Contents

Section 0. Executive Summary
Section 1. Introduction
Section 2. Background: NOHRSC OPPS And Data Assimilation
Section 3. Overview of SNOW-INFO Functionality
Section 4. Data Summary
Section 5. Processing Requirements
Section 6. RFC-NOHRSC Interaction Requirements
Section 7. SNOW-INFO Development and Implementation Schedule
Section 8. References

Recent snowmelt-flooding events in the Northeastern U.S. (1996) and in the Red River of the North (1997) have underscored the importance of seasonal snow covers in hydrological predictions. The National Weather Service's Office of Hydrology (OH) has long recognized the importance of representing snow accumulation and ablation processes in forecasts. For 25 years, the OH has maintained a snow accumulation and ablation model (SNOW-17 [Anderson, 1973]) that is designed to provide estimates of snowmelt input to other models within the NWS River Forecast System (NWSRFS). The SNOW-17 model, and its more recent state-space version, SNOW-43, describe snow accumulation and ablation processes in a simplified manner that allows the models to be driven only by inputs of precipitation and air temperature. Although these models were preceded by the development of more complex and complete representations of snow processes, simplification was necessary because precipitation and temperature were traditionally the only two variables that were reliably available. The simplified nature of the NWSRFS snow models leads to a greater dependence on empirical parameters to calibrate the models. This in turn leads to increased uncertainty in model results, particularly during extreme events or unusual conditions. Consequently, field observations of snow water equivalent and other snow pack characteristics are of great importance, both to evaluate the results of the snow model, and to update or correct the model. However, nationwide, there are very few routinely available observations of snow water equivalent. In the eastern and central U.S., objective procedures to effectively use the few snow water equivalent observations that do exist are lacking. And, as the Northeast floods of 1996 and the Red River of the North floods of 1997 clearly demonstrated, the uncertainties associated with the simplified NWSRFS snow models during extreme and unusual conditions are unsatisfactorily large.

The investigation following the Northeast snowmelt Floods of 1996 [Office of Hydrology, 1996] lead to the following seven recommendations (among others):

Rec. 1.1.a. A snow estimation and updating procedure is needed for the eastern U.S.

Rec. 1.1.b. Additional snow water equivalent measurements, especially from airborne snow surveys, would be helpful during periods of significant snow cover.

Rec. 1.4. Procedures need to be developed to use meteorological variables other than air temperature to estimate the amount of snowmelt, either to directly calculate snowmelt, or to determine and apply regional corrections to SNOW-17.

Rec. 1.5. New displays should allow forecasters to see what is taking place inside the snowmelt and rainfall-runoff models.

Rec. 1.6. Additional training should be provided to the RFCs on the physics of snow ablation.

Rec. 1.7.a. RFCs need to use procedures in mountainous areas that account for the effect of terrain on precipitation and air temperature.

Rec. 1.7.b. Improved displays need to be used that will depict the interaction between precipitation and terrain for the current event and compare this to isohyetal analyses of historical data.

These findings and recommendations are reviewed in detail in Appendix 1. This plan directly responds to these recommendations.

This plan describes a set of software tools to assist hydrologic forecasters in using many different types of snow information in a comprehensive, effective, and scientifically appropriate manner. We call this set of tools the Snow Information System, of SNOW-INFO. These software tools will provide an integrative framework to examine and analyze all available data related to snow accumulation and ablation processes. SNOW-INFO will allow forecasters to understand what can be known about current snow pack conditions given all available information, and to update relevant state variables in the NWSRFS snow models as they see fit.

SNOW-INFO will consist of four components (Figure 1): 1) a geospatial data base for storing various point, vector, and raster data, 2) a Data Transfer and Management software suite, 3) a Snow Analysis software suite, and 4) an Update NWSRFS software suite. Together, these tools will orchestrate the integration of a wide variety of snow information into a common environment, where forecasters will be able to easily view the data, manipulate the data, analyze the data, conduct ad hoc physical snow model runs to examine unusual or extreme scenarios, and provide updates to the NWSRFS snow model.

The National Operational Hydrologic Remote Sensing Center (NOHRSC) generates a variety of snow cover information products and distributes them widely. The NOHRSC operates two terrestrial gamma radiation detection systems on low-flying aircraft to infer snow water equivalent over a network of over 1800 flight lines covering portions of 28 states and five Canadian Provinces. Analysts at the NOHRSC use image data from the Geostationary Operational Environmental Satellite (GOES) and from the Advanced Very High Resolution Radiometer (AVHRR) on the NOAA polar orbiting satellites to digitally map the areal extent of snow cover for over 4000 river basins in the U.S. (including Alaska) and Canada. In addition to the airborne and satellite snow cover data, the NOHRSC has access to real-time, ground-based, snow water equivalent data collected in the Western U.S. by the National Resource Conservation Service, the California Department of Water Resources, and other Federal and State agencies. The ground-based, airborne, and satellite snow cover data are used by NOHRSC hydrologists to generate near real-time, alphanumeric and image products for electronic distribution to a variety of end-users for use in operational and research hydrology programs.

The NOHRSC is currently developing a comprehensive, operational four dimensional data assimilation (4DDA) framework for spatially distributed, physically based snow modeling (Appendix 2). Our purpose is to produce timely operational estimates of the distribution of snow water equivalent and its associated uncertainty over large hydrologic forecast regions. Our approach is to develop a distributed energy balance snow accumulation and ablation model for operational use at moderate (four km) spatial resolution and high temporal resolution (one to three hours) that will assimilate meteorological and snow pack data from satellite, airborne, and surface sources and analytical data from mesoscale atmospheric models. The 4DDA system will be run continuously at the NOHRSC; it will simulate the spatial distribution of snow water equivalent and other snow pack characteristics, and will estimate the uncertainty associated with these simulated values. Our approach combines two major elements: 1) a strong emphasis on physical understanding of snow properties and snow hydrology as the basis for modeling snow accumulation and melt, and 2) the comprehensive assimilation of remotely sensed satellite data with airborne, surface, and atmospheric model data to improve and update the model's representation of SWE. The 4DDA framework allows for data irregularly distributed in both space and time to be objectively used in the snow modeling. Operational products expected from this assimilation system include gridded estimates of snow water equivalent and other snow pack characteristics available on a three- or six-hour basis to RFCs and others. Various interim products used in the assimilation (e.g. various point snow observations, satellite-derived incident solar radiation, radar-derived snowfall estimates, etc.) will also be available. This will constitute a dramatic increase in the types and frequency of snow information products made available by the NOHRSC.

