PI: Paul Dirmeyer Center
for Ocean-Land-Atmosphere Studies dirmeyer@cola.iges.org
Co-I: Ben Kirtman Center for
Ocean-Land-Atmosphere Studies kirtman@cola.iges.org
Co-I: Charles J. Vörösmarty
University of New Hampshire charles.vorosmarty@unh.edu
Collaborator: Humberto R. da Rocha
Universidade São Paulo, Brazil humberto@model.iag.usp.br
Collaborator: José A. Marengo
CPTEC, Brazil marengo@cptec.inpe.br
March 1999 www.iges.org/lba
This research project brings together scientists at the Center for Ocean-Land-Atmosphere Studies (COLA) and the University of New Hampshire, in collaboration with colleagues at CPTEC and the University of São Paulo in Brazil to investigate seasonal-to-interannual variability and predictability in the Amazonian climate system (atmosphere, ocean, land surface and basin hydrology) with a suite of established coupled models and carefully designed sensitivity studies (see Tables 1-4 in the research plan).
The proposed work will attempt to determine the local and remote influences of land and ocean on Amazonian climate variability; determine the contributions of land and ocean to climate predictability over Amazonia as well as the potential limits of predictability; and determine where observational monitoring may be most effective to aid forecasting of climate anomalies, and associated risks (e.g., drought, fire, impacts to agriculture, transportation, etc.).
The work is divided into two phases. In Phase I, an offline land data assimilation (LDA) built on framework of the Global Soil Wetness Project will be conducted over South America using the COLA Surface Hydrology System (SHS) to generate uniform gridded ½ resolution analyses of surface water and energy balance terms, incorporating high-resolution rainfall data, and meteorological station data (e.g. from DNAEE) where available. This product will be validated against in situ measurements from LBA-DIS and other relevant data archives, and using a Water Transport Model (WTM) for basin-scale streamflow validation, and. Also, the Regional Spectral Model (RSM) framework will be incorporated into the COLA-GCM to provide improved resolution over South America without sacrificing consistency of model physics.
In Phase II, coupled model integrations will be used to investigate the interactions of components of the climate system, and their impact on seasonal variability and predictability. A proven two-tier forecast system will be used in hindcast mode. Tier one uses an anomaly coupled ocean-atmosphere model system, including a land surface model, to generate forecasts of SST. Tier two uses the forecast SSTs from the anomaly coupled model as boundary conditions for an atmosphere-land-river coupled model system. This framework will allow us to investigate the individual roles which the tropical Pacific, tropical Atlantic, and Amazon land surface play in determining regional climate by selectively removing feedback from the atmosphere in each region. We will investigate not only feedbacks of ocean anomalies on climate over land, but land anomalies on ocean circulation (via heat and moisture fluxes with the atmosphere, or through affects on freshwater discharge to oceans), and the impacts of climate variability on surface hydrology.
The proposed work will require four categories
of observational and analyzed data: land surface parameters necessary for
the SHS and WTM schemes, near surface meteorological data for the LDA effort
of Phase I, initial conditions of ocean and atmosphere for the coupled
model experiments of Phase II, and a wide range of data for validation
of all elements of the study.
1. Motivation
What are the sources and limits of climate
predictability over the Amazon region? The climate of South America and
its surrounding oceans is marked by strong asymmetry about the equator,
and varies on a range of spatial and temporal scales. The seasonal to interannual
variations are strongly affected by the unique coastal geometry, distributions
of biota, and the steep topography of the Andes. Rainfall patterns over
the Amazon Basin are complicated by interactions with the oceanic Inter-Tropical
Convergence Zones (ITCZ) over the eastern Pacific (El Niño related)
and western Atlantic Oceans, penetrating mid-latitude fronts, and the South
Atlantic Convergence Zone (SACZ). Because of the sharp meridional gradients
in the mean climate, small displacements in the maximum rainfall and Atlantic
sea surface temperature (SST) can lead to large climatic impacts, particularly
on the seasonal to interannual time scale.
a. Climate anomalies and the Amazon region
Climate anomalies can have significant impacts on ecology and hydrometeorology in the Amazon Basin and surrounding areas. Water resources in the entire Amazon Basin are affected by seasonal and interannual variations in rainfall (Marengo 1995). Droughts lead to reduced agricultural production and low river levels, which impede boat traffic and hydroelectric generation, as evidenced by this year's El Niño. Floods similarly disrupt agriculture and transportation. Changing land use practices are driven by, and may even affect climate variability. Slash and burn techniques for clearing forest for planting or grazing contribute vast quantities of smoke to the atmosphere while steadily changing the distribution of vegetation that regulates the energy and water balance at the earth's surface. Outside the Amazon Basin, periodic droughts in the Nordeste have grave consequences for local agriculture and society. The atmospheric general circulation over South America advects water vapor to subtropical and mid-latitude regions east of the Andes from over the Amazon Basin. Virtually the entire continent is linked in some way to the climate of Amazonia. An understanding of the causes for climate variations, be they local or remote, can greatly aid planning and resource management in the region.
