• Alexander, M. A., and S. D. Schubert, 1990: Regional earth–atmosphere energy balance estimates based on assimilations with a GCM. J. Climate,3, 5–14.

  • Berbery, E. H., E. M. Rasmusson, and K. E. Mitchell, 1996: Studies of North American continental scale hydrology using Eta Model forecast products. J. Geophys. Res.,101 (D3), 7305–7319.

  • Betts, A. K., J. H. Ball, A. C. J. Beljaars, M. J. Miller, and P. Viterbo, 1996: The land surface–atmosphere interaction: A review based on observational and global modeling perspectives. J. Geophys. Res.,101 (D3), 7209–7225.

  • ——, F. Chen, K. E. Mitchell, and Z. I. Janjić, 1997: Assessment of the land surface and boundary layer models in the two operational versions of the NCEP Eta Model using FIFE data. Mon. Wea. Rev.,125, 2896–2916.

  • ——, P. Viterbo, A. Beljaars, H.-L. Pan, S.-H. Hong, M. Goulden, and S. Wofsy, 1998a: Evaluation of the land–surface interaction in the ECMWF and NCEP/NCAR reanalysis models over grassland (FIFE) and Boreal forest (BOREAS). J. Geophys. Res.,103, 23 079–23 085.

  • ——, ——, and E. Wood, 1998b: Surface energy and water balance for the Arkansas–Red River basin from the ECMWF reanalysis. J. Climate,11, 2881–2897.

  • Black, T. L., 1994: The new NMC mesoscale Eta Model: Description and forecast examples. Wea. Forecasting,9, 265–278.

  • Chen, F., K. Mitchell, and J. Schaake, 1996: Modeling of land surface evaporation by four schemes and comparison with FIFE observations. J. Geophys. Res.,101 (D3), 7251–7268.

  • Climate Prediction Center, 1996a: Winter weather review. Weekly Weather Crop Bulletin, Vol. 83, No. 11, 24 pp. [Available from Climate Prediction Center, W/NP52, Attn: Weekly Weather and Crop Bulletin, NOAA/NWS/NCEP, 4700 Silver Hill Road, Stop 9910, Washington, DC 20233.].

  • ——, 1996b: Spring weather review. Weekly Weather Crop Bulletin, Vol. 83, No. 23, 28 pp. [Available from Climate Prediction Center, W/NP52, Attn: Weekly Weather and Crop Bulletin, NOAA/NWS/NCEP, 4700 Silver Hill Road, Stop 9910, Washington, DC 20233.].

  • ——, 1997: Winter weather review. Weekly Weather Crop Bulletin, Vol. 84, No. 10, 16 pp. [Available from Climate Prediction Center, W/NP52, Attn: Weekly Weather and Crop Bulletin, NOAA/NWS/NCEP, 4700 Silver Hill Road, Stop 9910, Washington, DC 20233.].

  • Coughlan, M., and R. Avissar, 1996: The Global Energy and Water Cycle Experiment (GEWEX) Continental-scale International Project (GCIP): An overview. J. Geophys. Res.,101 (D3), 7139–7147.

  • Dunn, L. B., and J. D. Horel, 1994: Prediction of central Arizona convection. Part II: Further examination of the Eta Model forecasts. Wea. Forecasting,9, 508–521.

  • Ehrendorfer, M., M. Hantel, and Y. Wang, 1994: A variational modification algorithm for three-dimensional mass flux non-divergence. Quart. J. Roy. Meteor. Soc.,120, 655–698.

  • Giorgi, F., S. W. Hosteler, and C. Shields Brodeur, 1994: Analysis of the surface hydrology in a regional climate model. Quart. J. Roy. Meteor. Soc.,120, 161–183.

  • Groisman, P. Ya., and D. R. Legates, 1994: The accuracy of the United States precipitation data. Bull. Amer. Meteor. Soc.,75, 215–227.

  • ——, D. R. Easterling, R. G. Quayle, V. S. Golubev, A. N. Krenke, and A. Yu. Mikhailov, 1996: Reducing biases in estimates of precipitation over the United States: Phase 3 adjustments. J. Geophys. Res.,101 (D3), 7185–7195.

  • Gutowski, W. J., Y. Chen, and Z. Ötles, 1997: Atmospheric water vapor transport in NCEP–NCAR reanalyses: Comparison with river discharge in the central United States. Bull. Amer. Meteor. Soc.,78, 1957–1969.

  • Helfand, H. M., and S. D. Schubert, 1995: Climatology of the simulated Great Plains low-level jet and its contribution to the continental moisture budget of the United States. J. Climate,8, 784–806.

  • Higgins, R. W., K. C. Mo, and S. D. Schubert, 1996: The moisture budget of the central United States in spring as evaluated in the NCEP/NCAR and the NASA/DAO reanalyses. Mon. Wea. Rev.,124, 939–963.

  • ——, Y. Yao, E. S. Yarosh, J. E. Janowiak, and K. C. Mo, 1997: Influence of the Great Plains low-level jet on summertime precipitation and moisture transport over the central United States. J. Climate,10, 481–507.

  • International GEWEX Project Office, 1995: Major activities plan for 1996, 1997, and outlook for 1998 for GCIP. IGPO Publ. Series 16, Int. GEWEX Project Office, Washington, DC, 246 pp. [Available from 1010 Wayne Avenue, Suite 450, Silver Springs, MD 20910.].

  • Janjić, Z. I., 1990: The step-mountain coordinate: Physical package. Mon. Wea. Rev.,118, 1429–1443.

  • ——, 1994: The step-mountain Eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev.,122, 927–945.

  • Kanamitsu, M., and S. Saha, 1996: Systematic tendency error in budget calculations. Mon. Wea. Rev.,124, 1145–1160.

  • La Rue, J. A., and R. J. Younkin, 1963: Large-scale precipitation volumes, gradients and distribution. Mon. Wea. Rev.,91, 393–401.

  • Lin, Y., K. E. Mitchell, E. Rogers, and G. J. DiMego, 1995: Impact of satellite moisture observations on forecasts made by NMC’s Eta model. Preprints, 14th Conf. on Weather Analysis and Forecasting, Dallas, TX, Amer. Meteor. Soc., 301–302.

  • Mesinger, F., 1996: Improvements in quantitative precipitation forecasts with the Eta regional model at the National Centers for Environmental Prediction: The 48-km upgrade. Bull. Amer. Meteor. Soc.,77, 2637–2649.

  • ——, and T. L. Black, 1992: On the impact on forecast accuracy of the step-mountain (Eta) vs. sigma coordinate. Meteor. Atmos. Phys.,50, 47–60.

  • ——, Z. I. Janjić, S. Ničković, D. Gavrilov, and D. G. Deaven, 1988:The step-mountain coordinate: Model description and performance for cases of alpine lee cyclogenesis and for a case of Appalachian redevelopment. Mon. Wea. Rev.,116, 1493–1518.

  • Peixoto, J. P., and A. H. Oort, 1992: Physics of Climate. American Institute of Physics, 520 pp.

  • Rasmusson, E. M., 1967: Atmospheric water vapor transport and the water balance of North America. Part I: Characteristics of the water vapor flux field. Mon. Wea. Rev.,95, 403–426.

  • ——, 1968: Atmospheric water vapor transport and the water balance of North America. Part II: Large-scale water balance investigations. Mon. Wea. Rev.,96, 720–734.

  • ——, and K. C. Mo, 1996: Large-scale atmospheric moisture cycling as evaluated from global NMC analysis and forecast products. J. Climate,9, 3276–3297.

  • Roads, J., S. Chen, A. Guetter, and K. Georgakakos, 1994: Large-scale aspects of the United States hydrologic cycle. Bull. Amer. Meteor. Soc.,75, 1589–1610.

  • ——, ——, M. Kanamitsu, and H. Juang, 1998: Vertical structure of humidity and temperature budget residuals over the Mississippi River basin. J. Geophys. Res.,103 (D4), 3741–3759.

  • Rogers, E., T. L. Black, D. G. Deaven, and G. J. DiMego, 1996: Changes to the operational early Eta analysis/forecast system at the National Centers for Environmental Prediction. Wea. Forecasting,11, 391–413.

  • Ropelewski, C. F., and E. S. Yarosh, 1998: The observed mean annual cycle of moisture budgets over the central United States (1973–92). J. Climate,11, 2180–2190.

  • Rosen, R. D., D. A. Salstein, and J. P. Peixoto, 1979: Variability in the annual fields of large-scale atmospheric water vapor transport. Mon. Wea. Rev.,107, 26–37.

  • Trenberth, K. E., 1991: Climate diagnostics from global analyses: Conservation of mass in ECMWF analyses. J. Climate,4, 707–722.

  • ——, and C. J. Guillemot, 1995: Evaluation of the global atmospheric moisture budget as seen from analyses. J. Climate,8, 2255–2272.

  • Wallace, J. M., 1975: Diurnal variations in precipitation and thunderstorm frequency over the conterminous United States. Mon. Wea. Rev.,103, 406–419.

  • Wang, M., and J. Paegle, 1996: Impact of analysis uncertainty upon regional atmospheric moisture flux. J. Geophys. Res.,101 (D3), 7291–7303.