This plan describes a set of software tools to assist hydrologic forecasters in using many different types of snow information, including but not limited to the NOHRSC 4DDA snow products, in a comprehensive, effective, and scientifically appropriate manner. We call this set of tools the Snow Information System, or SNOW-INFO. The fundamental purpose of this system is to maximize the potential benefit of all snow information that is available for operational forecasting. The software tools we describe here are intended to provide an integrative framework to examine and analyze all available data related to snow accumulation and ablation processes, including:

• Ground and airborne observations of snow water equivalent, snow depth, snowfall, and storm total snowfall
• Meteorological analyses and forecasts
• Satellite-observed snow cover
• NOHRSC-estimated spatial distribution of snow water equivalent
• Current NWSRFS SNOW-17 model states.

The tools we describe are intended to allow forecasters to understand what can be known about current snow pack conditions given all available information, and to update relevant state variables in the NWSRFS snow models as they see fit.

The Snow Information System will consist of a unique geospatial data base and three suites of software modules designed to provide particular sets of functions. The system will be implemented within AWIPS as a Local Application. The system will be developed using the NOHRSC GISRS software architecture (e.g. as used in the Integrated Hydrologic Automatic Basin Boundaries (IHABBS) and the Basin Unit Hydrograph (UHG) software), which facilitates the analysis, manipulation, and viewing of a wide variety of point, vector, and raster geographic data. A graphical-user-interface (GUI) consisting of menus and viewers is fundamental to this architecture. The SNOW-INFO data base will consist of a GISRS data base for point, vector and raster data. The GISRS data base has already been installed in each RFC as part of the IHABBS and UHG implementation, but will be expanded to include new raster layers relevant to SNOW-INFO, and will be moved to AWIPS. Many of the functions to be included in SNOW-INFO have already been developed for use at the NOHRSC, and will be migrated to SNOW-INFO. The functionality to be provided by SNOW-INFO is described in this plan as three suites of software modules:

• -Data Transfer and Management Suite
  • Move Data Module
  • Point Data Quality Control Module
  • Raster Data Editing Module
  • Archive Data Module
• -Snow Analysis Suite
  • Scientific Data Visualization Module
  • Spatial Data Interpolation Module
  • Ad Hoc Point Physical Snow Modeling Module
• -Update NWSRFS Suite
  • Basin Analysis Module
  • Comparative Analysis Module
  • NWSRFS Update Module

These six modules will provide mechanisms to gather all the relevant data into a common environment, view the data, manipulate the data, analyze the data, conduct ad hoc physical snow model runs to examine unusual or extreme scenarios, provide updates to the NWSRFS snow model, and produce high-quality graphical output of plots and maps. The modular architecture of GISRS allows for future modules to be easily added as necessary.

The development and implementation of the SNOW-INFO software at the RFCs will occur in two phases. All of the basic functionality of SNOW-INFO will be included in Phase One, to be completed by October, 2000. It will include all of the mechanics of the SNOW-INFO/GISRS data base, the complete Data Transfer and Management Suite, and core modules required for the Snow Analysis Suite and the Update NWSRFS Suite. Many of the functions planned for SNOW-INFO are already in place at the NOHRSC. Migration of the existing NOHRSC functions to SNOW-INFO at the RFCs will be straight-forward. Phase Two will occur gradually as the NOHRSC data assimilation system comes on-line and new NOHRSC products begin to be generated. Modules designed to use specific data assimilation products will be added to the Snow Analysis and Update NWSRFS Suites during Phase Two. Full implementation of Phase Two is scheduled for October, 2001.

2.1. The NOHRSC Operational Product Processing System (OPPS)

The NOHRSC OPPS currently produces raster maps of the areal extent of snow cover for the continental U.S., derived from image data from the NOAA Geostationary Operational Environmental Satellite (GOES) and from the Advanced Very High Resolution Radiometer (AVHRR) on the NOAA polar orbiting satellites. The NOHRSC also operates two terrestrial gamma radiation detection systems on low-flying aircraft to infer snow water equivalent over a network of over 1800 flight lines covering portions of 28 states and five Canadian Provinces. When sufficient ground or airborne observations of snow water equivalent (SWE) are available, these data are augmented with observations of zero SWE obtained from the satellite-derived snow line, and a SWE analysis is conducted to generate a gridded estimate of the distribution of SWE. The western U.S. is fortunate to have a network of National Resource Conservation Service (NRCS) snow pillows that provide routine SWE observations. A sufficient number of these sites exist to permit a simple analysis of the distribution of SWE throughout the western U.S. twice each week at the NOHRSC. Raster and alphanumeric summary products from these analyses are distributed to western RFCs to assist in hydrologic forecasting. The products can be used directly to update SNOW-17 or SNOW-43 in the NWSRFS if desired. Similar analytical products are generated in the eastern and central U.S. only when NOHRSC airborne gamma SWE surveys are conducted. These products typically cover much smaller regions.

2.2. Physically Based Data Assimilation for Snow Estimation

The simple fact of the matter is that the network of operationally available SWE observations in the central and eastern U.S. is extremely sparse. Most sources provide point observations irregularly distributed in space and in time, such that at any given time only a subset of these data are available. At any given time, there is generally an insufficient number of SWE observations to permit an adequate estimation of mean SWE over a basin by spatial analysis (i.e. interpolation). The methodology of SEUS cannot be readily implemented in the eastern U.S. due to SEUS's fundamental dependance on historical mean SWE distributions. In the eastern and central U.S., there is inadequate information on historical mean SWE to support this approach. In these regions, a different method of snow estimation is required. After consideration of what data are available in these areas, we have concluded that the most appropriate scheme is to provide forecasters with a comprehensive toolbox that allows them to examine, analyze, display, and rapidly comprehend a wide variety of supporting snow and hydrometeorological information. Using this toolbox, forecasters will be able to generate appropriate updates to the NWSRFS snow models.

A variety of sources provide data relevant to snow hydrology (Table 1). They have been categorized here loosely as: 1) those that provide direct or indirect observations of SWE, 2) those that provide observations of other snow pack characteristics, and 3) those that provide information directly related to snow accumulation and ablation processes, but provide no actual snow information. Table 1 provides an indication of the motivation for developing a physically based assimilation system to estimate snow properties. In general, we have the least information from actual observations of SWE, more information about other snow pack properties, and we have the most information about indirectly related hydrometeorological variables.