There are three large regions which appear to exert some influence over climate in Amazonia: the tropical Atlantic Ocean, the tropical Pacific Ocean, and Amazonia itself. Tropical Atlantic SST variability has a profound influence on the region's climate variability. Anomalous meridional gradients of SST over the equatorial Atlantic have a large impact on the total rainfall over northeastern Brazil through the modulation of the ITCZ's latitudinal position (Moura and Shukla 1981, Hastenrath and Greischar 1993, Nobre and Shukla 1996). The climate variability over the Amazon and Nordeste regions is further complicated by the fact that rainfall anomalies are also well correlated with extreme phases of the El Niño/Southern Oscillation (ENSO; Kousky et al. 1984, Marengo et al. 1993, Aceituno 1988). Impacts are not necessarily confined to regions in which the precipitation anomalies occur. Variations in river discharge and floodplain inundation result from rainfall anomalies that may be far upstream (Vörösmarty et al. 1996).
Most of the evidence for land feedbacks
on climate over tropical and subtropical continents come from a combination
of theory (Charney et al. 1977, Dickinson and Hanson 1984, Eltahir 1996)
and numerous deforestation modeling results (see Hahmann and Dickinson
1997 for a partial review). Some observational evidence of trends in rainfall
over Amazonia exists (Cauduro Dias de Paiva et al. 1995, Marengo et al.
1998), but it is not clear whether they can be ascribed to deforestation
as can local changes in surface fluxes (Gash and Nobre 1997). Although
there has been some gradual convergence, there is still a great deal of
variation between estimates of the climatic response to deforestation.
It should be of greater concern to the scientific community that so many
studies of the potential impacts on climate of tropical deforestation have
been undertaken before a reasonable understanding of the existing climate
has been attained. Without a better comprehension of the sources of climate
variability over Amazonia, their relative importance and potential predictability,
there is little foundation upon which to pin projections of deforestation
impacts.
b. The need for high resolution modeling over Amazonia
Understanding and predicting the climate of the tropical South American region requires sophisticated high resolution coupled ocean-land-atmosphere climate models that resolve the mesoscale details of the orography and vegetation, the physics and dynamics of the continental monsoon circulation, as well as the complex coupled interactions with the oceanic ITCZ and SACZ.
The complex terrain and steep topography of the South American region in many ways dictates the need for very high resolution models. The mesoscale model mountain/no-mountain experiments of Tanajura (1996) suggest that the Andes act as a barrier that separates the low level circulation associated with the subtropical high over the eastern South Pacific from the low level circulation over South America. Simulating the subtle circulation features associated with the Andes is beyond the resolution of current state-of-the-art coupled general circulation models (GCMs).
Simulating the ITCZ and SACZ also dictates the need for high resolution coupled models. For example, the northern branch of the ITCZ extends across much of the basin, merges with the continental monsoons of central and northern South America, but is meridionally confined to only a few degrees of latitude between 5oN-10oN. The distribution of rainfall over the South American continent also has great deal of mesoscale structure, particularly in the Amazon basin extending southeast, and in the vicinity of the Andes mountains. Even on the seasonal to annual time scales, large differences in the distribution of precipitation are seen between high resolution observed precipitation data sets and relatively low resolution model simulations, as well as between low and high resolution precipitation data sets (Costa and Foley 1998). Given these sharp rainfall gradients, any small spatial scale displacements lead to large droughts or floods.
Non-coupled regional model simulations of Amazon climate (da Rocha 1998) and deforestation impacts have been conducted. However, these studies have generally been focused on local land-atmosphere feedbacks, and not the larger scale influences of neighboring oceans.