  • Yarosh, E. S., C. F. Ropelewski, and K. E. Mitchell, 1996: Comparisons of humidity observations and Eta Model analyses and forecasts for water balance studies. J. Geophys. Res.,101 (D18), 23 289–23 298.

  • Zhao, Q., T. L. Black, and M. E. Baldwin, 1997: Implementation of the cloud prediction scheme in the Eta Model at NCEP. Wea. Forecasting,12, 697–712.

  • View in gallery

    (a) Mississippi River basin and subbasins corresponding to GCIP large-scale areas, as represented by the Eta Model grid points. (b) Distribution of cooperative stations reporting hourly precipitation within the Mississippi River basin.

  • View in gallery

    (a) Observed precipitation during summer. (b) Model forecast precipitation during summer. (c) Difference between model forecast and observed precipitation. (d)–(f) Same as (a)–(c) but for winter. Contour intervals for (a), (b), (d), and (e) are shown in the panels and units are mm month−1. Contour intervals in (c) and (f) are uniform and equal to 25 mm month−1; the zero contour was omitted.

  • View in gallery

    Scatter diagrams of seasonal mean model forecast precipitation vs seasonal mean observed precipitation for all gauges within the Mississippi River basin: (a) summer, (b) winter. Units are mm month−1.

  • View in gallery

    May 1995–Apr 1997 time series of basin-averaged observed precipitation (solid line) and model forecast precipitation (dashed line). Units are mm day−1.

  • View in gallery

    Diurnal cycle of basin-averaged observed precipitation during summer. Units are mm h−1.

  • View in gallery

    Summertime nighttime (0000–1200 UTC) precipitation minus daytime (1200–2400 UTC) precipitation for (a) observations and (b) model forecasts. Contour interval is 0.5 mm (12 h)−1.

  • View in gallery

    Vertically integrated moisture flux as estimated from EDAS for (a) spring, (b) summer, (c) autumn, and (d) winter. The lower-right arrow represents a moisture flux of 400 kg (m s)−1. Vectors with magnitude smaller than 50 kg (m s)−1 are not displayed. Only one of every three grid points are shown to avoid cluttering.

  • View in gallery

    Vertically integrated moisture flux convergence as estimated from EDAS for (a) spring, (b) summer, (c) autumn, and (d) winter. Contour interval is 2 mm day−1.

  • View in gallery

    Evaporation estimated as a residue of the moisture budget equation for (a) spring, (b) summer, (c) autumn, and (d) winter. Contour interval is 1 mm day−1.

  • View in gallery

    Annual cycle of the basin-averaged moisture budget terms: observed precipitation (short dashes with solid squares), moisture flux divergence (long dashes with solid circles), local changes of atmospheric water vapor content (thin solid line with open squares), and evaporation (thick solid line with open circles). Units are mm day−1.

  • View in gallery

    Basin-averaged moisture flux convergence as estimated from eight times daily analyses (solid lines), four times daily analyses (0000, 0600, 1200, and 1800 UTC; short dashed lines), and twice daily analyses (0000 and 1200 UTC; long dashed lines). Units are mm day−1.

  • View in gallery

    Basin-averaged differences of moisture flux convergence between the eight times daily analysis estimates and the four times daily analysis estimates (dashed lines) and differences between the eight times daily analysis estimates and the twice daily analysis estimates (solid lines). Units are mm day−1.

  • View in gallery

    Basin-averaged moisture flux convergence as estimated from EDAS (solid lines) and from a set of 2.5° × 2.5° grids (derived from the Eta Model full resolution) displaced in space (dotted lines). The average of all estimates from the 2.5° × 2.5° grids is presented as a heavy long dashed line. Units are mm day−1.

  • View in gallery

    (a) Cross section at 30°N of the 0600 UTC meridional component of moisture flux during summer. Contour interval is 20 g kg−1 m s−1. (b) The 900-hPa moisture flux at 0600 UTC. The lower-right arrow indicates a moisture flux of 200 g kg−1 m s−1. Vectors of magnitude smaller than 30 g kg−1 m s−1 are not displayed; (c) and (d) same as (a) and (b) but for the moisture flux difference between 0600 and 1800 UTC.

  • View in gallery

    Vertically integrated moisture flux convergence at (a) nighttime (average of 0600, 0900, and 1200 UTC) and (b) daytime (average of 1800, 2100, 0000 UTC). (c) Difference (0600, 0900, and 1200 UTC) − (1800, 2100, and 0000 UTC) moisture flux convergence. Contour interval is 0.2 mm h−1.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 31 31 1
PDF Downloads 14 14 0

Mississippi Moisture Budgets on Regional Scales

View More View Less
  • 1 Cooperative Institute for Climate Studies, Department of Meteorology, University of Maryland at College Park, College Park, Maryland
© Get Permissions
Full access

Abstract

Two years of regional analyses based on the Eta Data Assimilation System (EDAS) are used to examine the mesoscale features of the moisture budgets of the Mississippi River basin and its subbasins. Despite the short period, basic aspects of the regional-scale seasonal means, annual cycle, and even diurnal cycle of the atmospheric water cycle are represented. The ability of the Eta Model to resolve mesoscale features of the low-level circulation is an important factor in improving the estimates of moisture flux convergence at regional scales. It appears that the internal consistency of moisture budgets estimated from EDAS analyses for basins of nearly 5 × 105 km2 is comparable to that computed from radiosondes for basins of about 2 × 106 km2 or larger. In other terms, the spatial scale of basins where consistent moisture budgets can be estimated appears to be reduced by almost one order of magnitude.

Area-averaged evaporation estimates (computed as residuals of the moisture budget equation) for basins of about 5 × 105 km2 range from near zero during winter in the northern subbasins to about 5–6 mm day−1 during summer in the southern subbasin. It is suggested that the slightly negative estimates of evaporation in the northern subbasins during winter may partly result from an underestimation of observed precipitation due to the combined effect of wind and solid precipitation. No attempt was made at computing the model’s moisture budget, since changes in the surface parameterizations prevented having a period long enough to achieve stable results. Broad aspects of the diurnal cycle during summer were also examined through nighttime–daytime differences. Consistent with other studies over the central United States, results show that the nighttime development of moisture flux convergence is associated with an increase of intensity of the low-level jet. Interestingly, the nighttime convergence of moisture flux is offset by divergence during daytime and, as a result, overall moisture flux divergence is observed during summer.

A comparative analysis was made of the observed and model forecast precipitation to assess the model’s overall performance during the 2-yr period. It was found that the spatial patterns, intensity, and even the broad aspects of the summertime diurnal cycle of the model forecast precipitation are similar to those observed. Nevertheless, some deficiencies exist: a dry bias was obtained over the central United States during summer and winter; during summer, the southeastern United States had an excess of precipitation similar to that observed in the National Centers for Environmental Prediction global model; during winter, forecast precipitation in the northwestern United States appears to have biases in location and intensity, which can be related to the large-scale component of the model precipitation.

Corresponding author address: Dr. Ernesto H. Berbery, Department of Meteorology, 3411 Computer Space Sciences Building, University of Maryland at College Park, College Park, MD 20742-2425.

Email: berbery@atmos.umd.edu

Abstract

Two years of regional analyses based on the Eta Data Assimilation System (EDAS) are used to examine the mesoscale features of the moisture budgets of the Mississippi River basin and its subbasins. Despite the short period, basic aspects of the regional-scale seasonal means, annual cycle, and even diurnal cycle of the atmospheric water cycle are represented. The ability of the Eta Model to resolve mesoscale features of the low-level circulation is an important factor in improving the estimates of moisture flux convergence at regional scales. It appears that the internal consistency of moisture budgets estimated from EDAS analyses for basins of nearly 5 × 105 km2 is comparable to that computed from radiosondes for basins of about 2 × 106 km2 or larger. In other terms, the spatial scale of basins where consistent moisture budgets can be estimated appears to be reduced by almost one order of magnitude.

Area-averaged evaporation estimates (computed as residuals of the moisture budget equation) for basins of about 5 × 105 km2 range from near zero during winter in the northern subbasins to about 5–6 mm day−1 during summer in the southern subbasin. It is suggested that the slightly negative estimates of evaporation in the northern subbasins during winter may partly result from an underestimation of observed precipitation due to the combined effect of wind and solid precipitation. No attempt was made at computing the model’s moisture budget, since changes in the surface parameterizations prevented having a period long enough to achieve stable results. Broad aspects of the diurnal cycle during summer were also examined through nighttime–daytime differences. Consistent with other studies over the central United States, results show that the nighttime development of moisture flux convergence is associated with an increase of intensity of the low-level jet. Interestingly, the nighttime convergence of moisture flux is offset by divergence during daytime and, as a result, overall moisture flux divergence is observed during summer.

A comparative analysis was made of the observed and model forecast precipitation to assess the model’s overall performance during the 2-yr period. It was found that the spatial patterns, intensity, and even the broad aspects of the summertime diurnal cycle of the model forecast precipitation are similar to those observed. Nevertheless, some deficiencies exist: a dry bias was obtained over the central United States during summer and winter; during summer, the southeastern United States had an excess of precipitation similar to that observed in the National Centers for Environmental Prediction global model; during winter, forecast precipitation in the northwestern United States appears to have biases in location and intensity, which can be related to the large-scale component of the model precipitation.