The NOHRSC is currently developing a physically based four-dimensional data assimilation (4DDA) system to estimate the spatial and temporal distribution of SWE, using all available data to drive and guide the model (see Appendix 2 for details). In this system, knowledge of snow pack evolution with time is embodied in a spatially distributed, physical snow model. This permits the use of various observations irregularly distributed in time and space (e.g. Table 1) to drive and update the model. The model organizes and propagates forward information from previous observations. New observations are used to modify (update) the model states via one of many possible assimilation schemes. This approach accomplishes three important goals. First, it utilizes a strong physical understanding of snow properties and snow hydrology to provide estimates of SWE distributions for updating current empirical snow models used by forecasters. Second, estimates of mean basin SWE provided for short-term forecasting will be consistent with all available past and current snow observations. Third, this approach provides a much-needed framework within which new sources of information related to snow can be readily incorporated.

Several investigators have demonstrated the potential of using spatially distributed numerical snow models to estimate SWE over small to moderate sized basins under well-instrumented conditions [Bloschl and Kirnbauer, 1991; Tarboton et al., 1995; Cline et al., 1998]. We wish to capitalize on these developments at the NOHRSC, but we believe the 4DDA approach is necessary to provide a mechanism to compensate for the less ideal situation imposed by operationally available data over large forecast regions. This is a pragmatic approach driven both by our understanding of the limitations of available data and of the current capabilities of physically based, spatially distributed snow modeling [Bloschl and Kirnbauer, 1994; Cline, 1995, Cline, 1997b; Cline et al., 1997; Cline et al., 1998; Davis, et al., 1997].

Table 1. Types and sources of information relevant to snow hydrology. The data are categorized into three types: 1) observations of SWE, either directly or remotely ("SWE"), 2) observations of other snow properties, either directly or remotely ("SNOW"), and observations of other factors related in some way to snow pack conditions ("OTHER"). Sources of these data include the NOHRSC, the National Resource Conservation Service Snow Telemetry (SNOTEL) sites, weather stations, volunteer observers from the NWS Cooperative Observer program, and the NOAA GOES and AVHRR satellites. These data are collected from surface locations, airborne platforms including aircraft and weather balloons, and satellite platforms. Atmospheric model data are also considered a "platform" here. Most of these data are available throughout the U.S., with the exception of data from SNOTEL sites, which are almost exclusively located in the mountainous western U.S. Only satellite, radar, and model data are gridded; all other observations are collected at points. The temporal resolutions of these data range from nearly continuous, such as GOES satellite data, to once per season, as is often the case for airborne SWE observations.

2.3. Expected Products from the NOHRSC Data Assimilation System

The NOHRSC Data Assimilation System will be an important source of data for SNOW-INFO. The physical snow model forming the basis of the assimilation will run on a 1-3 hour time step and at 4-km spatial resolution. At each time step, all model state variables and gridded meteorological driving data (downscaled to 4-km resolution) will be available to SNOW-INFO for analysis by RFC forecasters. For a comparable example, consider the products from the NOAA Eta numerical weather prediction model, with a set of grids available every three hours.

The primary products that we expect to make available to SNOW-INFO include 4-km resolution rasters of assimilated snow water equivalence, snow depth, liquid water content, snow melt, and snow pack temperature. Other assimilation products that will be available include downscaled raster maps of numerical weather model data and satellite-derived incident solar radiation data [Tarpley et al., 1997], radar-derived snowfall estimates [Super and Holroyd, 1998], and various point snow observations. Appendix 2 provides more information on the data and processing anticipated in the NOHRSC Data Assimilation System. These NOHRSC Data Assimilation System products will be made available to SNOW-INFO using mechanisms described below.

3.1. Introduction

The fundamental purpose of SNOW-INFO is to maximize the potential benefit of all snow information that is available for operational forecasting. The system we describe here is intended to provide an integrative framework to examine and analyze all available data related to snow accumulation and ablation processes, allowing forecasters to understand what can be known about current snow pack conditions given all available information, and to update relevant state variables in the NWSRFS snow models as they see fit.

SNOW-INFO will consist of a geospatial data base and three suites of software modules designed to provide particular sets of functions (Figure 2). The system will be developed using the NOHRSC GISRS software architecture (e.g. as used in IHABBS and UHG), which facilitates the analysis, manipulation, and viewing of a wide variety of point, vector, and raster geographic data. A graphical-user-interface (GUI) consisting of menus and viewers is fundamental to this architecture. The SNOW-INFO data base will consist of a GISRS data base for point, vector and raster data. SNOW-INFO will be implemented as a Local Application within each RFC AWIPS system; the SNOW-INFO software will run on AWIPS CPUs. The functionality to be provided by SNOW-INFO is contained within three suites of software modules:

• Data Transfer and Management Suite
  • Move Data Module
    • transfer data from NOHRSC to RFC SNOW-INFO data base
    • populate RFC SNOW-INFO data base with relevant local data
    • transfer relevant local data to NOHRSC for assimilation
  • Point Data Quality Control Module
    • automatically quality control point data in SNOW-INFO data base
    • facilitate manual QC of point data in SNOW-INFO data base by RFC
  • Raster Data Editing Module
    • modify values in raster data sets
  • Archive Data Module
    • remove data older than a specified age and archive it at RFC

• Snow Analysis Suite
  • Scientific Data Visualization Module
    • query any point, vector, or raster data
    • plot data time series
    • graph frequency histograms
    • display maps (e.g. rasters overlain with relevant vectors and points)
    • display rasters draped over topographic surfaces (e.g. "pseudo 3D)
    • animate sequences of rasters
  • Spatial Data Interpolation Module
    • interpolate irregular series of selected points to create a raster
    • convolve irregular series of points with existing raster to update raster
  • Ad Hoc Point Physical Snow Modeling Module
    • simulate current or future snow characteristics at points
    • test scenarios, e.g. What if QPF rain occurs on given snow pack?

• Update NWSRFS Suite
  • Basin Analysis Module
    • summarize relevant gridded fields on a basin-by-basin basis
  • Comparative Analysis Module
    • compare basin analyses of relevant grids to SNOW-17/43 states
  • NWSRFS Update Module
    • write update mod for SNOW-17/43

In brief, these six modules provide mechanisms to gather all the relevant data into a common environment, view the data, manipulate the data, analyze the data, conduct ad hoc physical snow model runs to examine unusual or extreme scenarios, provide updates to the NWSRFS snow model, and produce high-quality graphical output of plots and maps. The modular architecture of GISRS allows for future modules to be easily added as necessary.

3.2. The SNOW-INFO Data Base

The SNOW-INFO geospatial data base is central to all SNOW-INFO tasks, and will point, vector and raster (e.g. gridded) data. Point, vector and raster data relevant to SNOW-INFO will be stored in the RFC's existing GISRS data base. This data base has already been installed at each RFC as part of the IHABBS and the UHG software implementations, and will be moved to the AWIPS environment when SNOW-INFO is deployed.

