Close Window

PC-05 Abstract

The Impacts of Land-Use/Land-Cover in Amazonia on Hydrometeorological Processes at Different Spatial and Temporal Scales

Roni Avissar — Rosenstiel School of Marine and Atmospheric Science - University of Miami (US-PI)
Pedro Leite Silva Dias — USP - Universidade de Sao Paulo (SA-PI)
Maria Assunção Faus da Silva Dias — IAG/USP (SA-PI)

Abstract

General circulation models (GCMs) are typically

used to simulate hydrometeorological processes at the seasonal-to-interannual

time scales. However, these models are not equipped to simulate the microscale

and mesoscale land-atmosphere interactions expected to be very important in

Amazonia, where convective clouds and precipitation play a key role. This is

because their grid resolution is much lower than the spatial scale of

convection, and than the landscape heterogeneity resulting from land use / land

cover change in Amazonia.


To fulfill the major objective of this NASA

Hydrometeorological Program (namely, to improve hydrometeorological predictions

at the seasonal-to-interannual time scales), it is crucial to bridge the gap

between these scales. To do that, we propose to nest a state-of-the-art regional

climate model (which will be validated with observations collected as part of

LBA) within a state-of-the-art GCM. This two-way interactive coupled model will

offer the unique capability of simulating explicitly (at a high resolution)

convection and precipitation in Amazonia, as well as their effects on the

general circulation. It will be used to produce ensembles of multi-year

simulations for various scenarios of land use / land cover change. Because this

type of coupled model has a time step of integration of the order of minutes, it

will be possible to study the hydrometeorological processes that it simulates at

time scales varying from hours to years.


This modeling study, which is expected to be

completed within three years, will contribute to the education of

Brazilian students and the interactions between the PI and Brazilian scientists

through a broad cooperative agreement recently developed between Rutgers

University and the University of São Paulo.


 


Introduction/Justification

Observations suggest the existence of

landscape-induced mesoscale circulations. These circulations are generated by

surface temperature contrasts resulting from landscape discontinuities 10-200 km

wide. Numerical studies have indicated that these circulations can be as strong

as sea breezes (e.g., Avissar and Chen, 1993; Lynn et al., 1995; Avissar and

Liu, 1996).


Clouds and precipitation associated with such

mesoscale circulations are expected to have a crucial impact on land-atmosphere

interactions. This is because clouds modify significantly the energy balance at

the ground surface by affecting the incoming solar and atmospheric radiation.

Precipitation change the soil-water content and, as a result, the water

availability for evapotranspiration and the surface energy fluxes. Furthermore,

both clouds and precipitation generated by these mesoscale circulations are not

distributed uniformly over the region in which they develop, introducing complex

nonlinear feedbacks that one cannot evaluate without the help of complex

hydrometeorological models (e.g., Avissar and Liu, 1996).


In Amazonia, where intensive deforestation is

taking place and convection is a very important aspect of the regional

hydrometeorology, both microscale and mesoscale landscape heterogeneities

created by the contrast of undisturbed forest and agricultural areas (pasture)

are observed. Such heterogeneities are clearly seen in satellite-derived picture

of the region where the joint LBA Mesoscale Intensive Field Campaign (LBA-MIFC)

- TRMM Validation Site will take place. Preliminary simulations performed by the

PI and his research team with the Regional Climate Model Version 2 (RegCM2) and

the climate version of the Regional Atmospheric Modeling System (ClimRAMS) over

this domain indicate a significant impact of the landscape on the precipitation

distribution and amount (unpublished). It should also be emphasized that using

satellite images of Amazonia, Cutrim et al. (1995) found a clear signal of the

impact of deforestation on cumulus clouds, supporting these preliminary modeling

results.


As emphasized in Section III of this NRA, from

an hydrometeorological standpoint, one of the major objective of LBA is to

better understand, and possibly quantify, the impacts of land use / land cover

change in Amazonia on hydrometeorological prediction at the seasonal-to-interannual

time scale.