Corresponding author address: Dr. Ernesto H. Berbery, Department of Meteorology, 3411 Computer Space Sciences Building, University of Maryland at College Park, College Park, MD 20742-2425.

Email: berbery@atmos.umd.edu

1. Introduction

Estimates of the atmospheric water cycle are important for reasons that range from climate diagnostics and water resources management to validation of models and climate change scenarios using general circulation models (GCMs). Despite the conceptual simplicity of the problem, which can be expressed in a relatively simple water balance equation [see (1) or (2)], reliable descriptions have been limited to broadscale patterns usually averaged over areas large enough to smooth out internal heterogeneities (Trenberth and Guillemot 1995).

The area-averaged atmospheric branch of the hydrological cycle (Rasmusson 1968; Peixoto and Oort 1992) can be written as
i1520-0493-127-11-2654-e1
or using Gauss’ divergence theorem
i1520-0493-127-11-2654-e2
where
i1520-0493-127-11-2654-eq1
γ is the perimeter of the area A, 〈 · 〉 represents the area average, and n is a unit vector normal to the perimeter with an outward direction. The vertical integrations are performed from the top of the atmosphere (T) to the surface (S); W is the vertically integrated water vapor (or precipitable water), Q is the vertically integrated water vapor flux vector, q is the specific humidity, p is pressure, and g is the gravity; E and P are evaporation and precipitation over the area; V is the horizontal wind.

In an early study Rasmusson (1968) evaluated the line integrals from radiosonde data for basins of different size over North America and concluded that the water balance estimates were reliable for averages over basins whose area was typically 2 × 106 km2 or larger. The limiting factor in describing regional features was the density of rawinsonde observations and their sampling frequency. A historical background is presented by Ropelewski and Yarosh (1998), who then proceed to use observations to compute and compare the terrestrial and atmospheric water budgets.

Water balances have also been estimated from global model output, that is, simulations, short-term forecasts, or analyses resulting from four-dimensional data assimilation (4DDA) techniques. With 4DDA techniques it is possible to integrate the information from a diverse set of observations and model forecasts that skillfully fill in regions of sparse data (at the cost of certain model dependency). Still, early results describing moisture transports on hemispheric or global scales (e.g., Rosen et al. 1979) were limited due to the inaccuracy of the models’ representation of the wind (and moisture flux) divergence, particularly in the Tropics. Subsequent improvements in 4DDA and associated models, including their increasing resolution, have resulted in increasingly reliable estimates of the atmospheric water cycle (Trenberth 1991; Trenberth and Guillemot 1995).

Estimates of the water cycle at subcontinental scales have already been derived from global analyses. Roads et al. (1998) describe the vertical structure of moisture (and energy) budgets over the Mississippi River as estimated from the National Centers for Environmental Prediction (NCEP) global analyses, reanalyses, and a GCM. They suggest that residuals are large and can be used as a measure of the analysis skill. Gutowski et al. (1997) also computed atmospheric water balances based on NCEP reanalyses for the Upper Mississippi River and the Ohio River basins (whose size is about 5 × 105 km2; see also Fig. 1a) and made comparisons with observed river discharge. They found that the terrestrial and atmospheric components have an imbalance of about 10% for the combined basins, but the basins individually have biases around 40% and are of opposite signs, largely attributable to errors in the atmospheric component.

Water cycle climatologies often differ from model to model (Trenberth and Guillemot 1995; Wang and Paegle 1996; Higgins et al. 1996). In part the problem stems from the inadequate resolution in current global models as will be discussed here. However, the resolution of the current European Centre for Medium-Range Weather Forecasts global operational model now approaches typical resolutions of regional mesoscale models. Nevertheless, other effects, as too simple or inaccurate parameterizations may also play important limiting roles (Betts et al. 1998a).

The development of high-resolution regional models now provides the possibility for more accurate evaluations of the area-averaged water balance. Giorgi et al. (1994) used simulations from a regional circulation model nested into a GCM to estimate hydrologic budgets for drainage basins in the United States and found that the model could reproduce basic characteristics of the major basins, although they noted that better accuracy was still needed. Berbery et al. (1996) showed that the atmospheric moisture budget can also be estimated from a series of short-term forecasts from a regional mesoscale model. Their work showed that NCEP’s Eta Model forecasts were consistent with observations, with the added benefit of higher temporal and spatial resolution. In particular they found that the Eta Model successfully reproduced observed aspects of the Great Plains low-level jet (LLJ) and its diurnal cycle, a crucial aspect related to the atmospheric water balance.

These results have a degree of dependency of the corresponding regional models; a way to reduce the influence is by using regional model analyses as opposed to regional model forecasts/simulations. There is still a degree of model dependence in the regional analyses and the results also depend on the data assimilation system, but it is expected that the influence will be smaller. This is the case of the Eta Data Assimilation System (EDAS) currently used with the Eta Model for operational short-term forecasts at NCEP (Rogers et al. 1996). In an early evaluation, Yarosh et al. (1996) compared water balances from radiosonde data with “synthetic soundings” from analysis and forecasts of a 1994 preoperational version of the Eta Model. They found that model-derived estimates compare favorably with radiosonde estimates, although some degradation of the model estimates with increased forecast time is observed.

The objective of this study was, first, to document the atmospheric water cycle for subcontinental scales (105–106 km2) over the United States as determined from Eta Model regional analyses. Second, to examine the quality of the model’s precipitation forecast as a general measure of its ability to reproduce relevant features of the atmospheric water cycle. Last, to estimate the effects of model resolution on moisture budget estimates. These subjects are relevant for their intrinsic value in understanding the hydrologic cycle, but also possess an added value as diagnostic and baseline information useful for model development and possible future efforts to perform a regional reanalysis over North America based on the Eta Model. In this context, this diagnostic study addresses the usefulness of Eta Model output for studies of short-term climate variability. Specifically, the results of this research shed light on the current strengths and limitations of the Eta Model data assimilation system from a climatological rather than synoptical or case study point of view.

The Eta Model and associated data assimilation system, as well as the regions where this study is focused, are described in section 2. In section 3 the seasonal mean, annual cycle, and diurnal cycle of observed and model forecast precipitation are analyzed. Different aspects of the moisture budgets as estimated from EDAS analyses and observed precipitation are discussed in section 4. A brief description of the diurnal cycle of moisture flux, also computed from the EDAS analyses, is presented in section 5, while conclusions are given in section 6.

2. Basic information

a. Eta Model

The Eta Model, whose products are used in this study, is a mesoscale regional model being used operationally at NCEP. Its most novel feature is its vertical coordinate, η, a generalization of the σ coordinate (Mesinger et al. 1988). The model’s domain covers all of North and Central America and portions of the adjacent oceans with a grid resolution of about 48 km. The model’s 38 levels are unevenly distributed in the vertical to provide a detailed description of the vertical structures, in particular over lower terrain regions (see Rogers et al. 1996, their Fig. 1, for domain size and vertical levels). While the model domain and resolution are changed as needed for operational purposes, they were not changed during the 2-yr period covered by this study that extends from May 1995 through April 1997.

The model’s dynamics and physics have been discussed in several articles (Mesinger and Black 1992; Black 1994; Janjić 1990, 1994; Rogers et al. 1996). As with most operational models, physical packages are under continuous development and therefore the model was subject to some changes over the period of this study. The most relevant changes came into effect in October 1995, when fixes to the code were done, a cloud prediction model was added (Zhao et al. 1997), and vertically integrated water vapor from satellite estimates was included in the analysis system. In January 1996 an extended Pan–Mahrt land surface scheme was incorporated (Chen et al. 1996; Betts et al. 1997). Changes that took effect in February 1997 include an improved computation of cloud fraction and radiation effects, and improved soil moisture and bare soil evaporation.1 The aforementioned changes to the model’s code affected several variables resulting from the parameterizations, particularly evaporation. For these reasons it was chosen not to offer a comparison between the model-computed evaporation and the estimated evaporation estimated as a residual from the moisture budget equation.

The initial conditions are described in the next section; one-way interaction boundary conditions are taken from the global operational model forecast of the preceding cycle.

b. Assimilation system

The Eta Model has a concurrent analysis system known as the Eta Data Assimilation System, implemented operationally in April 1995. This four-dimensional data assimilation system blends the information from the heterogeneous mix of observations that now exists over the United States. The assimilation scheme begins with a global model 6-h forecast 12 h prior to the Eta Model forecast, followed by subsequent adjustments based on optimal interpolation from observations every 3 h during the 12-h period. In addition to the conventional observational data, information is assimilated from aircraft, wind profilers, radar, and, as mentioned earlier, vertically integrated water vapor based on satellite measurements (Rogers et al. 1996; Lin et al. 1995). EDAS has two basic advantages. First, it uses the same physics and dynamics for the analysis as for the model. Second, the observational data are incorporated every 3 h, allowing for a better and timelier ingest of information.