Point data to be used in SNOW-INFO will be stored as a flat file in the GISRS data base. The flat file will be populated with data extracted from the RFC's local Informix data base. The existing RFC GISRS data base will become substantially larger in size as new raster layers relevant to SNOW-INFO are produced and stored. The addition of the SNOW-INFO data base to the RFC computer systems will almost certainly require additional disk space (see Section 5 for estimates of disk space requirements).

3.3. Data Transfer and Management Suite

Data that will be involved in SNOW-INFO can be conveniently categorized as either dynamic or static, as either point, vector, or raster, and as either originating locally via the RFC, or as originating from other sources such as the NOHRSC. All vector data relevant to SNOW-INFO is relatively static (e.g. basin boundaries or rivers), and will be simply contained and available within the GISRS data base. Similarly, many raster data sets are relatively static, such as digital elevation data, and these also will simply be stored and available. The SNOW-INFO Data Transfer and Management Suite is primarily concerned with dynamic point and raster data that change frequently.

The NOHRSC currently produces (using the OPPS) both point and raster data sets that are now generally available to RFCs via standard ftp and the internet. As the NOHRSC data assimilation system comes on line, new point and raster data sets will be generated that will be available through AWIPS. One function of the SNOW-INFO Data Transfer and Management Suite is to orchestrate the delivery of these data sets to the SNOW-INFO data base in a timely manner. A second function is to populate the SNOW-INFO data base with snow-related point data obtained locally by the RFC (e.g. cooperative observer data), and to transfer these data to the NOHRSC where they can be assimilated. A third function is to conduct quality control of the point data in the SNOW-INFO data base. The final function is to archive data sets that are no longer needed on-line in the SNOW-INFO data base.

These functions will be provided by four modules: 1) Move Data, 2) Point Data Quality Control, 3) Raster Data Editing, and 4) Archive Data.

3.3.1. Move Data Module. This module will orchestrate the transfer of data from the NOHRSC to the RFC SNOW-INFO data base, as well ensure that locally available snow data brought into the RFC are made available to SNOW-INFO and shipped to the NOHRSC for assimilation (Figure 1).

Our general working model for how point data enter the RFC is as follows. Point data arrive at the RFC in a variety of formats, including SHEF, METAR, faxes, and even hand-written messages. At some stage, nearly all point data are formatted into SHEF messages and passed to an incoming SHEF directory, where relevant SHEF physical elements are decoded and posted into the RFC Informix data base. The RFC is responsible for ensuring that relevant point data are stored in their Informix data base.

The Move Data module includes a Data Extractor program which extracts point data relevant to SNOW-INFO from the local RFC Informix data base and makes it available to SNOW-INFO. The specific schema of the local Informix data base in unimportant to SNOW-INFO, but it must be specified by the RFC when configuring SNOW-INFO. A variety of point data relevant to the objectives of SNOW-INFO might be ingested, including the physical elements listed in Table 2.

Table 2. SHEF physical elements relevant to the objectives of SNOW-INFO.

This mechanism ensures that all relevant data in the RFC's Informix data base can be used by SNOW-INFO. If data pertaining to snow or snow processes are ingested by the RFC but not stored in Informix, then a) these data will not be available in SNOW-INFO, and b) these data will not be assimilated at the NOHRSC.

Raster data sets produced by the NOHRSC will be shipped directly to appropriate storage directories managed by the SNOW-INFO GISRS data base. The data are expected to be shipped through an AWIPS firewall (e.g. LDADS) using standard ftp over the AWIPS LAN. The raster data transfer will be initiated by the NOHRSC on a regular schedule determined by the time step used by the NOHRSC snow model (i.e. 1-3 hours). The Move Data module will also manage new raster data as they enter the system by recording attributes and relevant meta-data in the GISRS data base.

Following automatic quality control of locally-obtained point data, described in the next section, the Move Data module will transfer a copy of these data to the NOHRSC via AWIPS-ftp, where they will be assimilated in the next model time step. This results in the following data cycle: as soon as locally obtained, SHEF-formatted data are decoded and posted to the local RFC Informix data base, the data are extracted and ingested into SNOW-INFO. There, they are automatically quality controlled, and a copy is shipped to the NOHRSC, where it will be immediately assimilated into the physical snow model. The next snow raster shipped from the NOHRSC to the SNOW-INFO data base will contain information from those point data.

3.3.2. Point Data Quality Control Module. This module will automatically quality control all point data contained in the SNOW-INFO data base. It will also facilitate additional manual quality control of any point data (Figure 2).

The NOHRSC has developed a comprehensive automatic quality control system for point data. It is currently used in the NOHRSC OPPS. This QC system will be included in the SNOW-INFO Data Transfer and Management Suite. Every data point is attributed with a 16-bit long word, with all bits initially set to zero. A battery of tests are conducted on every point (the specific nature of the tests varies with the type of data), and each test is assigned a particular bit in the QC word. When a test is passed, the appropriate bit is set to 1. There are four major categories of automatic QC tests performed:

• 1) Reasonable Values. For a given data type, the point value must fall within a reasonable range. For example, an air temperature of 200^oC would fail, and the appropriate bit for this test would remain set at zero.
  
• 2) Internal Consistency. For a given data type, the point value must be consistent with values of other data types at the same location and time. For example, for point x,y at time t, the dew point temperature cannot exceed the air temperature. If this test is failed, appropriate bits in both the dewpoint temperature and the air temperature QC words are set to zero.
  
• 3) Temporal Consistency. For a given data type, departures from the previous value exceeding some threshold are considered spurious, and an appropriate bit is set to zero. For example, if the current air temperature at point x,y is 60^oF, but the value at the previous hour was 10^oF, the test would fail and the appropriate bit in the current value remains set at zero. Consideration is given to the time interval since the previous observation.
  
• 4) Spatial Consistency. For a given data type, excessive departures of a given point value from the mean of its nearest neighbors are considered spurious. For example, if the air temperature of point x,y is 60^oF but the mean of nine nearest neighbors is 20^oF and the standard deviation is 10^oF, the value of point x,y is considered spurious and the appropriate bit for that datum remains set at zero. Consideration is given to the distances of the nearest neighbor to the target point.
  

The nature of the automatic QC tests performed and the criteria for each test vary according to data type. The only test required for all data types is the reasonable value test. More than one test may be performed for each category - e.g. there may be a number of appropriate internal consistency tests performed on a particular data type. Up to 12 tests can be performed on a single datum. The remaining four bits in the QC word are reserved for QC administration, such as indicating whether or not QC has been performed on a value. The nature of the individual tests for each data type is based on sound physical reasoning and remains fixed. This scheme allows for flexible use of the quality control information in subsequent processing. All tasks that use the quality controlled point data contain configurable QC filtering options that specify which QC tests must be passed and which tests can be ignored. Thus, for one application, QC results other than reasonable values could be ignored, while for another application only values passing all tests might be used.

