The purpose of this paper is to review the published objectives of ACSYS, particularly with respect to hydrology, and to suggest how APDA can best contribute to ACSYS, particularly with respect to ACSYS hydrologic objectives. I will also indicate what I know of how ACSYS objectives relative to land surface hydrology have evolved. Finally, two examples are discussed that suggest how the data APDA might make available to the ACSYS community could contribute to meeting ACSYS hydrological objectives.

1.0 ACSYS OBJECTIVES

The ACSYS objectives, per the September, 1994 Initial Implementation Plan (IIP), are:

I. To understand the interactions between the Arctic Ocean circulation, ice cover, and the hydrological cycle;

II. To initiate long-term, climate research and monitoring programs for the Arctic;

III. To provide a scientific basis for an accurate representation of Arctic processes in global climate models.

To address these objectives, the IIP identifies five ACSYS subprograms. The first two (Arctic Ocean Circulation Programme and Arctic Sea Ice Programme) are primarily oceanographic in nature. The third is the Arctic Atmosphere Programme.

The ACSYS subprogram to which APDA is most relevant, is titled the Hydrological Cycle of the Arctic Region. It consists of two tasks. The first is to assemble a regional data base for all components of the freshwater balance over the Arctic basin, "including the Arctic Ocean basin and the catchments of Arctic Rivers". This first task in turn consists of three elements:

a) ACSYS Solid Precipitation Climatology Project -- to define the space-time distribution of precipitation over the Arctic Basin, including the oceans. This element, among other things, focuses on problems in estimation of solid precipitation, especially wind catch deficiency of conventional gauges;

b) Arctic river discharge data -- to define the climatology and interannual variability of the freshwater influx to the Arctic basin, focusing on the period 1978 to present. This element is being closely coordinated with the Global River Data Center in Koblenz;

c) The Arctic Precipitation Data Archive (APDA), "to provide an observational data base for hydrological modeling". The APDA effort is to be based primarily on station data, with a daily time frequency where possible, and monthly elsewhere. The focus is to be on the period 1978 to present, for consistency with the river runoff data.

The second task within the ACSYS Hydrological Cycle component is to develop hydrological models of selected Arctic River basins and validate against appropriate observational data sets. It in turn consists of two elements:

a) Catchment scale hydrological modeling (essentially, predict streamflow given precipitation and other meteorological variables, at a daily or shorter times scale); and

b) Use of (atmospheric) model-derived precipitation to back out mean areal precipitation given moisture flux divergences. This technique is particularly applicable to estimation of precipitation over large areas (in excess of about 10⁵ km²) of the ocean.

A fifth ACSYS subprogram, the ACSYS Modeling Program, seeks to integrate the other four subprogrammes, and to address directly the ACSYS objective of improving the representation of Arctic processes in global climate models. APDA has an important role to play in this subprogram as well, especially given the inherent sparseness of the archival data. Any contribution APDA can make to improving estimates of the precipitation climatology of the Arctic and its space-time variability should be extremely useful in evaluation of climate model performance in the ACSYS region.

2.0 HYDROLOGY IN ACSYS

The information in this section is based on discussions with Knut Aagaard, Roger Colony, and Victor Savchentko, whose help is appreciated, as their involvement in the project considerably predates mine. The SSG view of the role of hydrology in ACSYS has evolved somewhat since the initiation of the project. Initially, the desire was to provide an estimate of the riverine portion of the freshwater balance of the Arctic basin, with a time scale of at least one year. Subsequently, that view was expanded to include:

a) the interannual variability of the freshwater flux;

b) the role of snow and the spatial variability of the land surface snow cover on albedo and (land surface) energy fluxes;

c) the dynamics of land-atmosphere-ocean interactions at high latitudes (e.g., for eventual incorporation in GCMs and numerical weather prediction models; e.g., in support of ACSYS Objective III).

An ancillary issue is the role of ephemeral accumulations of snow on sea ice. Its contribution to the Arctic basin freshwater balance, and its effect on the sea ice thermal balance, need to be better understood.