Eight EDAS analyses per day, as well as 12–36-h forecast products have been retrieved and processed in the model’s native grid since April 1995. (Native grid refers to the model’s staggered, Arakawa E grid.) Monthly values of all terms relevant for moisture and energy budgets are being computed for each analysis and forecast time, so that the “mean” diurnal variability is preserved. The variables being processed fall into two categories: those that are the result of—or are derived from—the data assimilation system, for example, wind, temperature, and moisture flux; and those resulting from parameterizations in the model, for example, precipitation, evaporation, and soil moisture. We will focus mostly on the first type in this article, an exception being precipitation, which is being evaluated in section 3 as an estimate of the overall performance of the model.

c. Basins

The areas of interest are the Mississippi River basin and the so-called Global Energy and Water Cycle Experiment (GEWEX) Continental-Scale International Project (GCIP) large-scale areas that encompass the following drainage basins (Fig. 1a): (a) Arkansas–Red River, (b) Upper and Lower Missouri River, (c) Upper Mississippi River, and (d) Ohio River. The area of the Mississippi River basin is 3.1 × 106 km2, which is covered by about 1215 Eta Model grid points. The area of the Missouri River basin is 1.36 × 106 km2 and it is covered by about 540 grid points. Each of the other basins are about 0.5–0.6 × 106 km2 in area and are covered by about 215 grid points each. The number of grid points is about one order of magnitude larger than those from typical global analyses. The Mississippi Delta is not included in these basins because of the difficulty of identifying basin boundaries in this region of complex drainage and diversion, and no reliable measurements of streamflow are available south of Vicksburg (Kentucky). Each of the smaller basins possesses unique topographic and climatological features. According to the GCIP Implementation Plan (International GEWEX Project Office 1995; see also Coughlan and Avissar 1996) the Arkansas–Red River basin has large east–west climatic gradients, and precipitation has large contributions from convective systems and large diurnal variability. There are minimal orographic effects on precipitation over most of the Upper Mississippi River basin, and winter snow accumulation and spring snowmelt are significant factors in the hydrology of this basin. The Ohio River basin is probably the most complex of all. It has important orographic effects from the Appalachian Mountains and it has the heaviest precipitation in the whole Mississippi River basin leading to frequent winter to spring flood events. The Missouri River basin, with its semiarid steppe-type climate, has a major terrain effect associated with the Rockies, and significant reservoir storage and regulation of streamflow.

3. Precipitation analysis

Evaluation of the Eta Model 12–36-h precipitation forecasts over the United States was done for the 2-yr period covered in this study, May 1995–April 1997. Monthly observed precipitation was computed from hourly precipitation data from a network of about 2600 precipitation gauge stations, which is available from the National Climatic Data Center [Fig. 1b; see Berbery et al. (1996) for a description of this dataset]. Throughout this article the units of choice are millimeters per day. Two exceptions are (a) millimeters per month, used when seasonal means are presented, and (b) millimeters per hour, used in the analysis of the diurnal cycle.

a. Summer and winter mean fields

Figure 2 depicts the mean observed and model forecast precipitation during summer and winter over the continental United States. Observed precipitation during summer (Fig. 2a) is largest in the eastern part of the country, with a maximum toward Florida. The model precipitation forecast (Fig. 2b) shows similar overall features, although model precipitation over Florida is more intense. This deficiency appears to be a common feature of other NCEP models as well. For example, Higgins et al. (1996) reported a similar excess of precipitation in the NCEP–NCAR reanalysis. The differences between the precipitation fields are shown in Fig. 2c; the model had an excess of precipitation along the eastern coast and Great Lakes, with a deficit of precipitation over the central United States. An analysis of daily forecasts for Colorado during the summer of 1996 (not shown) suggests that the dry bias was related at least in part to the model’s failure to trigger convection because of erroneous low moisture in the lower levels arising from low model evaporation. This is consistent with the findings of Dunn and Horel (1994) who observed a similar model behavior over Arizona that they attributed to an inadequate model representation of the increase of moisture in lower levels prior to the onset of convection.

During winter, maximum precipitation is observed in the northwest and southeast United States in both the observations and model forecasts (Figs. 2d,e). The largest positive differences (Fig. 2f) appear in the northwestern region and secondarily over the eastern coast. Differences near the West Coast appear in intensities but are also due to shifts in the location of maxima. Examination of the model’s partition of precipitation into convective and nonconvective (or large scale) contributions shows that the deficiencies in the northwest and northeastern coast rainfall are associated with the large-scale component; that is, they are not of convective origin. As during summer, a slight tendency toward negative differences (dry model bias) is noted in the central United States.

b. Precipitation estimates at the basins

Quantitative evaluation of the Eta Model daily forecast precipitation is a continuous task at NCEP, and has been documented in several articles (e.g., Mesinger 1996; Rogers et al. 1996). Berbery et al. (1996) also performed an evaluation of the model precipitation for the period August 1993–March 1994 using equitable threat scores and bias scores applied to monthly precipitation estimates computed from the 12–36-h forecasts. Scatter diagrams of observed versus model forecast precipitation are used as a general measure of the quality of the forecasts in this article. The model forecast precipitation was interpolated to the gauges’ locations;thus each point in the scatter diagrams represents the seasonal mean observed and model forecast precipitation at any gauge within the Mississippi River basin. During summer the scatter diagram of points within the Mississippi River basin (Fig. 3a) shows a cloud of points with a spatial correlation of 0.56 and a slope of 0.7, indicating a slight tendency toward higher values of observed precipitation that is consistent with Fig. 2. A better agreement between model and observations is achieved during winter (Fig. 3b), with a spatial correlation of 0.87, a slope of 0.82, and less dispersion. A similar analysis for the subbasins (not shown) exhibits a larger dispersion and, as for the Mississippi River basin, the larger spatial correlations are observed during winter.

The differences are reduced when precipitation is averaged over a basin: Time series of observed and model forecast precipitation averaged for each basin are presented in Fig. 4. In general, the seasonal cycle and intensity of both observations and model forecasts are similar, although the latter have a tendency for smaller values than the former. Precipitation differences for the Mississippi River basin are typically of the order of tenths of millimeters per day with an overall average of about 0.36 mm day−1; differences become larger during summer and for the smaller basins averaging around 0.5 mm day−1.

c. Summertime diurnal cycle of precipitation

Wallace (1975) showed that the diurnal cycle of precipitation over the United States is rather complex in the sense that important regional differences exist. In general, there is a nighttime maximum over the Great Plains that shifts eastward toward the afternoon. Figure 5, which depicts the diurnal cycle of basin-averaged observed precipitation for each of the Mississippi River subbasins, shows that the Mississippi River basin has a minimum of precipitation in the morning (around 1600 UTC), with a rapid increase during the afternoon (peaking around 2200 UTC) after which it decreases slowly during the nighttime hours. This basin-averaged diurnal cycle is in fact a combination of a large variety of diurnal regimes of precipitation (Wallace 1975; Higgins et al. 1997). Focusing on smaller areas a better representation of the different features described by Wallace (1975) is obtained. The Arkansas–Red River basin (Fig. 5e) shows two peaks of precipitation, one around 0700 UTC (about 0200 local time2) that has been related to the Great Plains LLJ, and a second one at about 2300 UTC (1800 local time). This is consistent with the description presented in Betts et al. (1998b) who noted three regimes over the basin (with maxima at midnight, afternoon, and after sunset).

The Missouri River basin, and particularly its lower portion (Figs. 5c,d) exhibits a precipitation maximum toward the late afternoon, also consistent with Wallace’s results. There is no well-defined diurnal cycle over the Upper Mississippi River basin (Fig. 5b), although there are indications of increased afternoon precipitation. The Ohio River basin (Fig. 5f) has a marked diurnal cycle with maximum precipitation in the afternoon (around 2100 UTC or 1600 local time) and a minimum that extends from local midnight to about 1000 LT. These regional features are consistent with the description by Wallace (1975) and others, indicating that the two years used in this study reproduce the fundamental features of the diurnal cycle of precipitation. In other words, the 2-yr averages seem stable enough to describe adequately regional features of the moisture budgets.

The Eta Model precipitation forecasts are accumulated in 6-h periods, so that only the broadscale features of the diurnal cycle can be analyzed. The difference between nighttime and daytime precipitation is presented in Fig. 6 for both observations and model forecasts. Nighttime is defined as the period between 0000 and 1200 UTC (approximately 1900 through 0700 local time of the following day) and daytime as the period between 1200 and 2400 UTC (0700 through 1900 local time). Observed precipitation (Fig. 6a) shows larger nighttime precipitation in the western portion of the Mississippi River basin, while toward the southeast precipitation is increasingly a daytime phenomenon. The model precipitation (Fig. 6b) reveals a similar pattern and, in particular, is able to reproduce the maximum night–day precipitation difference near the western part of Arkansas–Red River and Lower Missouri River basins. It also shows agreement toward the southeast, with increased daytime precipitation. The patterns on the West Coast are not clear; the model tends to have a somewhat increased nighttime precipitation (here, local time is 1700–0500), while only a hint is noticed in observations in northern California. However, in both cases the threshold between night and day precipitation is not well defined. While there is an overall agreement with observations, the model fails to place the transition zone over the Great Plains that in the forecast precipitation is shifted westward. The affected areas are the eastern part of the Arkansas–Red River and the Upper Mississippi River basins.

Evaluation of the 6-h accumulated precipitation is more complex. The model reproduces the broader features of the diurnal cycle of precipitation, but differences also exist, particularly in the central region (not shown). On the western side of the Mississippi River basin, there is increased precipitation both in the model and observation estimates at 0000–0600 UTC and 0600–1200 UTC while both depict decreased precipitation at 1200–1800 UTC; the period 1800–2400 UTC shows disagreement, with decreased precipitation in observations while either no change or increase of precipitation is observed in the model estimates. On the eastern side of the basin, agreement is observed for 0000–0600 UTC and 0600–1200 UTC (decreased precipitation) and 1800–2400 UTC (increased precipitation); the 6 h between 1200 and 1800 UTC do not have a clear identifiable sign in the observations.