A manual QC component will be added to the Point Data Quality Control module for SNOW-INFO. Two bits in the QC long word will be dedicated to manual QC - one indicating whether or not manual QC has been performed, and one indicating the outcome (pass/fail). The criteria for manual QC will be entirely up to the user. The Point Data Quality Control module will provide tools to facilitate manual QC. It will permit all point data to be queried, viewed in relation to a time series, etc. as described below in the Scientific Data Visualization module. Sets of irregular points may be interpolated using the Spatial Data Interpolation module to examine how a given value relates to neighboring values. Using these tools, the user will be able to set the manual QC bit to pass or fail for any reason. As with the automatic QC flags, the relevance of the manual QC bit in subsequent processing is configurable.

3.3.3. Raster Data Editing Module. This module will allow raster data sets to be edited, i.e. the values of individual pixels or groups of pixels can be modified and the results saved to a new raster (Figure 2).

Under some circumstances, a user may wish to edit values in a raster data set. For example, a forecaster familiar with a basin may be certain that there is continuous snow cover throughout the basin, and want to change values in a satellite-derived snow cover raster from cloud to snow. We have developed a Raster Data Editing module that provides this ability at the NOHRSC, and will include this within the Data Transfer and Management Suite. Editing can be performed on an individual pixel, or on groups of pixels within an existing polygon (such as a basin) or a user-defined polygon. The edited raster will only be saved as a new raster file.

3.3.4. Archive Data Module. This module will remove data older than a specified age and archive it on a user-designated device, such as a tape drive (Figure 2).

This module has been developed for the NOHRSC OPPS and will be included in the Data Transfer and Management Suite. Each data type (e.g. SWE, snow depth, etc.) in any category (point, vector, or raster) is assigned a configurable time-to-archive parameter. When data become older than the specified age, they are automatically moved out of the SNOW-INFO data base and into an archive directory. From there they are written to a specified archiving media such as a tape drive, CD-ROM, etc. Options are available to remove but not archive specific data types, and the time-to-archive for specific data types can be set to infinity, meaning they are permanent until removed manually. This module allows for flexible use of on-line data storage resources. If the SNOW-INFO user wants to retain certain on-line data, such as point observations of particular variables at particular stations, for the entire water year, while retaining other on-line data such as Eta model grids for only the last week, the Archive Data module permits this.

3.4. Snow Analysis Suite

In contrast to the Data Transfer and Management Suite, which will generally be transparent to the SNOW-INFO user, the Snow Analysis Suite will be the primary interface to SNOW-INFO. The Snow Analysis Suite will serve three major functions. First, it will provide a comprehensive set of scientific data visualization tools designed to facilitate examination and understanding of all data within SNOW-INFO. Second, it will provide flexible spatial data interpolation tools to facilitate analysis of irregular point data, and in Phase Two will permit raster data to be updated using irregular point analyses. Third, it will provide an interactive point energy and mass balance modeling system to conduct ad hoc analyses of current and future snow cover conditions, to test scenarios, and to develop correction schemes for the NWSRFS snow model. The tools in the Snow Analysis Suite are intended to be used interactively in conjunction with the Update NWSRFS Suite, described below, to provide guidance to update SNOW-17 or SNOW-43. The Snow Analysis Suite will consist of three modules: 1) Scientific Data Visualization, 2) Spatial Data Interpolation, and 3) Ad Hoc Point Physical Snow Modeling (Figure 2).

3.4.1. Scientific Data Visualization Module. This module will provide a comprehensive, generic, and flexible set of data query and display capabilities, allowing the SNOW-INFO user to examine all data contained within the SNOW-INFO data base, interim data products created within the Snow Analysis Suite, and NWSRFS snow model state information.

The basic form of data display within the GISRS architecture of SNOW-INFO is a map, which typically might consist of a raster (e.g. satellite-derived snow cover) overlain by a set of vectors (e.g. basin boundaries) and by a set of points (e.g. meteorological stations). The SNOW-INFO data base permits great flexibility in choosing the set of points to be displayed on a map. For example, stations that have reported snow depth in the last 24 hours might be displayed on one map, while the next map might display all stations that have ever reported any type of snow information. From here, SNOW-INFO will permit many types of data queries to be made. Different types of appropriate data displays will be provided that will depend on the nature of the query.

Existing GISRS architecture permits a comprehensive cursor query of raster data. As the user moves the screen cursor over a displayed map, corresponding geographic coordinates are displayed together with pixel values of user-selected raster data sets. For example, the user might display a raster of satellite-derived snow cover, and choose to query gridded snow water equivalent , air temperature, and precipitation from the Eta model. By moving the screen cursor over the snow cover map, pixel values for all four variables are displayed with the latitude, longitude, and elevation of the pixel.

The Scientific Data Visualization module will provide several fundamental querying tools that will generate time-series plots of point data and frequency histograms. For example, a user can select a time window and a variable, then go to a map with stations displayed and select one station at a time from the map to view time series information for that station. A user could display a frequency histogram of point values in the SNOW-INFO data base by specifying a variable (e.g. snow depth), a region (e.g. a forecast group), and a time. Alternately, the user might draw a polygon on a displayed raster to generate a histogram of the values within the polygon.

Additional types of graphical displays will be included in the Scientific Data Visualization module that will be aimed at facilitating interpretation and comprehension of SNOW-INFO data. One type is basic raster animation - displays of sequences of rasters to observe evolving patterns over time. A second type of display that SNOW-INFO will provide is surface rendering and texture mapping, which is familiar to GIS users and is often referred to as "pseudo-3D" or "draped images". These plots are generally used to display various raster data in relation to topography using a digital elevation model (DEM) as the base of an XYZ plot. Such displays often improve cognition of the location of particular features (e.g. snow cover) relative to topographic features (e.g. a basin).

3.4.2. Spatial Data Interpolation Module. This module will provide methods of analyzing irregular point data and predicting values at other points or to generate rasters by spatial interpolation. It will also permit existing rasters to be updated with point values by convolving a set of irregular points with the raster.

Three methods of interpolation will be provided in this module. The first is simple inverse-distance-weighted (e.g. "1/D") interpolation, where a distance function provides the only control on interpolated values. The user may specify parameters for the distance function, such as the search radius, the number of neighbors to include, and distance weights. This simple form of interpolation is appropriate when information about the underlying processes controlling the spatial distribution of values is unknown or not available.