Many areas of Amazonia and the Pantanal region, when viewed at GCM grid resolutions have greater than 10% open water coverage. The data set of Matthews (1989) indicates numerous areas of wetlands greater than 20% coverage. The low resolution of global GCMs cannot resolve even large features such as the Amazon flood plain or the Pantanal, which may have significant impacts on regional energy and water fluxes between land and the atmosphere. Patchiness in the pattern of vegetation and deforestation, evidenced in satellite images over Rôndonia, is also impossible for low-resolution GCMs to represent. Higher resolution will allow better representation of these areas in the models. Data sets now exist, or are being compiled as part of LBA, which make high-resolution studies more justifiable than ever before.
Coupling to river flux models over regions
and sub-basins also requires higher resolutions than traditional GCMs can
supply. For instance, the ½ river topology of R-HydroNET (Vörösmarty
et al. 1997) could be used in dynamical prediction models if the atmospheric
component were at a comparable resolution. Higher resolution in climate
models will greatly facilitate such hydrologic studies.
2. Goals
The work we are proposing entails:
This proposal addresses a number of the
priority topics spelled out in the NRA. We will utilize model experiments
to assess the interaction of the land surface in the Amazon region with
the global scale atmosphere and ocean using an established two-tier seasonal
prediction scheme with improved parameterizations of the land surface and
terrestrial hydrology. An established land surface data assimilation methodology
will be enhanced with data from pre-LBA and LBA-DIS data sets to estimate
the terms of the surface water and energy budgets over the Amazon basin
and surrounding regions. This will include estimation of the seasonal cycles
and interannual variability in the hydrologic cycle and creation of analyses
of land surface conditions to initialize seasonal coupled model integrations.
Water transport modeling will be included to assess the sensitivity of
streamflow to climate variability, and as a means of validation of simulated
hydrologic balance. Coupling of atmosphere, ocean, land surface and water
transport models will create a fully closed hydrologic system over the
Amazonian region. This closure will allow investigation of the interactions
between these components, and a uniquely thorough assessment of the roles
of each component within the climate system. This work therefore supports
the goals of several international scientific efforts, most notably IGBP,
and WCRP.
3. Research Envisioned
The proposed research consists of two phases,
each expected to require approximately half of the three-year period to
complete (see the Management Plan at the end of this proposal). Phase I
must be completed in order to execute Phase II. Each phase is described
below.
a. Phase I
Using the framework developed in the Global Soil Wetness Project (GSWP; IGPO 1995, Dirmeyer et al. 1998), we will perform global and regional (Amazon region) assimilation of observed, remotely sensed, and analyzed meteorological and radiance data into the COLA surface hydrology system (SHS) for the period 1986-present. The older COLA land surface scheme as used in GSWP performed quite well in simulating the mean and annual cycle of runoff over the Amazon Basin (Oki et al. 1997). As in GSWP, this land data assimilation (LDA) will be performed without coupling to an atmospheric model. Our focus will be on assimilation over the South America, capitalizing on unique data available over the region from LBA-DIS and other sources (e.g., C. Willmott, personal communication). Time varying remote sensing data of vegetation cover properties will be combined with taxonomic information on vegetation and soils to update the parameterizations of the terrestrial biosphere and soil hydrology models.
The LDA effort will produce two important data sets. The first is a multi-year data set of global and regional surface hydrologic and energy data. Being the product of a land surface model driven by gridded observations and analyses, the multi-year data set will be among the best continental-scale data sets that can be produced. Using the Water Transport Model (WTM) of Vörösmarty et al. (1989, 1996), we will compare our surface hydrologic estimates to streamflow data, and compare selected grid point simulations to LBA field site measurements for purposes of validation. The data will be available for use and comparison by the community after submission to the appropriate LBA-DIS center, according to the LBA-DIS regulations. Second, as a product of this assimilation, land surface initial conditions and boundary conditions for the coupled model experiments of Phase II will be created. They will be fully consistent with the global and regional model, since the same land surface scheme will be used both for the assimilation and the coupled modeling.
To address the need for high resolution,
we have chosen to use a regional spectral model (RSM, Juang et al., 1997)
incorporated into the COLA GCM. The primary reason for choosing the RSM
procedure over nested grid-point models is that the physics and dynamics
of the regional model are identical to those of the host global model.