General circulation models (GCMs) are typically

used to simulate hydrometeorological processes at the seasonal-to-interannual

time scales. However, these models are not yet properly equipped to account for

the very important microscale and mesoscale processes and feedbacks that occur

between the atmosphere and the ground, as a result of / land cover change. This

is because their grid resolution (of the order of hundreds of kilometers) is

much lower than the spatial scale at which land cover varies, and than the

spatial scale at which convective clouds and precipitation (which play a key

role in the hydrometeorology of Amazonia) develop (of the order of a few

kilometers).


Therefore, to fulfill the major objective of

this NASA Program, it is crucial to bridge the gap between these scales. To do

that, two possible solutions come to mind: (i) develop a sophisticated

parameterization of land-atmosphere interactions capable of accounting for

high-resolution land-cover variation (and associated microscale and mesoscale

hydrometeorological processes), which then would need to be introduced in a GCM

and run for the relevant time scale; or (ii) develop a sophisticated model,

capable of resolving microscale and mesoscale hydrometeorological processes (at

least over the region of interest), and at the same time be also capable of

simulating the general circulation at the relevant time scale.


 


Research Objectives

We propose to use the latter approach, namely to

nest a state-of-the-art regional climate model (which will be validated with

observations collected as part of LBA) within a state-of-the-art GCM. This

two-way interactive coupled model will offer the unique capability of simulating

explicitly (at a high resolution) convection and precipitation in Amazonia, as

well as their effects on the general circulation. It will be used to produce

ensembles of multi-year simulations for various scenarios of land use / land

cover change. Because this type of coupled model has a time step of integration

of the order of minutes, it will be possible to study the hydrometeorological

processes that it simulates at time scales varying from hours to years.


The two models to be coupled are (i) the climate

version of the Regional Atmospheric Modeling System (ClimRAMS); and (ii) the

NASA - Goddard Institute for Space Studies (GISS) GCM. Both models are

well-known, well-documented, and quite robust. The PI has been working with RAMS

for many years, and he was on sabbatical at GISS, where he learned the GISS GCM.


 


Methodology

The proposed study will consist of three tasks,

to be accomplished within three years (estimated time for each task is given in

parenthesis).


 


Task 1: Evaluation of ClimRAMS (1.5 year)

For this purpose, we propose to use the data

sets being collected as part of the LBA campaign. An important feature available

in ClimRAMS for this particular project, is its two-way nesting capability,

which will be used here to simulate up to five embedded grids: (i) A

very-low-resolution grid covering a large part of the South-American continent

and part of the oceans surrounding it; (ii) A low-resolution grid covering part

of the Amazonian Basin; (iii) A medium-resolution grid covering Rondonia (iv) A

high-resolution grid centered around the highly-deforested part of Rondonia; and

(v) A very-high-resolution grid covering most of the LBA Mesoscale IFC site.

While convective cells will be more or less resolved in the very-high-resolution

grid which, therefore, will provide great details on precipitation and clouds in

the region of interest, the purpose of the low-resolution grid is to assimilate

large-scale observations (and, possibly, an analysis produced by either NCEP,

ECMWF, or INPE) of various meteorological conditions, and provide large-scale

forcing to the higher-resolution grids.


The model so configured will be run at least for

the LBA-MIFC periods (January-February 1999 and June-July 2000). If computing

resources permit, the entire period between the two MIFCs will also be

simulated.


To evaluate the ability of ClimRAMS to simulate

the hydrometeorology of this region, the results of this simulation will be

compared with observations, with a particular focus on clouds and surface energy

and water balance. Because the area scanned by the S-Pol radar will be

particularly rich in observations, the evaluation will concentrate on the

results produced in the very-high resolution grid.


As part of this task, a sensitivity analysis of

the importance of assimilation frequency (spatial and temporal) of various

variables will be performed. Also, the importance of grid resolution will be

estimated, by eliminating progressively the higher resolution grids. This will

provide a crucial information for Task 3, in terms of the size of the domain

that we will be able to simulate at a high to very-high resolution, given

available computer resources.