The second method is elevational detrending, used by the NOHRSC to analyze snow water equivalent observations in the western U.S. This method relates point values to their elevations, removes any elevation trend, interpolates the deviations, and then adds the elevation trend back in to the interpolated values. This method is useful for interpolating variables that generally exhibit a strong elevational dependence, such as snowfall in mountainous regions.

The third method is a weighted topographic regression approach, similar to the method used in the PRISM (Parameter-elevation Regressions on Independent Slopes Model). In this approach, a weighted multiple regression function is developed for each target point from nearby stations, and the value of each target point is predicted with this function. In the multiple regression, greater weight is given to stations with location, elevation, and topographic characteristics similar to that of the target point. The PRISM method has been shown to be an effective interpolator for precipitation and other hydrometeorological variables in mountainous regions [Daly, et al., 1994]. The PRISM method will be generalized somewhat here to permit interpolation of a variety of variables.

The Spatial Data Interpolation module will include an interpolation error analysis tool to estimate the uncertainty associated with any interpolated raster. This tool will include simple jackknifing procedures, whereby the interpolation is repeated several times with a set of points withheld from the interpolation each time. The differences between the withheld and interpolated values provide a measure of the quality of the interpolated raster. This tool will also permit errors from the weighted topographic regression method to be evaluated.

Snow water equivalent analyses will permit sampling of pseudo observations of zero snow water equivalent from satellite-derived snow line information. This is currently used in the elevational detrending method at the NOHRSC for the western U.S., and is a valid approach to augmenting a sparse network of snow water equivalent observations prior to interpolation.

As described above, locally obtained snow data at the RFC will be shipped automatically to the NOHRSC for assimilation, and the assimilated information will be contained within the next set of raster data sent to the RFCs. The length of this data cycle will depend on the assimilation processing interval used at the NOHRSC (e.g., 1-3 hours). In the time period between the RFC's receipt of new local point data and receipt of a new assimilated raster containing this information from the NOHRSC, forecasters may want to update their most recent raster with the new point data. The Spatial Data Interpolation module will permit this using a convolution technique in which point values are interpolated "into" an existing raster. In this approach, introduced values influence their surrounding region in a manner specified by the interpolation parameters.

3.4.3. Ad Hoc Point Physical Snow Modeling Module. This module will allow SNOW-INFO users to conduct ad hoc analyses of current and future snow cover conditions, to test scenarios, and to develop correction schemes for the NWSRFS snow model using an interactive point energy and mass balance modeling system.

This module will be based on a one dimensional (point) energy and mass balance model for a snow cover called SNTHERM [Jordan, 1990]. This model is based heavily on the work of Anderson [1976]; Jordan [1990] later added several improvements, including a more complete representation of water flow through the snow pack. SNTHERM can be considered a thorough encoding of our current understanding of snow energy and mass exchanges, and has frequently been used as a standard for comparison to other snow models (e.g. Harrington et al., [1995]).

The SNTHERM model calculates snow pack state conditions that will satisfy energy and mass balance requirements given the previous state of the snow pack and external meteorological conditions. Coupled heat and mass flow is distributed vertically within the modeled snow pack using a control volume approach. State variables represented in the current version of SNTHERM include SWE, snow depth, snow density, snow temperature (both internal and snow surface), change in snow pack heat content, and snow grain size. The model is driven by measured or estimated inputs of air temperature, relative humidity, wind speed, precipitation, incident solar radiation, and incident longwave radiation. The last two inputs are estimated internally if they cannot be provided. Bloschl and Kirnbauer [1991] demonstrated that vertically distributed, physically based models of this type are most appropriate for representing snow pack characteristics during periods of rapid change, an important consideration for providing snow guidance products for short term hydrologic forecasting.

SNTHERM has been successfully tested in a variety of environments (e.g. Jordan, [1990]; Cline, [1997a]), establishing confidence that the model physics are correct. It has few major parameters requiring calibration or tuning. In general, it accurately predicts how a snow pack will respond to meteorological conditions. These positive characteristics also make the model sensitive to poor data inputs. Nonetheless, it is a useful investigative tool for gaining insight into how snow packs may respond to certain conditions, and may be used to help diagnose the behavior of simpler snow models like SNOW-17/43. If input data of reasonable quality are available, it may be used to estimate current snow pack conditions as an independent model to SNOW-17/43. In an ad hoc investigative/diagnostic mode, a snow pack can easily be prescribed for SNTHERM, either from field observations, from the SNOW-17/43 models, or a hypothetical example, and various meteorological conditions can be tested to gain insight into snow pack response. These capabilities permit a wide variety of scenario testing. For example, a model snow pack can be constructed, and the effects of rainfall expected from a quantitative precipitation forecast can be studied.

The Ad Hoc Physical Snow Modeling module will provide a graphical interface to SNTHERM that will permit rapid initialization and analysis of ad hoc SNTHERM model runs. To run the SNTHERM model, tools will be provided to assist in identifying points with sufficient meteorological input data to drive SNTHERM, and, alternately, to extract input data from rasters within the SNOW-INFO data base (i.e. Eta, RUC, and satellite-derived incident solar radiation). Although this module is intended for ad hoc snow pack analyses, users will have the option of running the model continuously for specific points if they desire. Used in conjunction with the spatial interpolation tools, it will be possible for users to compare SNTHERM output to SNOW-17/43 model states and develop correction schemes for SNOW-17/43. The Update NWSRFS Suite described below will facilitate this last task.

3.5. Update NWSRFS Suite

This set of three modules will form a crosswalk between SNOW-INFO and the NWSRFS snow models. The first module will analyze and summarize raster data (or groups of irregular points) on a basin-by-basin (or basin elevation zones) basis. The second module will facilitate comparison of SNOW-17/43 model states to summarized SNOW-INFO data. The third module will generate SNOW-17/43 modification files that can be used to update the NWSRFS snow models.

3.5.1. Basin Analysis Module. This module will statistically summarize point or raster data within vector polygons defining basins or basin sub-areas.

This module is currently used at the NOHRSC to summarize satellite-derived rasters of areal extent of snow cover and gridded snow water equivalent estimates for each basin or basin sub-area (e.g. elevation zone). For SNOW-INFO, it will be expanded to allow summarization of a series of irregular point data. Thus, for example, the mean snow water equivalent for a basin could be calculated from a series of cooperative observer data or from a series of SNTHERM model runs at several points within the basin. Vector polygon basin boundaries for each RFC have either been calculated using IHABBS software or have been otherwise provided by the RFCs, and are available within the SNOW-INFO GISRS data base.