This is particularly advantageous since the identical land surface model
can be used at all scales. The RSM, developed and used at the National
Centers for Environmental Prediction (NCEP), has proven to be comparable
in skill with the operational Eta weather prediction model (Juang et al.,
1997). Implementation and testing of the global/regional model system will
be conducted with enhanced resolution over the South American domain. We
plan to implement a resolution of ½ over most of South America and
adjacent regions of the Pacific and Atlantic Oceans, with a region of higher
¼ resolution over the Amazon Basin.
b. Phase II
We will bring together the developments of Phase I to investigate the predictability of climate over the Amazon basin and surrounding areas, and document the seasonal to interannual variability of the region. This will be done in two ways. We will use a two-tiered approach to make historical forecasts (hindcasts) of climate and check the validity of the results as a function of location and season to determine the practical and theoretical limits of predictability over Amazonia. We will also perform sensitivity studies where greater control is exercised over the boundary conditions of the climate model to isolate the relative impacts of the Pacific Ocean, Atlantic Ocean, and regional land surface conditions on the climate of the Amazon Basin.
To understand climate predictability, one must first demonstrate that the two-tiered coupled model-RSM prediction system produces skillful forecasts. We will perform a series of two tiered coupled model-RSM hindcasts of South American climate variability. Tier one of the prediction system is a global anomaly coupled GCM (the COLA atmospheric GCM coupled to the Modular Ocean Model of Pacanowski 1995) which, in an earlier Pacific-only version, has been successfully used for ENSO prediction (Kirtman et al. 1997). It should be noted that the complete land surface parameterization is used in tier one (see Table 1). A more recent version of the global anomaly coupled model is currently being tested for SST forecasts in both Pacific and Atlantic. Integrations of 6-9 months will be performed for each period of interest, with the first three months discarded to help remove the influence of initial conditions on the climate simulations.
In addition to control experiments (Control
1 in Table 2), a number of sensitivity experiments will be conducted
using the tier-one models. The interactive nature of selected components
of the surface will be shut off to investigate the separate roles that
the (a) Pacific Ocean , (b) Atlantic Ocean, and (c) Amazonian land mass
play in determining climate variability. The goal of these sensitivity
studies is to examine how the climate of the Amazon interacts with, and
is affected by the nearby oceans. Several different case studies will be
made. For example, we will repeat the forecast/hindcast of selected El
Niño and La Niña years without an interactive Atlantic Ocean.
Similarly, we will deactivate the Pacific Ocean or the land surface processes
in Amazonia. By comparing these sensitivity studies we can identify the
relative roles played by the Atlantic and Pacific SST anomalies and local
land surface feedbacks in determining South American anomalies in the climate
system. Cases where there are strong SST anomalies in the Atlantic will
also be considered. For the prescribed Amazon, climatological soil moisture
determined from the LDA of Phase I will be specified over South America
to assess the potential influence of land surface feedbacks on regional
climate, including ocean circulation. In addition, we will also investigate
how simulated freshwater discharge from the Amazon basin affects the simulation
of the Atlantic circulation. The proposed fresh water discharge experiments
are not intended to fully explore how river runoff affects ocean circulation.
However, these experiments will identify the potential importance of Amazon
discharge on local ocean climate. Table 2 provides a concise depiction
of the planned experiments.
| Tier 1
Components |
Model | Initial Conditions | Coupling |
| Ocean | MOM | COLA-ODA (Obs. SST prior to 1986) | Anomaly coupled
to GCM
(Coupled to WTM in River 1 experiment) |
| Atmosphere | COLA-GCM | NCEP | Anomaly coupled
to MOM
Directly coupled to SHS |
| Land | SHS | COLA-LDA | Directly coupled
to GCM
(WTM coupled to MOM in River 1 experiment) |
| Tier 1 Experiments | Control 1 | Prescribed Pacific 1 | Prescribed Atlantic 1 | Prescribed Amazon 1 | River 1 |
| Pacific | Coupled to GCM | Climatology | Coupled to GCM | Coupled to GCM | Coupled to GCM |
| Amazon | Coupled to GCM | Coupled to GCM | Coupled to GCM | LDA Climatology | Coupled to GCM (and WTM coupled to MOM) |
| Atlantic | Coupled to GCM | Coupled to GCM | Climatology | Coupled to GCM | Coupled to GCM |
| Cases to Examine | El Niño and La Niña years, large Atlantic anomalies | El Niño and La Niña years | Large Atlantic SST anomalies | Selection from Control 1 | Selection from Control 1 |
| Objective | Baseline predictability integrations | How Atlantic variability interacts with Amazon | How Pacific variability interacts with Amazon | How Amazon land variability affects SST | How river discharge affects SST |
In tier two, we will use the RSM procedure
incorporated into the COLA GCM (see Table 3). While the tier one global
coupled forecasts are mainly designed to produce SST forecasts, the focus
of tier two will be on ensemble climate prediction over the Amazon and
environs using the RSM. The tier one forecasts of SST anomaly will be used
as ocean boundary conditions for tier two. Probabilistic hindcasts of rainfall,
temperature, circulation features, and streamflow will be made and evaluated
for skillfulness. Validation against observed meteorological and hydrological
data will be performed.
