circulation Models (GCM) are a useful tool to assess until which extent land state variations affect fluxes or variables in the overlying atmosphere. With the use of models, it is relatively easy to turn on/off processes to study the isolated and combined effect of each model parameter or parametrization allowing a better understanding of the models themselves and of the physical mechanisms involved in the coupling. The Global Land-Atmosphere Coupling Experiment (GLACE) was a model-based experiment that was the pioneer in pinpointing regions in the globe where land and atmosphere are strongly coupled.

As evaporation is dominant process in local coupling, Guo et al. 2006 used the concept of evaporation variability and coherence to show that both need to co-exist in order to have a useful metric regarding on coupling diagnostic.

As an attempt to better understand the spatial distribution of the model and to compare model results with observations, the effects of soil moisture in the variation of precipitation and temperature is studied via the evaporative fraction. This method allows the distinction whether the evaporation regime is soil moisture controlled or atmospheric controlled. It was shown for some spots in the US, that the evaporation can be controlled by either soil moisture or the atmosphere only in the central part of the country, coinciding with the hotspots encountered in GLACE. For each of the spots, the relation between the variability of temperature versus the variability of precipitation behaves differently, result that agrees with observations. Based in this information, two diagnostic indices where created indicating i) where and which soil moisture range is influencing in the temperature and ii) if soil moisture is in the transition regime between soil and atmospheric control on evaporation. Comparing model (with and without interactive land) indices results with observations, it is clear that both are complementary and must exist together in the regions of strong land-atmosphere coupling.

As the soil moisture seems to be important for the temperature and precipitation variability, the second phase of GLACE or GLACE2, will study the impact of land initialization in temperature and precipitation forecasts in the monthly time scale. A pilot study made with one single model (Koster et al. 2004), showed that land initialization contributed to improve the forecast skill in the monthly time scale in many regions in the globe, showing its potential value in terms of predictability.

13. Future Conditions – Sonia Seneviratne

As land-atmosphere coupling is important for some regions in the globe (Koster et al. 2004), a question that is always raised is that if this coupling strength or pattern will change in the future. When future scenarios are taken into account, the increased temperature variability in central and Eastern Europe can be explained in terms of changes in land atmosphere coupling. These scenarios create a hotspot of soil moisture temperature coupling in regions displaced northward compared to the present conditions, suggesting a migration of the transition zones. Heat waves in Europe are also linked to soil moisture variations and can be more frequent in the climate change context.

The reliability of future coupling assessment depends on the ability of the models to capture present day coupling, which depends on how well the modeled coupling compares to the nature (observational) coupling, even though it is being indirectly measured. The correlation between temperature and evapotranspiration indicates high temperature soil moisture coupling where the values are negative (high temperature is a response of low evapotranspiration – evapotranspiration is soil moisture controlled). Positive correlation indicates that evapotranspiration is energy limited. The regions with negative correlation in the model agree in general with the hostpots (Koster et al. 2004). The method can be applied for both, models and observations. The necessity of development of diagnostics using observations which can be compared to model results at least in a qualitative way would help in a better understanding of land-atmosphere interactions for the present and consequently for the future climate conditions.
Organizers and links