 

Task 2: Model Coupling (0.5 year)

The coupling approach to be used in this project

will rely on the fact that ClimRAMS is already equipped with a data assimilation

package. This package will also be tested as part of the evaluation procedure

described under Task 1. The output of the GISS-GCM (at the spatial and temporal

resolution that will be available when this task starts) will be assimilated in

ClimRAMS, thus providing the large-scale atmospheric background needed to

correctly calculate the hydrometeorological processes over the limited area

simulated at high resolution. The ClimRAMS output will be averaged over this

high-resolution domain and fed back into the GISS-GCM through the

parameterization of subgrid-scale fluxes, following a procedure suggested by

Avissar and Chen (1993). According to this approach, the divergence of the

subgrid-scale fluxes (assumed to be generated by microscale turbulence and

parameterized accordingly in all current GCMs) will be substituted with the

microscale and mesoscale fluxes calculated explicitly with ClimRAMS. While this

approach is not very sophisticated, except for a few adjustments that will need

to be made on some of the GCM parameterizations (e.g., radiation and land), it

requires only a modest effort to implement. At a later stage, we will consider

develop a more efficient model, which will utilize modern techniques that have

been developed to nest grids (as, for example, used in ClimRAMS).


 


Task 3: Impacts of Deforestation on

hydrometeorology (1 year)


Under this last task, we propose to evaluate the

impact of land use / land cover change on diurnal-to-interannual

hydrometeorological prediction at various spatial scales (from Rondonia to the

entire Earth).


For that purpose, we will produce at least two

ensembles of simulations with the nested models, each consisting of several

two-year simulations (i.e., realizations): (i) A reference-case ensemble, using

the current deforestation estimate as used in Task 1; and (ii) A test-case

ensemble, using the same version of the model (with the same initial conditions

and same forcing), but assuming no deforestation. If time and computer resources

permit, we will perform additional ensembles of simulations, assuming

deforestation increases of 25%, 50%, 75%, etc.


Initial conditions, for all ensembles of

simulations, will be assessed from an ensemble of 17-year simulations

(1979-1996) produced with the GISS GCM Version SI95, using observed

monthly-averaged sea-surface temperatures. These simulations, as well as the

GISS GCM Version SI95, have been described in details in Hansen et al. (1997).

Forcing, which include sea-surface temperature, land cover types, and emission

of greenhouse gases, will be derived from monthly mean data sets, at the highest

spatial resolution that will be available when this task is initiated. These

forcing conditions are also available at GISS.


The effects of land use / land cover change will

be assessed by calculating the differences in several climatological conditions

(including air temperature and precipitation) between the ensembles of

simulations, at different time scales (daily, weekly, monthly, seasonal, and

interannual) and different space scales (Rondonia to entire Earth).


 


Bibliography

Avissar, R. and F. Chen, 1993. Development and

Analysis of Prognostic Equations for Mesoscale Kinetic Energy and Mesoscale (Subgrid-Scale)

Fluxes for Large Scale Atmospheric Models, J. Atmos. Sci., 50,

3751-3774.


Avissar, R. and Y. Liu, 1996. A

three-dimensional numerical study of shallow convective clouds and precipitation

induced by land-surface forcings, J. Geophys. Res., 101,

7499-7518.


Cutrim, E., D. Martin, and R. Rabin, 1995.

Enhancements of cumulus clouds over deforested lands in Amazonia. Bull. Amer.

Meteorol. Soc., 76, 1801-1805.


Hansen, J., and 42 co-authors, 1997. Forcings

and chaos in interannual to decadal climate change, J. Geophys. Res., 102,

25679-25720.


Lynn, B.H., D. Rind, R. Avissar, 1995. The

importance of mesoscale circulations generated by subgrid-scale landscape

heterogeneities in general circulation models. J. Climate, 8,

191-205.


April 6, 1999

 







Close Window