3.5.2. Comparative Analysis Module. This module will provide a mechanism for current NWSRFS snow model states to be ingested into SNOW-INFO, where they can be compared to basin or basin sub-area summary statistics from the Basin Analysis module.

The primary task for this module is to allow SNOW-17/43 model state information to be either manually or automatically entered into the SNOW-INFO processing engine. A simple interface will be provided to allow the user to manually enter an individual basin ID, time, and associated model states, or to specify a text file containing this information for single or multiple basins. Use of this module will require a modification to NWSRFS to extract complete SNOW-17/43 model state information and write this information to a file that SNOW-INFO can access. We propose that the Office of Hydrology make this modification to NWSRFS.

Once the SNOW-17/43 model states are in the SNOW-INFO data base, data visualization tools may be used to map the information (if multiple basins are involved), and evaluate differences between the NWSRFS model states and SNOW-INFO Basin Analysis results. Snow analysis tools may be used to further identify reasons for observed discrepancies.

3.5.3. NWSRFS Update Module. This module will write a NWSRFS modification file to update model states for selected basins.

The user will first be able to select basins to be updated from the list of basins currently under analysis in the Update NWSRFS Suite. The update values can then be edited by the user as desired. When the user is satisfied with the selected basin updates, a file will be written containing the update information.

NWSRFS snow model updates generated from SNOW-INFO will generally be derived either from NOHRSC-assimilated snow water equivalent rasters (either as is, or modified using the SNOW-INFO Edit Raster or Spatial Data Interpolation modules), or from ad hoc raster information generated from various point data using tools from the Snow Analysis software suite. In either case, the update information will not account for any biases that are calibrated into the NWSRFS snow model. When NWSRFS is properly calibrated, the SNOW-17/43 models should provide a relatively unbiased estimate of mean areal snow water equivalent for a basin. Thus, SNOW-INFO will be most appropriate for updating the NWSRFS snow models in basins that are properly calibrated.

3.6. On-Line Help and SNOW-INFO Technical Manual.

SNOW-INFO will include both an on-line help functionality and a web-based Technical Manual that will demonstrate relevant analyses that can be performed with the tools in SNOW-INFO. Appropriate use of SNOW-INFO will require a sound background in snow physics and in geospatial analysis. The Technical Manual will contain required background information on these topics. The NOHRSC will provide support for SNOW-INFO, and we suggest that part of this support should include a regular training program that will provide each RFC with a review of snow accumulation and ablation processes, current SNOW-INFO capabilities, and current NOHRSC data assimilation methods and products.

3.7. Interactive and Command Line Modes

Each of the modules described above include an interactive user interface. It is important to note, however, that all of the functions described can be run from a command line by entering the function and its required arguments. This allows for functions to be run automatically from a cron, for example.

The NOHRSC will continue to ship point data sets to RFCs via SHEF-formatted text messages. These data will include airborne snow water equivalent observations and satellite-derived basin areal extent of snow cover. Where relevant, NRCS SNOTEL data will also be shipped in SHEF format.

Raster data sets to be shipped from the NOHRSC directly to the RFC SNOW-INFO data base will include the following:

• satellite-derived snow cover (available daily)
• various NOHRSC-assimilated data sets, including snow water equivalent and snow depth (available on a 1-3 hour basis)
• satellite-derived incident solar radiation (available hourly)
• selected variables from numerical weather analyses and forecasts from the Eta and RUC models (including 2-m height air temperature, relative humidity, wind vector components, barometric pressure, precipitation - downscaled and formatted for use in SNOW-INFO, and available every 3 hours.

• digital elevation data
• forest cover and type data
• PRISM historical monthly precipitation data, 1961-1990
• STATSGO soils data

Local RFC point data that will be extracted from Informix by SNOW-INFO and passed on to the NOHRSC will include all snow water equivalent, snow depth, and snowfall observations from cooperative observers and other data sources.

Most of the tools we have described for SNOW-INFO are not particularly CPU intensive. Spatial interpolation generally requires the greatest computational resources, but for most situations these requirements are not excessive. Some visualization tools are CPU or memory intensive (such as movie loops/animations).We expect that all RFCs will have sufficient computational resources within AWIPS to run SNOW-INFO processes. Storage space is another issue, however. The SNOW-INFO data base will contain numerous raster data sets for the RFC spatial domain. Since raster data shipped form the NOHRSC will accumulate within the SNOW-INFO data base each day, the time period for which these data can be stored on-line will be strictly limited by available disk space. Data storage requirements will vary by the size of each RFC's spatial domain (Table 3) and the desired on-line time period for various data sets. It should be anticipated that all RFC AWIPS systems will require additional hard disk space when SNOW-INFO is implemented.

The SNOW-INFO system will also require a dedicated archive media device (e.g. a tape drive).

Table 3. Estimated disk space requirements for raster data are shown . Individual file sizes are based on uncompressed 2-byte, 4-km resolution raster data. Depending on the variable, raster data might be stored as 1, 2, or 4-byte data, so 2-byte is used as an "average" file size. Raster data in SNOW-INFO are compressed using run-length encoding, so actual file sizes will be somewhat less than shown here. Total daily storage requirements are based on the following assumed list of files:

• Three 3-hourly NOHRSC-assimilated variables (snow cover, snow water equivalent, and snow depth), shipped to the RFC eight times each day (3x8=24)
• Hourly satellite-derived solar insolation, shipped from the NOHRSC to the RFC up to twelve times per day (1x12=12)
• Six 3-hourly variables from the Eta or RUC models shipped eight times per day (6x8=48)
• Twenty-five static rasters, such as digital elevation data, PRISM data, etc. (1x25=25)
The approximate number of days of dynamic raster data that can be stored is based on a 9 Gb hard drive. The actual number of days should be greater than shown here because of file compression.