4. Moisture budgets

The computation of vertically integrated moisture flux convergence is probably one of the most delicate aspects when estimating moisture budgets. Much of the problem appears related to the introduction of errors in the postprocessing stage of the analysis (Trenberth 1991), in particular when variables are interpolated from the model’s grid to a different grid where the diagnostics are performed. According to Trenberth (1991) and Ehrendorfer et al. (1994) the vertical interpolation to pressure levels appears to be the primary source of noise in the divergence that can alter radically the mass field. Apparently, the errors due to vertical interpolation are largest over mountain areas (Alexander and Schubert 1990). Trenberth asserts that the budget errors tend to be systematic, so that anomalies would be more reliable than the total fields. While there are ways to reduce these errors by slightly modifying the winds so that they meet conservation properties (see, e.g., Ehrendorfer et al. 1994), an alternative is to work directly with the analyses in the model’s original grid, thus avoiding the errors introduced by the interpolation schemes. We have followed the latter alternative, analyzing the moisture fields of the EDAS analyses in the Eta Model’s original grid and performing all computations with schemes that are consistent with those used in the model.

a. Mean seasonal fields

The seasonal mean moisture flux and its convergence, as estimated from the regional EDAS analyses, are illustrated, respectively, in Figs. 7 and 8. Due to the high density of the information available that would clutter the figures, and for display purposes only, moisture flux is depicted on a coarser grid than that of the model. The moisture flux during spring (Fig. 7a) is related to the Great Plains LLJ (see Helfand and Schubert 1995 and Higgins et al. 1997 for further descriptions) that most often occurs during the warm months. During this season moisture flux convergence (Fig. 8a) is largest over the Mississippi River basin, with two maxima up to 4 mm day−1. Another maximum is present near the Gulf of Mexico coastal areas, associated with the moisture flux from the Gulf into the continent. It can also be noted that some small-scale spurious centers appear to be related to topography effects in the western boundary of the Missouri River basin. However, the centers are more localized than for global models (Rasmusson and Mo 1996) and their values do not appear to affect the basin estimates. Another region that appears to have topographically related errors is observed near the Sierra Nevada in California.

A significant change is observed in the structure of the moisture flux and its convergence during summer (Figs. 7b, 8b). The northward flux from the Gulf of Mexico (Fig. 7b) into the continent has become more localized in longitude east of the Rockies and continues to be an important source of moisture for the Great Plains. Although the input from the Gulf is still large as in spring, now an outflow has developed in the northern boundary of the basin, mostly due to eastward flux. The net result is that now divergence of moisture flux dominates in most of the Mississippi River basin, a fact that only occurs during summer.

Autumn and winter (Figs. 7c,d) are characterized by a decrease in the intensity of the low-level jet that is replaced by eddy transports (as shown in Berbery et al. 1996) associated with the southward displacement of the upper-level jet stream. At the same time, an increase of westerly and northwesterly fluxes is observed over the Mississippi River basin. Berbery et al. (1996) showed that at 30°N, near the coast of the Gulf of Mexico, the meridional stationary and transient fluxes have an opposite summer to winter evolution, so that while the stationary component decays, the transient component increases becoming larger than the stationary component. Moisture flux convergence (Figs. 8c,d) depicts a clear pattern of divergence over the Atlantic Ocean and Gulf of Mexico, and convergence over land areas particularly toward the east. The contrast is larger during winter, when values are between −2 and +4 mm day−1 over land areas and are as high as −6 mm day−1 over water.

From Eqs. (1) or (2) evaporation can be estimated as a residual between observed precipitation, moisture flux convergence computed from EDAS analyses, and the local changes of atmospheric water vapor content (also called precipitable water, and computed from EDAS analyses too). The local changes of atmospheric water vapor content are small as noted in Fig. 10 and in the literature (Rasmusson 1967; Roads et al. 1994; Ropelewski and Yarosh 1998) and will not be discussed here although they are included in the computations. Although observed precipitation was available only over the continental United States, it was preferred over the model forecast precipitation to reduce the impact of the model’s biases in the moisture budgets. Thus, in this study evaporation estimates derived as a residual from the moisture budget equation are limited to the same region. Recent model-based moisture budget studies have started including an extra term, sometimes called analysis increment, that completes the model balance (Kanamitsu and Saha 1996; Roads et al. 1998). This term compensates other uncertainties due to the model biases when moisture flux convergence, precipitation, and evaporation are computed from the model. Since in this study observed precipitation is used, the analysis increments have not been included as they might obscure the physical balance of the equation. The model’s moisture budget is not discussed in this article for the reasons explained in section 2b: Changes to the land surface schemes in different occasions may have affected some parameterized variables and particularly evaporation, so that no stable results would be obtained.

During spring (Fig. 9a) evaporation values derived from the balance computations range between 2 and 3 mm day−1 with the larger values mostly to the south and in particular over the Arkansas–Red River and Ohio River basins. The remainder of the Mississippi River basin has values typically between 0 and 2 mm day−1. Evaporation is maximum during summer (Fig. 9b), with values as high as 7 mm day−1 over the Arkansas–Red River basin and along the coast of the Gulf of Mexico. In the rest of the Mississippi basin, values range between 2 and 6 mm day−1.

Evaporation would be expected to decrease during autumn because of dying vegetation (Betts et al. 1996), diminished soil moisture, and less radiative heating. Figure 9c shows that evaporation has indeed decreased, although, not unexpectedly, large values of evaporation persist in the southern parts of the basin. During winter (Fig. 9d) the derived evaporation field within the Mississippi River basin is small and relatively uniform, with values generally between 0 and 1 mm day−1. It has been documented (La Rue and Yonkin 1963; Groisman and Legates 1994) that observed precipitation is underestimated by significant amounts during winter due to the combined effect of wind and snowfall. The consequence for computed cold season moisture budgets is that the negative bias in observed precipitation may lead to underestimation of evaporation with negative values appearing in some regions. This appears to be the case in the higher-latitude basins, where slightly negative values are computed.

Groisman et al. (1996) developed a set of correction terms for the precipitation field to compensate for rainfall undercatchment. Mean monthly correction terms during winter estimated from a 20-yr time series (1973–92) range from 1.0 (no correction) in the southern United States to about 1.45 near the Great Lakes. In tests with the correction coefficients that we performed, the negative values of evaporation were reduced but not eliminated. A possible explanation is that the factors employed are long-term mean corrections and may not be sufficient in particular winters, as was the case for 1995–96 and 1996–97, when precipitation was above normal over the northern tier of the United States [e.g., South Dakota had a precipitation excess of about 150%–200%; Climate Prediction Center (1996a), (1997)].

Larger, but more confined, centers of negative evaporation are also seen over the Rockies during all seasons, particularly during autumn and winter. Also in winter, a center of about −1 mm day−1 is noted east of the Appalachian Mountains. These topography-related problems may be the result of inaccurate model products and uncertainties in observed precipitation.

b. Annual cycle

The mean annual cycle of the individual terms of the basin-averaged moisture budget is presented in Fig. 10. The most striking feature is the consistency between the independent estimates of moisture flux divergence and observed precipitation, which depict an opposite evolution in all basins. For example, the peak in precipitation during May in all basins is consistent with a peak in moisture flux convergence (negative divergence) and a similar behavior is observed for other months as well. The resulting evaporation, derived as the difference between large terms, shows a smoother evolution than the balancing terms, increasing our confidence in the validity of the computations. Note that the local change in water content is small for all the basins, and does not play a significant role in the climatological budget.

As expected, evaporation estimated as a residual from the moisture budget in the Mississippi River basin (Fig. 10a) is a maximum during the summer months, with area-averaged values close to 4 mm day−1. During the cold season values range between 0.2 and 0.5 mm day−1, with marked changes in the transition months (May to June and September to October). The Arkansas–Red River basin (Fig. 10e) shows the largest evaporation of all basins during summer with area-averaged values of about 5–6 mm day−1. All subbasin terms have similar annual evolution as the larger basin, indicating physical consistency even at these spatial scales. Both precipitation and evaporation display a two-peak warm season maximum (in June and September).

Slightly negative evaporation is obtained during December and January for the Missouri River basin and its subbasin, the Lower Missouri (Figs. 10c,d). However, the negative values are comparatively small, about −0.2 mm day−1. Not surprisingly, the areas most affected are those to the north, in agreement with the biases in observed precipitation found by Groisman and Legates (1994). In support is the fact that negative evaporation in the Lower Missouri River basin is smaller in magnitude than that of the whole Missouri River basin.

The Upper Mississippi River basin (Fig. 10b) depicts a longer transition from winter to summer months, with evaporation increasing steadily from March through July. Negative evaporation of −0.6 mm day−1 is obtained during March. The larger value (compared to other basins) is consistent with the explanation presented for the Missouri River basin, since the Upper Mississippi is probably the most affected by undercatchment due to its northern position. Finally, the Ohio River basin (Fig. 10f) depicts precipitation peaking in May and evaporation in August. The transition from winter to summer in evaporation is similar to that observed in the Upper Mississippi River. The Ohio River basin is the only one that has convergence of moisture flux in the June–July–August average despite the divergence maximum in August (all other basins have divergence).