3.0 POTENTIAL ROLE OF APDA IN ACSYS HYDROLOGY

There are at least four ways in which APDA can contribute to ACSYS. The first is to help refine the climatology of the Arctic drainage basin, especially those portions that are ungauged. The freshwater balance of the major rivers can be estimated directly from stream gauge information. The estimation problem exists primarily with respect to the 30 percent or so of the Arctic basin land mass, runoff from which is not gauged. Estimation of runoff from these areas will quite likely require a modeling effort, and a key forcing will be precipitation.

The second area is the estimation of ocean precipitation. Direct estimation of the climatology of ocean precipitation, e.g., via drifting ice stations, is highly problematic, due to sparse coverage, nonoverlapping periods of record, and nonstationary locations. For water budget purposes, calculation of large area P-E via a convergence-divergence computation appears to be a preferable approach. This approach is, however, limited to large areas. It depends on the accuracy of estimation of wind and water vapor profiles, especially in the lower atmosphere. Such estimates are a potential contribution of APDA.

The third area in which APDA can support ACSYS is provision of forcing and validation data for testing and evaluation of improved land surface models for the Arctic region. An initial step is off-line model testing (see Section 4.0 for an example), for which precipitation is a key forcing. Validation would be via streamflow and snow (primarily areal extent and water equivalent) data.

The fourth area is snow-albedo feedbacks. It is well known that the albedo of snow is strongly dependent on its microphysical properties, which in turn depend on the age of the surficial snow cover. Probably the best hope for such large areas as the ACSYS region are satellite remote sensing products, however surface data are needed for comparison and validation. As noted above, snow areal extent and water equivalent data also play an important role in hydrological model evaluation.

4.0 EXAMPLE APPLICATIONS

4.1 Grid based hydrological model of the MacKenzie River

As an example of the application of macroscale hydrological models to predict the flow of large northern rivers, selected results from a study of the Mackenzie River basin are summarized from a presentation by Pauwels et al. (1996). In their study, a grid-based version of the 2-layer Variable Infiltration Capacity Model (Nijssen et al., 1996) was applied to the Mackenzie River basin (Figure 1). Precipitation data were from a station data set provided by G.W. Kite of the Canadian National Hydrology Research Institute. Digital terrain data were from a 5 minute global digital elevation model, and soils and vegetation data were from the global one degree ISLSCP data set (Sellers et al., 1995).

Figure 1: Grid-based VIC-2L model schematic.

Figure 2 shows the mean simulated and observed monthly streamflow near the mouth of the MacKenzie River, aggregated from the model's daily time step. The annual volumes are preserved fairly closely, as are the winter flows, but the model overpredicts the spring/summer peak, and the summer recession is underpredicted. Among the suspected reasons for the discrepancies are the absence of a frozen soil algorithm in the present version of the model, and misspecification of the spatial distribution of precipitation, especially snow accumulation at the end of the winter season. The results are encouraging to the extent that they indicate the potential for this type of model to provide estimates of freshwater flows for large Arctic drainages. Eventually, we expect that a similar approach can be used to model the hydrology of the entire Arctic basin.

Figure 2: Predicted and observed mean monthly discharge for Mackenzie River near mouth.

4.2 Gridding of station precipitation data for hydrological modeling

One problem that will inevitably be encountered in ACSYS off-line land surface hydrological modeling efforts, (such as that described above for the MacKenzie River) is estimation of precipitation forcings in data-sparse regions. The first order problem is to preserve the climatological spatial mean precipitation fields. Nijssen et al. (1996), in application of the VIC-2L model to the Columbia River basin, rescaled station data to match the long-term mean precipitation estimated by Daly et al. (1994) using the PRISM (Precipitation-elevation Regressions on Independent Slopes Model). A similar approach is in progress for the MacKenzie River basin, using a topographically-derived spatial mean precipitation product developed by R. Soulis and S. Solomon at the University of Waterloo.