RFC	Single 2-byte raster file size (Mb)	NOHRSC Snow Products (24) (Mb)	Satellite Insolation Data (12) (Mb)	Eta/RUC Met Data (48) (Mb)	Dynamic Daily SubTotal (Mb)	Static Data Sets (25) (Mb)	Total (Mb)	Approx. Days of On-Line Dynamic Data Storage
ABRFC	0.8	19.2	9.6	38.4	67.2	20.0	87.2	132
AKRFC	7.7	184.8	92.4	369.6	646.8	192.5	839.3	12
CBRFC	1.4	33.6	16.8	67.2	117.6	35.0	152.6	75
CNRFC	1.2	28.8	14.4	57.6	100.8	30.0	130.8	87
LMRFC	1.2	28.8	14.4	57.6	100.8	30.0	130.8	87
MARFC	0.6	14.4	7.2	28.8	50.4	15.0	65.4	177
MBRFC	3.0	72.0	36.0	144.0	252.0	75.0	327.0	34
NCRFC	2.6	62.4	31.2	124.8	218.4	65.0	283.4	39
NERFC	1.4	33.6	16.8	67.2	117.6	35.0	152.6	75
NWRFC	1.5	36.0	18.0	72.0	126.0	37.5	163.5	70
OHRFC	1.4	33.6	16.8	67.2	117.6	35.0	152.6	75
SERFC	1.6	38.4	19.2	76.8	134.4	40.0	174.4	65
WGRFC	1.8	43.2	21.6	86.4	151.2	45.0	196.2	58




---

SECTION 6. RFC-NOHRSC INTERACTION REQUIREMENTS

• The NOHRSC will develop and implement the SNOW-INFO software at participating RFCs as a Local Application in AWIPS.
• The NOHRSC will provide general software support.
• The NOHRSC will provide periodic training on the physics of snow, geospatial analysis, and how to effectively use the SNOW-INFO software system.
• Implementation of SNOW-INFO will require a modification to NWSRFS to output all SNOW-17/43 model states to a file for ingestion into the SNOW-INFO data base, as described in Section 3.5.2.
• Implementation of SNOW-INFO will require forecasters to configure the software to schedule tasks such as data archiving.
• The need for additional AWIPS hardware resources (i.e. hard disk space, archive device, and possibly memory) at the time of SNOW-INFO implementation should be anticipated.
• The implementation of SNOW-INFO at the RFCs involves a symbiotic relationship - SNOW-INFO also serves as a mechanism to transfer locally obtained snow data to the NOHRSC to satisfy data assimilation objectives. RFCs will be expected to insert all locally obtained snow data into their Informix data base in a timely manner.

---

SECTION 7. SNOW-INFO DEVELOPMENT AND IMPLEMENTATION SCHEDULE

The development and implementation of the SNOW-INFO software at the RFCs will occur in two phases. All of the basic functionality of SNOW-INFO will be included in Phase One, to be completed by October, 2000. It will include all of the mechanics of the SNOW-INFO/GISRS data base, the complete Data Transfer and Management Suite, and core modules required for the Snow Analysis Suite and the Update NWSRFS Suite. Most of the GISRS geospatial data base component of SNOW-INFO is already in place at the RFCs and will be moved to AWIPS. Most of the data transfer and management functions planned for SNOW-INFO are already in place at the NOHRSC. Migration of the NOHRSC data transfer and management functions to SNOW-INFO at the RFCs will be straight-forward, with only a few relatively minor modifications required. Phase Two will occur gradually as the NOHRSC data assimilation system comes on-line and new NOHRSC products begin to be generated. Modules designed to use specific data assimilation products will be added to the Snow Analysis and Update NWSRFS Suites during Phase Two. Full implementation of Phase Two is scheduled for October, 2001.

---

SECTION 8. REFERENCES

Anderson, E.A., National Weather Service River Forecast System - Snow Accumulation and Ablation Model, NOAA Technical Memorandum, NWS, HYDRO-17, Silver Spring, Maryland, 1973.

Anderson, E.A., A point energy and mass balance model of a snow cover, NOAA Technical Report NWS 19, Office of Hydrology, National Weather Service, Silver Spring, Maryland, 1976.

Blöschl, G., and R. Kirnbauer, Point snowmelt models with different degrees of complexity - internal processes, Journal of Hydrology, Vol. 129, 127-147, 1991.

Blöschl, G., and R. Kirnbauer, Entering the era of distributed snow models, Nordic Hydrology, 25, 1-24, 1994.

Cline, D., Studies Supporting the Development of Spatially Distributed, Physically Based Snowmelt Models for Continental Alpine Areas, Ph.D. Dissertation, University of Colorado, 1995.

Cline, D., Snow surface energy exchanges and snowmelt at a continental, midlatitude Alpine site, Water Resources Research, 33(4), 689-701, 1997a.

Cline, D., Sub-resolution energy exchanges and snowmelt in a distributed SWE and snowmelt model for mountain basins, EOS, Transactions, American Geophysical Union, 78(46) Supplement, p. 210, 1997b.

Cline, D., Elder, K., and R. Bales, Scale effects in a distributed SWE and snowmelt model for mountain basins, Proceedings, 65th Western Snow Conference, 317-328, 1997.

Cline, D., R. Bales, and J. Dozier, Estimating the spatial distribution of snow in mountain basins using remote sensing and energy balance modeling, Water Resources Research, 34(5), 1275-1285, 1998.

Daly ,C., R.P. Neilson, and D.L. Phillips, 1994: A statistical-topographic model for mapping climatological precipitation over mountainous terrain. J. Appl. Meteor., 33, 140-158.

Davis, R., J. McKenzie, and R. Jordan, Distributed snow process modeling: An image segmentation approach, Hydrologic Processes, 9, 865-875, 1995.

Harrington, R., R. Jordan, and D. Tarboton, A comparison of two energy balance snowmelt models, EOS, Transactions, American Geophysical Union, 76(46) Supplement, p. 185, 1995.

Jordan, R., User's Guide for USACRREL One-Dimensional Snow Temperature Model (SNTHERM.89). Hanover: U.S. Army Cold Regions Research and Engineering Laboratory, 1990.

Office of Hydrology, Northeast Floods of January 1996. Natural Disaster Survey Report, U.S. Department of Commerce, National Oceanic and Atmospheric Administration, National Weather Service, Silver Spring, Maryland, 1998.

Super, A., and E. Holroyd, III, Snow Accumulation Algorithm for the WSR-88D Radar: Final Report, Denver: Bureau of Reclamation, 1998.

Tarboton, D., T. Chowdhury, and T. Jackson, A spatially distributed energy balance snowmelt model, in Biogeochemistry of Seasonally Snow-Covered Catchments, IAHS Publication #228, 141-155, 1995.

Tarpley, D., R. Pinker, I. Laszlo, and K. Mitchell, GOES surface and cloud products for validation of regional NWP models, GEWEX Continental-Scale International Project (GCIP) Meeting Abstracts, University Consortium for Atmospheric Research/National Center for Atmospheric Research, November 5, Boulder, CO, p. 39, 1997.





NOHRSC Mission Statement  |  Contact

National Weather Service National Operational Hydrologic Remote Sensing Center Office of Water Prediction 1735 Lake Drive W. Chanhassen, MN 55317
Contact NOHRSC Glossary Credits Information Quality Page last modified: Feb 14, 2025 - cloud 1	About Us Disclaimer Privacy Policy FOIA Career Opportunities