Gutowski et al. (1997) used NCEP reanalysis to evaluate the hydrologic cycle of the Upper Mississippi and Ohio River basins and computed an annual mean moisture flux convergence for the first basin of about 30.5 mm month−1, about 40% larger than stream discharge from the basin. In the second basin, computed moisture flux convergence was around 27.1 mm month−1, which is about 33% less than stream discharge. The results presented here (see also Table 1) for the Eta Model show smaller moisture flux convergence for the Upper Mississippi River basin (12.2 mm month−1) and larger convergence for the Ohio River basin (63.4 mm month−1) and are thus in closer agreement with the observed stream discharge. The latter is significantly larger than Gutowski et al.’s, but at the same time is consistent with the heavy precipitation observed during the springs of 1996 and 1997. According to Climate Prediction Center (1996b) during some of those spring months precipitation was twice the normal and produced floods in the Ohio River basin.

c. Effects of diurnal variability

As it will be shown in section 5, all moisture budget terms have a diurnal variability that, if not properly resolved, may produce biases affecting the balance among terms. The most sensitive term is the convergence of moisture flux that is closely related to the diurnal cycle of the low-level jet in the Great Plains.

To evaluate the biases that can arise from inadequate temporal sampling, moisture flux convergence, computed using four and two analyses per day, were compared with estimates from eight analyses per day (Fig. 11). The estimates from four analyses per day are relevant because global analyses are available with that frequency. Estimates from two analyses per day are needed because some older global analyses are available with that frequency, but also because moisture budgets estimated from radiosondes, which are launched twice daily, serve as a reference (see, e.g., Yarosh et al. 1996).

Figure 11 presents the moisture flux convergence as computed from the eight analyses per day (solid line). The dotted line shows the estimate using four analyses per day and the dashed line the estimate using two analyses per day. To simplify the comparison, Fig. 12 is added showing the differences resulting from using four analyses per day instead of eight (dashed line) and the differences when two analyses are used (solid line). According to Figs. 11a and 12a, the averages for the entire Mississippi River basin are not affected significantly when using four analyses per day, but for two analyses per day, differences of the order of 0.1–0.3 mm day−1 appear throughout the year. As expected, important deviations are observed in the smaller basins. Differences of the order of 0.5–1 mm day−1 are common in most basins when comparing the estimate from eight analyses per day with those from four and two analyses per day. Surprisingly, the Arkansas–Red River basin (Figs. 11e, 12e) is not the most affected basin, although it is in the path of the LLJ. The Upper Mississippi River basin (Figs. 11b, 12b) and the Ohio River basin (Figs. 11f, 12f) are the most affected basins when using two analyses per day, with deviations occasionally exceeding 1 mm day−1. In the latter basin, even four analyses per day produce estimates with biases of about 0.3 mm day−1.

Table 1 presents the annual mean moisture flux convergence for each basin as estimated from the eight, four, and two analyses per day. The results suggest that annual means are not significantly affected by the time resolution, although differences are about 6–8 mm month−1. A similar table showing the summer averages (Table 2) indicates that the basin most affected by time resolution is the Ohio River basin. In the Arkansas–Red, differences are small although according to Fig. 12 differences exist in the transition months during autumn. The largest errors in sampling are observed at individual months and thus may be relevant for studies of intermonthly variability.

d. Effects of spatial resolution

When moisture flux convergence is computed as the net flux across the perimeters of the basins, spatial resolution may become an important issue. The high spatial resolution of the Eta Model can be used to replicate coarser grids, and thus it is possible to provide an estimate of the errors arising from computing moisture flux convergence at lower resolutions. Sets of 2.5° × 2.5° grids3 were constructed and moisture flux convergence was computed for each. This is not an experiment running the model at lower resolution that would have other implications, but is only the analysis of sampling errors. Figure 13 presents the moisture flux convergence as estimated from the full resolution (solid line) and from the set of 2.5° × 2.5° grids (dotted lines). The dashed line is the average resulting from all estimates in the coarser grids. A first point to note is that the average curve does not agree with that from the full resolution estimate, suggesting that the errors are not random. The larger basins, Mississippi (Fig. 13a) and Missouri (Fig. 13c), have the smallest biases, but nevertheless during summer months, average differences are about 0.2–0.3 mm day−1. Moreover, the indicated biases are sometimes of opposite sign, depending on the basin being analyzed. The most notable cases are those of the Arkansas–Red River and the Upper Mississippi River basins where smaller estimates of moisture flux convergence are obtained from coarser resolutions (Figs. 13e,b) while the opposite is true for the Ohio River basin (Fig. 13f). It is presumed that the reason for this opposite behavior may be related to the typical distribution of moisture fluxes on the perimeter of each of these basins. Both Arkansas–Red and Upper Mississippi River basins are in the path of the LLJ during summer and replaced by transients in winter; undersampling might result then in a lower estimate of the moisture influx and thus of the moisture flux convergence. On the other hand, the Ohio River basin has an important moisture outflow (as noted from Fig. 8); in this case, coarser grids might underestimate the outgoing moisture flux and therefore moisture flux convergence would be exaggerated.

Table 3 presents the annual average of moisture flux convergence as estimated from the full resolution and as the average of estimates from 2.5° × 2.5° grids. Differences between estimates are larger than those related to time sampling discussed in the previous section. Most affected are the Arkansas–Red River basin where moisture flux divergence is doubled when reducing the resolution, and the Ohio River basin whose moisture flux convergence is increased by about 50% when the resolution is reduced. Similar tests with 1° × 1° grids (not shown) reveal that differences, although smaller, may still exceed 1 mm day−1 in individual months in the smaller basins.

5. Summertime diurnal variability of the moisture flux convergence

It was shown in section 3c that the broadscale features of the diurnal cycle of precipitation are reproduced by the Eta Model. The impact of the diurnal variability of moisture flux convergence on the regional estimates of the atmospheric water balance was discussed in section 4c. In fact, all terms in the water balance over the Great Plains have a diurnal cycle, and several studies have related the nighttime maximum of moisture inflow due to the LLJ with a nocturnal maximum in storm activity and precipitation (see, e.g., Wallace 1975). Helfand and Schubert (1995) using an atmospheric general circulation model reproduced the Great Plains LLJ in their springtime simulations and highlighted its marked diurnal cycle. They found that associated with the nighttime increase of the LLJ, the wind at lower levels (at the model level σ = 0.97) depicts a pattern of divergence over the Gulf of Mexico and convergence over the Great Plains (their Fig. 8). Higgins et al. (1997) prepared composites of LLJ cases from NCEP reanalyses and also noted at the model’s σ = 0.98 level a weak nighttime wind convergence over the Great Plains (their Fig. 8).

The pronounced mesoscale nature of the LLJ and its marked diurnal variability revealed by EDAS are illustrated in Fig. 14. Figure 14a shows a cross section along 30°N of the meridional component of moisture flux at approximately local midnight (0600 UTC). Most of the northward flux is found east of the Rockies below 700 hPa, in a longitudinal band about 10°–15° in width. A southward low-level jet off the coast of California near 120°W related to the eastern boundary of the Pacific Ocean anticyclone (Helfand and Schubert 1995) also appears in the analysis. The nighttime spatial pattern at 900 hPa (Fig. 14b) reveals two northward branches: one branch is the one already discussed that lies east of the Rockies and curves eastward over the Great Lakes and southern Canada; the second appears over the western Atlantic. The moisture flux between the two branches over the eastern United States is small and consistent with Higgins et al.’s (1997) results suggesting a suppression of precipitation over the eastern region during LLJ events.

At 1800 UTC the intensity of the moisture flux east of the Rockies decays to almost one-half its maximum value (Fig. 14c). Changes continue to be confined to a narrow band near the Rockies and suggest a height decrease of the maximum wind during daytime. The marked decrease in moisture flux intensity confirms the results of Berbery et al. (1996) who found a night to day decay of the same order of magnitude for August 1993. The spatial distribution of the moisture flux difference between 0600 and 1800 UTC is depicted in Fig. 14d. The region that exhibits the largest changes in moisture flux is found along the east of the Rockies, with maximum values over Texas. On the other hand, the moisture flux branch over the western Atlantic does not show signs of diurnal variability.

Helfand and Schubert (1995) and Higgins et al. (1997) discussed the diurnal cycle of wind divergence at a lower level representing the boundary layer. Here we present the patterns of vertically integrated moisture flux convergence that have been averaged over three analysis times to better represent nighttime and daytime mesoscale characteristics of the field (Fig. 15). It was shown in Figs. 8 and 10 that during summer vertically integrated moisture flux divergence predominates over the Mississippi River basin. However, as illustrated in Fig 15a, during nighttime the LLJ has associated convergence that achieves a maximum over the state of Arkansas and the central and southern part of the Mississippi River basin.