While such rescaling shows promise for assuring consistency of the hydrologic forcings (especially precipitation) with climatology, there remains the problems of serial completeness of station records, and sparseness of stations. This problem, while not unique to the Arctic, is particularly acute. The approach used by the ongoing global soil wetness project is temporal disaggregation of gridded monthly station data (generally more available than the daily time step data required for hydrological model forcings) using global reanalysis fields from numerical weather prediction models. The shortcoming with this approach is that biases in the weather prediction models, typically in the direction of too many wet days, are transmitted to the gridded daily fields. Schnur and Lettenmaier (1995) have developed an alternative three step procedure. Grid cell averages of a daily global precipitation data set archived by the U.S. Climate Analysis Center (CAC) are used to disaggregate monthly gridded values when a minimum station threshold (typically about five stations per two degree grid cell) is met. When fewer stations are present, the parameters of a stochastic daily precipitation model are estimated from each of the daily CAC stations, and the parameters are interpolated. Stochastic sequences of precipitation are then generated using the interpolated parameters, and the stochastic sequences are used to disaggregate the monthly gridded precipitation. Typical Northern Hemisphere results are shown in Figure 3 for January, 1987.

Figure 3: Monthly precipitation for January, 1987 at two degree resolution summed from daily gridded precipitation generated from global CAC station data set after estimating missing values by using a stochastic rainfall model, and after scaling daily sequences to Hulme (1995) gridded monthly precipitation data set

5.0 RECOMMENDATIONS FOR APDA

ACSYS Subprograms 4 (Hydrological cycle of the Arctic region) and 5 (ACSYS Modeling Program) will both be highly dependent on APDA precipitation products. The highest priority for APDA should be to assemble the available archival precipitation data, at daily time frequency where it is available, for the entire ACSYS region. For this purpose, the ACSYS region should be based on a hydrological definition, that is, all of the Arctic region that drains to the Arctic Oceans. Preliminary indications are that the Arctic precipitation archive produced by the Canadian Atmospheric Environment Service should be suitable for ACSYS purposes, assuming that data access issues can be resolved. Precipitation data for the Russian Arctic are much more problematic, and APDA will likely need to focus considerable effort on acquiring and processing the Russian data.

In addition to precipitation, other variables will be required for hydrological purposes, the most important of which is temperature. For most purposes, daily temperature maxima-minima will suffice. In addition to precipitation and temperature, other surface meteorological variables, such as humidity, surface pressure, wind, and solar radiation, are of interest will be required for land surface modeling. It is unlikely that these variables will be available at enough locations to facilitate production of station-based gridded fields. A more promising approach is to utilize gridded weather prediction model analysis or reanalysis fields.

Finally, snow estimates (both areal extent and water equivalent) will be critical for ACSYS modeling. Remote sensing products seem to offer the greatest promise, given the large areas involved. Snow areal extent products based, for instance, on visible and near-infrared channels of AVHRR are reasonably well established. One disadvantage, or course, is lack of availability during cloudy periods. Passive microwave algorithms for snow water equivalent have shown reasonable promise in areas with short vegetation (e.g., tundra), although the algorithms tend to be fairly site specific. Global snow products are already available through the National Snow and Ice Data Center (NSIDC). APDA coordination with NSIDC to determine what, if any, additional Arctic snow data should be made available to the ACSYS community would be highly advisable

REFERENCES:

Daly, C., R.P. Neilson, and D.L. Phillips, 1994, "A statistical topographic model for mapping climatological precipitation over mountainous terrain". Journal of Applied Meteorology 33(2): 140-158.

Hulme, M., 1995, "Estimating global changes in precipitation". Weather, 50, 34-42.

Nijssen, B., D.P. Lettenmaier, X. Liang, S.W. Wetzel, and E.F. Wood, 1996, "Streamflow simulation for continental-scale watersheds". Water Resources Research, in review, 1996.

Pauwels, V., E.F. Wood, and D.P. Lettenmaier, "Large-scale hydrological modeling over high latitude basins". poster paper presented at Second International Conference on the Global Energy and Water Cycle, Washington, D.C., June, 1996.

Sellers, P.J., B.W. Meeson, J. Closs, J. Collatz, F. Corprew, D. Dazlich, F.G. Hall, Y. Kerr, R. Koster, S. Los, K. Mitchell, J. McManus, D. Myers, K.J. Sun, and P. Try, 1995, "An overview of the ISLSCP Initiative I global data sets". on ISLSCP Initiative I global data sets for land-atmosphere models, 1987-88, Volumes 1-5, National Aeronautics and Space Administration (5 volume CD).

Schnur, R., and D.P. Lettenmaier, 1995, "A global gridded dataset of daily soil moisture". EOS, Transactions of the American Geophysical Union, 76(46) (Supplement), F127 (abstract).