| Tier 2
Components |
Model | Initial Conditions | Coupling to Atmosphere |
| Ocean | Prescribed from Tier 1 | Prescribed from Tier 1 | None |
| Atmosphere | GCM + RSM | NCEP | Directly coupled to SHS |
| Land | SHS + WTM | COLA-LDA | Directly coupled to GCM |
In order to assess the relative importance
of Atlantic and Pacific SST in influencing South American climate, we will
specify climatological SST in either the Atlantic or the Pacific. By comparing
the results with a control simulation that has tier one forecast SST in
both basins, we can assess the impact of SST variability (see Table 4).
We will similarly specify land surface conditions (vegetation properties,
soil moisture) from our Phase I LDA to isolate the regional land surface's
role in governing the region's climate. The upper limit of predictability
for the RSM system will be assessed using integrations with "perfect" boundary
conditions, which represent the ideal case where the future state of land
and ocean are known exactly in advance of the forecast. In both hindcasts
and sensitivity studies, the WTM will be used to produce time variations
in river flow for validation and comparison. Appropriate subsets of the
data and analyses produced as a result of the two tier system will also
be submitted to the respective LBA-DIS center, following the regulations
that establish terms of data protocol, data policies, metadata, etc.
| Tier 2 Experiments | Control 2 | Prescribed Pacific 2 | Prescribed Atlantic 2 | Prescribed Amazon 2 | Perfect 2 |
| Ocean | Control 1 | Prescribed Pacific 1 | Prescribed Atlantic 1 | Control 1 | Observed |
| Land | Interactive | Interactive | Interactive | Climatology | LDA |
| Operational-style forecast | How Pacific variability impacts climate | How Atlantic variability impacts climate | How Amazon variability impacts climate | Upper limit of predictability (perfect BCs) |
Validation efforts will be greatly enhanced by the participation of Dr. José Marengo of CPTEC. Dr. Marengo brings experience in the LBA-Data Information System, and will collaborate in the hydrological validation of Phase I LDA integrations and both the meteorological and hydrological validation of Phase II hindcasts. He is also working independently with scientists at the University of São Paulo and CIRES to estimate the current water balance of the region, its seasonal and interannual variations using aerologic, precipitation and river data. Those results will be used for cross-validation.
We will also collaborate with Prof. H.
R. da Rocha at the University of São Paulo. Prof. da Rocha is investigating
the impact of land surface and atmospheric heterogeneity below the scale
of hydrostatic models and its bearing on convective precipitation with
the mesoscale RAMS model coupled to SiB2 (Sellers et al. 1996a). We will
investigate how changes in Amazonian water balance would affect convection
within and beyond Amazonia, using the global-scale modeling results from
tier two as initial and lateral boundary conditions for RAMS integrations.
Comparison between comparable integrations with RSM and RAMS, particularly
for those cases where RSM performs poorly, will shed light on regional
modeling uncertainties, and the roles that different physics and coupling
strategies may have on climate simulation over Amazonia. This collaborative
work would compliment the studies proposed above, but would not be directly
supported by an award from this opportunity.
Aceituno, P., 1988: On the functioning of the Southern Oscillation in the South American sector. Part I: Surface climate. Mon. Wea. Rev., 116, 505-524.
Charney, J. G., W. J. Quirk, S. H. Chow, and J. Kornfield, 1977: A comparative study of the effects of albedo change on drought in semi-arid regions. J. Atmos. Sci., 34, 1366-1385.
Costa, M. H., and J. A. Foley, 1998: A comparison of precipitation datasets for the Amazon basin. Geophys. Res. Let., 25, 155-158.
da Rocha, H.R., M. A. F. S. Dias, L. Bounoua, J. Collatz, O. M. R. Cabral, and C von Randow, 1999: A biosphere-atmosphere coupled mesoscale model (SiB2-RAMS). Part I: building and validation for southeast Brazil (in preparation).