On the other hand, the daytime pattern (Fig. 15b) shows that the area of convergence shrinks leaving flux divergence as the dominant pattern. These changes are better reflected in the nighttime–daytime difference (Fig. 15c). Different features are observed over each basin; the Arkansas–Red has an east–west seesaw pattern with nighttime moisture flux convergence in the east and divergence to the west, and the reverse pattern during daytime. It would appear that other areas do not have a well-defined diurnal cycle, but this is the result of presenting vertically integrated variables. In fact, all of the basins exhibit distinct diurnal changes in the vertical profiles of moisture flux convergence. A detailed description of the vertical structure and how it relates to physical aspects of the boundary layer is being currently analyzed. Results will be reported elsewhere.

6. Summary and conclusions

Eta Model products (analyses and forecasts) have been used to investigate the atmospheric water balance at regional scales for the Mississippi River basin. A comparative analysis of the Eta Model 12–36-h forecast precipitation, and observed precipitation from a network of about 2600 stations, was performed to evaluate the performance of the model. The results showed close correspondence, but there was a dry model bias over the central United States during both summer and winter. Model estimates of precipitation for the Mississippi River basin are closest to observations during winter. Outside the Mississippi River basin, summertime forecast precipitation over the southeastern United States in general, and Florida in particular, is excessive. This bias is similar to that observed on the NCEP global model (Higgins et al. 1996). Forecast precipitation over the northwestern United States in winter exhibits biases in location and intensity that can be related to the large-scale component of the model precipitation.

A comparison of our moisture budget estimates and those computed directly from the model products was avoided, because the latter were not stable enough due to changes in physical schemes in the model.

The main findings of this study can be summarized as follows:

  • The results strongly suggest that the Eta Model products are a valuable tool for regional climate studies that require higher resolution than that provided by current global models. Additionally, it has been shown that the spatial and temporal resolution of the Eta Model analysis products provide reliable estimates of moisture budgets for spatial scales that are smaller than those possible from radiosondes or current analyses provided by global models. More specifically, it was found that moisture budgets estimated for basins of the order of 5 × 105 km2 from EDAS analyses appear to have similar internal consistency as those computed for basins of the order of 2 × 106 km2 from radiosonde observations alone (as in Rasmusson 1968).
  • Biases were found when estimating moisture budgets at regional scales when the frequency of analyses was reduced from eight times daily to four or two analyses per day. Similarly, biases were produced when computing moisture flux convergence at coarser grids. In both cases, the results are not systematic in the sense that they vary from basin to basin.

The diurnal cycle during summer was examined in the context of nighttime–daytime differences. Broad aspects of the observed diurnal cycle of precipitation were reproduced in the model forecasts although the transition zone in the latter is shifted to the west over the central United States. Consistent with other studies over the central United States, the nighttime development of moisture flux convergence is associated with an increase of intensity of the LLJ. Interestingly, the nighttime convergence of moisture flux is offset with divergence during daytime and, as a result, net moisture flux divergence is observed during summer (i.e., the continent acts as a moisture source for the atmosphere during summer months).

Part of the improvement in the moisture budget results presented here is due to the high temporal and spatial resolution of the regional analyses. There are indications that the higher resolution also allows a better definition of the land surface fields (like vegetation) that may have a positive impact on the model performance. Not surprisingly, some problems remain. Estimates of moisture flux convergence, and consequently evaporation derived as a residual, exhibit spurious small-scale centers near major terrain features. However, the centers are considerably more localized than those previously derived from global analyses (Rasmusson and Mo 1996) and have a minor effect on basin averages on scales examined in this study. Slightly negative evaporation during the cold season was derived for some of the northern basins. This bias can be attributed, at least partially, to underestimation of the observed precipitation.

Acknowledgments

Support by Ken Mitchell, Eric Rogers, and the Environmental Modeling Center (NCEP/EMC) is deeply appreciated. We thank Pavel Groisman for facilitating the correction terms for winter precipitation and Evgeney Yarosh for making them available in an adequate format. Comments by Roni Avissar are also appreciated. The comments of three anonymous reviewers helped clarify several aspects of the paper. This work was partially supported by NOAA Grants NA76GP0291 and NA76GP0479.

REFERENCES

  • Alexander, M. A., and S. D. Schubert, 1990: Regional earth–atmosphere energy balance estimates based on assimilations with a GCM. J. Climate,3, 5–14.

  • Berbery, E. H., E. M. Rasmusson, and K. E. Mitchell, 1996: Studies of North American continental scale hydrology using Eta Model forecast products. J. Geophys. Res.,101 (D3), 7305–7319.

  • Betts, A. K., J. H. Ball, A. C. J. Beljaars, M. J. Miller, and P. Viterbo, 1996: The land surface–atmosphere interaction: A review based on observational and global modeling perspectives. J. Geophys. Res.,101 (D3), 7209–7225.

  • ——, F. Chen, K. E. Mitchell, and Z. I. Janjić, 1997: Assessment of the land surface and boundary layer models in the two operational versions of the NCEP Eta Model using FIFE data. Mon. Wea. Rev.,125, 2896–2916.

  • ——, P. Viterbo, A. Beljaars, H.-L. Pan, S.-H. Hong, M. Goulden, and S. Wofsy, 1998a: Evaluation of the land–surface interaction in the ECMWF and NCEP/NCAR reanalysis models over grassland (FIFE) and Boreal forest (BOREAS). J. Geophys. Res.,103, 23 079–23 085.

  • ——, ——, and E. Wood, 1998b: Surface energy and water balance for the Arkansas–Red River basin from the ECMWF reanalysis. J. Climate,11, 2881–2897.

  • Black, T. L., 1994: The new NMC mesoscale Eta Model: Description and forecast examples. Wea. Forecasting,9, 265–278.

  • Chen, F., K. Mitchell, and J. Schaake, 1996: Modeling of land surface evaporation by four schemes and comparison with FIFE observations. J. Geophys. Res.,101 (D3), 7251–7268.

  • Climate Prediction Center, 1996a: Winter weather review. Weekly Weather Crop Bulletin, Vol. 83, No. 11, 24 pp. [Available from Climate Prediction Center, W/NP52, Attn: Weekly Weather and Crop Bulletin, NOAA/NWS/NCEP, 4700 Silver Hill Road, Stop 9910, Washington, DC 20233.].

  • ——, 1996b: Spring weather review. Weekly Weather Crop Bulletin, Vol. 83, No. 23, 28 pp. [Available from Climate Prediction Center, W/NP52, Attn: Weekly Weather and Crop Bulletin, NOAA/NWS/NCEP, 4700 Silver Hill Road, Stop 9910, Washington, DC 20233.].

  • ——, 1997: Winter weather review. Weekly Weather Crop Bulletin, Vol. 84, No. 10, 16 pp. [Available from Climate Prediction Center, W/NP52, Attn: Weekly Weather and Crop Bulletin, NOAA/NWS/NCEP, 4700 Silver Hill Road, Stop 9910, Washington, DC 20233.].

  • Coughlan, M., and R. Avissar, 1996: The Global Energy and Water Cycle Experiment (GEWEX) Continental-scale International Project (GCIP): An overview. J. Geophys. Res.,101 (D3), 7139–7147.

  • Dunn, L. B., and J. D. Horel, 1994: Prediction of central Arizona convection. Part II: Further examination of the Eta Model forecasts. Wea. Forecasting,9, 508–521.

  • Ehrendorfer, M., M. Hantel, and Y. Wang, 1994: A variational modification algorithm for three-dimensional mass flux non-divergence. Quart. J. Roy. Meteor. Soc.,120, 655–698.

  • Giorgi, F., S. W. Hosteler, and C. Shields Brodeur, 1994: Analysis of the surface hydrology in a regional climate model. Quart. J. Roy. Meteor. Soc.,120, 161–183.

  • Groisman, P. Ya., and D. R. Legates, 1994: The accuracy of the United States precipitation data. Bull. Amer. Meteor. Soc.,75, 215–227.

  • ——, D. R. Easterling, R. G. Quayle, V. S. Golubev, A. N. Krenke, and A. Yu. Mikhailov, 1996: Reducing biases in estimates of precipitation over the United States: Phase 3 adjustments. J. Geophys. Res.,101 (D3), 7185–7195.

  • Gutowski, W. J., Y. Chen, and Z. Ötles, 1997: Atmospheric water vapor transport in NCEP–NCAR reanalyses: Comparison with river discharge in the central United States. Bull. Amer. Meteor. Soc.,78, 1957–1969.

  • Helfand, H. M., and S. D. Schubert, 1995: Climatology of the simulated Great Plains low-level jet and its contribution to the continental moisture budget of the United States. J. Climate,8, 784–806.

  • Higgins, R. W., K. C. Mo, and S. D. Schubert, 1996: The moisture budget of the central United States in spring as evaluated in the NCEP/NCAR and the NASA/DAO reanalyses. Mon. Wea. Rev.,124, 939–963.

  • ——, Y. Yao, E. S. Yarosh, J. E. Janowiak, and K. C. Mo, 1997: Influence of the Great Plains low-level jet on summertime precipitation and moisture transport over the central United States. J. Climate,10, 481–507.

  • International GEWEX Project Office, 1995: Major activities plan for 1996, 1997, and outlook for 1998 for GCIP. IGPO Publ. Series 16, Int. GEWEX Project Office, Washington, DC, 246 pp. [Available from 1010 Wayne Avenue, Suite 450, Silver Springs, MD 20910.].