Dickinson, R. E., and B. Hanson, 1984: Vegetation-albedo feedbacks. Climate Processes and Climate Sensitivity, Geophysical Monograph 29, Maurice Ewing Volume 5, American Geophysical Union, 180-186.
Dirmeyer, P. A., A. J. Dolman, and N. Sato, 1999: The Global Soil Wetness Project: A pilot project for global land surface modeling and validation. Bull. Amer. Meteor. Soc., (in press).
Eltahir, E. A. B., 1996: Role of vegetation in sustaining large-scale atmospheric circulations in the tropics. J. Geophys. Res., 101, 4255-4268.
Gash, J. H. C., and C. A. Nobre, 1997: Climatic effects of Amazonian deforestation: Some results from ABRACOS. Bull. Amer. Meteor. Soc., 78, 823-830.
Hahmann, A. N., and R. E. Dickinson, 1997: RCCM2-BATS model over tropical South America: Applications to tropical deforestation. J. Climate, 10, 1944-1964.
Hastenrath, S., and A. Greischar, 1993: Circulation mechanisms related to Northeast Brazil rainfall anomalies. J. Geophys. Res. Atmos., 98, 14917-14923.
Juang, H.-M., S.-Y. Hong, M. Kanamitsu, 1997: The NCEP regional spectral model: an update. Bull. Amer. Met. Soc., 78, 2125-2143.
Kirtman, B. P., J. Shukla, B. Huang, Z. Zhu and E. K. Schneider, 1997: Multiseasonal predictions with a coupled tropical ocean global atmosphere system. Mon. Wea. Rev., 125, 789-808.
Kousky, V. E., T. Kayano and I. F. A. Cavalcanti, 1984: A review of the Southern Oscillation: Oceanic-atmospheric circulation changes and related rainfall anomalies. Tellus, 36A, 490-502.
Marengo, J., 1995: Variations and change in South American streamflow. Climatic Change, 31, 99-117.
Marengo, J. A., L. M. Druyan, and S. Hastenrath, 1993: Observational and modeling studies of Amazonia interannual climate variability. Clim. Change, 23, 267-286.
Matthews, E., 1989: Global data bases on distribution, characteristics and methane emission of natural wetlands: Documentation of archived data tape. NASA Tech. Memo 4153, Goddard Institute for Space Studies, New York, New York, 18 pp.
Moura, A. D., and J. Shukla 1981: On the dynamics of droughts in northeast Brazil: Observations, theory, and numerical experiments with a general circulation model. J. Atmos. Sci., 38, 2653-2675.
Nobre, P., and J. Shukla, 1996: Variations of sea surface temperature, wind, and rainfall over the tropical Atlantic and South America. J. Climate, 9, 2464-2479.
Oki, T., T. Nishimura, and P. Dirmeyer, 1997: Validating estimates of land surface parameterizations by annual discharge using Total Runoff Integrating Pathways. J. Japan Soc. Hydrol. Water Resour., 9, 416-425.
Pacanowski, R. C., 1995: MOM2 documentation, user's guide and reference manual. GFDL Ocean Technical Report #3, Geophysical Fluid Dynamics Laboratory, Princeton, NJ, 232 pp.
Tanajura, C. A. S., 1996: Modeling and analysis of the South American summer climate. Ph. D. Dissertation, University of Maryland, College Park, 164 pp.
Vörösmarty, C.J., B. Moore, M.P. Gildea, B. Peterson, J. Melillo, D. Kicklighter, J. Raich, E. Rastetter, and P. Steudler. 1989. A continental-scale model of water balance and fluvial transport: Application to South America. Global Biogeochemical Cycles, 3, 241-65
Vörösmarty, C. J., C. J. Willmott, B. J. Choudhury, A. L. Schloss, T. K. Stearns, S. M. Robeson, and T. J. Dorman, 1996: Analyzing the discharge regime of a large tropical river through remote sensing, ground-based climatic data, and modeling. Water Resour. Res., 32, 3137-3150.
Vörösmarty, C. J., C. Fernandez-Jauregui, and M. Concepcion-Donoso, 1997: Prototype Version 1.0 of the Regional, Electronic Hydrometeorological Data Network (R-HydroNET v1.0) for South America, Central America, and the Caribbean. [Project supported by UNESCO Regional Office for Science and Technology (Montevideo), Water Center for the Humid Tropics of Latin American and the Caribbean (Panama City), and the US Committee on Scientific Hydrology].