  • Janjić, Z. I., 1990: The step-mountain coordinate: Physical package. Mon. Wea. Rev.,118, 1429–1443.

  • ——, 1994: The step-mountain Eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev.,122, 927–945.

  • Kanamitsu, M., and S. Saha, 1996: Systematic tendency error in budget calculations. Mon. Wea. Rev.,124, 1145–1160.

  • La Rue, J. A., and R. J. Younkin, 1963: Large-scale precipitation volumes, gradients and distribution. Mon. Wea. Rev.,91, 393–401.

  • Lin, Y., K. E. Mitchell, E. Rogers, and G. J. DiMego, 1995: Impact of satellite moisture observations on forecasts made by NMC’s Eta model. Preprints, 14th Conf. on Weather Analysis and Forecasting, Dallas, TX, Amer. Meteor. Soc., 301–302.

  • Mesinger, F., 1996: Improvements in quantitative precipitation forecasts with the Eta regional model at the National Centers for Environmental Prediction: The 48-km upgrade. Bull. Amer. Meteor. Soc.,77, 2637–2649.

  • ——, and T. L. Black, 1992: On the impact on forecast accuracy of the step-mountain (Eta) vs. sigma coordinate. Meteor. Atmos. Phys.,50, 47–60.

  • ——, Z. I. Janjić, S. Ničković, D. Gavrilov, and D. G. Deaven, 1988:The step-mountain coordinate: Model description and performance for cases of alpine lee cyclogenesis and for a case of Appalachian redevelopment. Mon. Wea. Rev.,116, 1493–1518.

  • Peixoto, J. P., and A. H. Oort, 1992: Physics of Climate. American Institute of Physics, 520 pp.

  • Rasmusson, E. M., 1967: Atmospheric water vapor transport and the water balance of North America. Part I: Characteristics of the water vapor flux field. Mon. Wea. Rev.,95, 403–426.

  • ——, 1968: Atmospheric water vapor transport and the water balance of North America. Part II: Large-scale water balance investigations. Mon. Wea. Rev.,96, 720–734.

  • ——, and K. C. Mo, 1996: Large-scale atmospheric moisture cycling as evaluated from global NMC analysis and forecast products. J. Climate,9, 3276–3297.

  • Roads, J., S. Chen, A. Guetter, and K. Georgakakos, 1994: Large-scale aspects of the United States hydrologic cycle. Bull. Amer. Meteor. Soc.,75, 1589–1610.

  • ——, ——, M. Kanamitsu, and H. Juang, 1998: Vertical structure of humidity and temperature budget residuals over the Mississippi River basin. J. Geophys. Res.,103 (D4), 3741–3759.

  • Rogers, E., T. L. Black, D. G. Deaven, and G. J. DiMego, 1996: Changes to the operational early Eta analysis/forecast system at the National Centers for Environmental Prediction. Wea. Forecasting,11, 391–413.

  • Ropelewski, C. F., and E. S. Yarosh, 1998: The observed mean annual cycle of moisture budgets over the central United States (1973–92). J. Climate,11, 2180–2190.

  • Rosen, R. D., D. A. Salstein, and J. P. Peixoto, 1979: Variability in the annual fields of large-scale atmospheric water vapor transport. Mon. Wea. Rev.,107, 26–37.

  • Trenberth, K. E., 1991: Climate diagnostics from global analyses: Conservation of mass in ECMWF analyses. J. Climate,4, 707–722.

  • ——, and C. J. Guillemot, 1995: Evaluation of the global atmospheric moisture budget as seen from analyses. J. Climate,8, 2255–2272.

  • Wallace, J. M., 1975: Diurnal variations in precipitation and thunderstorm frequency over the conterminous United States. Mon. Wea. Rev.,103, 406–419.

  • Wang, M., and J. Paegle, 1996: Impact of analysis uncertainty upon regional atmospheric moisture flux. J. Geophys. Res.,101 (D3), 7291–7303.

  • Yarosh, E. S., C. F. Ropelewski, and K. E. Mitchell, 1996: Comparisons of humidity observations and Eta Model analyses and forecasts for water balance studies. J. Geophys. Res.,101 (D18), 23 289–23 298.

  • Zhao, Q., T. L. Black, and M. E. Baldwin, 1997: Implementation of the cloud prediction scheme in the Eta Model at NCEP. Wea. Forecasting,12, 697–712.

Fig. 1.
Fig. 1.

(a) Mississippi River basin and subbasins corresponding to GCIP large-scale areas, as represented by the Eta Model grid points. (b) Distribution of cooperative stations reporting hourly precipitation within the Mississippi River basin.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 2.
Fig. 2.

(a) Observed precipitation during summer. (b) Model forecast precipitation during summer. (c) Difference between model forecast and observed precipitation. (d)–(f) Same as (a)–(c) but for winter. Contour intervals for (a), (b), (d), and (e) are shown in the panels and units are mm month−1. Contour intervals in (c) and (f) are uniform and equal to 25 mm month−1; the zero contour was omitted.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 3.
Fig. 3.

Scatter diagrams of seasonal mean model forecast precipitation vs seasonal mean observed precipitation for all gauges within the Mississippi River basin: (a) summer, (b) winter. Units are mm month−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 4.
Fig. 4.

May 1995–Apr 1997 time series of basin-averaged observed precipitation (solid line) and model forecast precipitation (dashed line). Units are mm day−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 5.
Fig. 5.

Diurnal cycle of basin-averaged observed precipitation during summer. Units are mm h−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 6.
Fig. 6.

Summertime nighttime (0000–1200 UTC) precipitation minus daytime (1200–2400 UTC) precipitation for (a) observations and (b) model forecasts. Contour interval is 0.5 mm (12 h)−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 7.
Fig. 7.

Vertically integrated moisture flux as estimated from EDAS for (a) spring, (b) summer, (c) autumn, and (d) winter. The lower-right arrow represents a moisture flux of 400 kg (m s)−1. Vectors with magnitude smaller than 50 kg (m s)−1 are not displayed. Only one of every three grid points are shown to avoid cluttering.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 8.
Fig. 8.

Vertically integrated moisture flux convergence as estimated from EDAS for (a) spring, (b) summer, (c) autumn, and (d) winter. Contour interval is 2 mm day−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 9.
Fig. 9.

Evaporation estimated as a residue of the moisture budget equation for (a) spring, (b) summer, (c) autumn, and (d) winter. Contour interval is 1 mm day−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 10.
Fig. 10.

Annual cycle of the basin-averaged moisture budget terms: observed precipitation (short dashes with solid squares), moisture flux divergence (long dashes with solid circles), local changes of atmospheric water vapor content (thin solid line with open squares), and evaporation (thick solid line with open circles). Units are mm day−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 11.
Fig. 11.

Basin-averaged moisture flux convergence as estimated from eight times daily analyses (solid lines), four times daily analyses (0000, 0600, 1200, and 1800 UTC; short dashed lines), and twice daily analyses (0000 and 1200 UTC; long dashed lines). Units are mm day−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 12.
Fig. 12.

Basin-averaged differences of moisture flux convergence between the eight times daily analysis estimates and the four times daily analysis estimates (dashed lines) and differences between the eight times daily analysis estimates and the twice daily analysis estimates (solid lines). Units are mm day−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 13.
Fig. 13.

Basin-averaged moisture flux convergence as estimated from EDAS (solid lines) and from a set of 2.5° × 2.5° grids (derived from the Eta Model full resolution) displaced in space (dotted lines). The average of all estimates from the 2.5° × 2.5° grids is presented as a heavy long dashed line. Units are mm day−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 14.
Fig. 14.

(a) Cross section at 30°N of the 0600 UTC meridional component of moisture flux during summer. Contour interval is 20 g kg−1 m s−1. (b) The 900-hPa moisture flux at 0600 UTC. The lower-right arrow indicates a moisture flux of 200 g kg−1 m s−1. Vectors of magnitude smaller than 30 g kg−1 m s−1 are not displayed; (c) and (d) same as (a) and (b) but for the moisture flux difference between 0600 and 1800 UTC.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Fig. 15.
Fig. 15.

Vertically integrated moisture flux convergence at (a) nighttime (average of 0600, 0900, and 1200 UTC) and (b) daytime (average of 1800, 2100, 0000 UTC). (c) Difference (0600, 0900, and 1200 UTC) − (1800, 2100, and 0000 UTC) moisture flux convergence. Contour interval is 0.2 mm h−1.

Citation: Monthly Weather Review 127, 11; 10.1175/1520-0493(1999)127<2654:MMBORS>2.0.CO;2

Table 1.

Mean annual moisture flux convergence. Temporal resolution.

Table 1.
Table 2.

Mean summer moisture flux convergence. Temporal resolution.

Table 2.
Table 3.

Mean annual moisture flux convergence. Spatial resolution.

Table 3.

1

Because we are using model products from an operational environment, it is not possible to provide a formal reference for some of the model’s changes. Updates are posted at NCEP’s Environmental Modeling Center Web site (http://nic.fb4.noaa.gov:8000/).

2

Since the description of the diurnal cycle focuses on the central United States, “local time” is taken to be the central daylight saving time.

3

Strictly, these are not 2.5° × 2.5° grids, but grids resulting from selecting one of every five Eta Model grid points in each direction. Thus, the additional effects of interpolating to a regular latitude–longitude grid were avoided.

Save