In March 2003, the Canadian government announced a reorganization of the Meteorological Service of Canada, which included a significant change in the way atmospheric research will be conducted. The locus of this reorganization involves the establishment of six national labs, each with a mandate for research in key areas including; marine/coastal issues, severe weather, remote sensing, complex terrain, hydrometeorology and the arctic. These labs will be established in conjunction with storm prediction centres (SPCs) located in each regional headquarters. A critical component of each national lab will be to ensure that the science and new techniques developed are successfully transferred to the production centres. The Arctic lab will be housed in Edmonton, Alberta. It will focus on improved understanding and prediction of high-impact weather that affects people in Canada's arctic. The Arctic lab will provide a focal point for meteorologists and research scientists from within MSC to collaborate with scientists in universities and other research organizations. The presentation will provide an overview of the Arctic lab, including the development, organization, infrastructure, partnerships and linkages, research and development goals, modeling capabilities and interactions with the other national labs. The focus will be on how the Arctic lab will conduct its business and opportunities for collaboration and cooperative projects.
There are a number of observing sites in the Canadian Arctic for which surface winds often exceed their geostrophic value. One such location is Paulatuk, NT, at 69 21N, 124 04W on Damley Bay, approximately 10km north of a steep (300m elevation change) east-west escarpment. The strongest winds at Paulatuk are invariably associated with a southerly gradient and, from surface observations, appear to have characteristics similar to downslope winds that occur over more southerly, and mountainous locales. However, little work has been done to determine if the observed wind strength is related to atmospheric stability and the vertical wind-profile in a manner that would also be consistent with downslope winds.
In order to test this relationship, three cases of varying wind strength were selected and the Advanced Regional Prediction System (ARPS) model used to simulate winds using a 2-dimensional configuration, and 1 km grid spacing. The domain was initialized using radiosonde data from Inuvik, NT (400km west of Paulatuk). Results were promising as, for all cases, the predicted maximum wind speed was close to observed values.
Results of the three case studies are presented, including a subjective comparison to expected features based on theories of downslope winds.
Since 2000 the National Center for Atmospheric Research and the Byrd Polar Research Center of the Ohio State University have been collaborating in the Antarctic Mesoscale Prediction System (AMPS): an experimental, real-time mesoscale modeling system covering Antarctica. AMPS was designed to support flight forecasters at McMurdo Station, to serve scientific programs around the continent, and to serve as a vehicle for the polar physical parameterizations. Given the complex topography of the critical forecast regions, efforts have been made to employ the highest possible resolution in the model's nested grids, to the limits of hardware and real-time constraints. Currently, the system features a 3.3-km grid over the Ross Island vicinity and 10-km grids over the western Ross Sea and the South Pole. This presentation will examine the results of such high-resolution modeling, and how it is reflected in verifications. Also addressed will be the impacts of running fine meshes over the continent.
Efforts are ongoing to evaluate the northern high-latitude performance of existing atmospheric reanalyses, with an emphasis on surface variables associated with the hydrologic budget. NCEP- 1 is known to suffer from a number of shortcomings, in particular, excessive summer precipitation and evaporation over land areas. At least in part, this overly-vigorous hydrologic cycle is associated with excessive downwelling solar radiation, and the updating of soil moisture by the forecasted precipitation. Nevertheless, in some areas of the Arctic, de- biased NCEP precipitation forecasts have considerable skill. Summer precipitation rates from NCEP-2 also have a high summer bias. Overall, the performance of ERA-15 is better relative to both NCEP-1 and NCEP-2. Early results from ERA-40 reveal further improvements in many surface variables for both land and sea ice areas (e.g., surface temperature), but precipitation forecasts are about as skillful as those from ERA-15. Validation of the NARR is underway, and will be reported on during the workshop.
The weather of the northern countries depends from statement of the atmosphere over Arctic. We decide to find out the structure of variability of SLP over Arctic, the synoptic nature of her components and temporal tendency of nearest climatic changing.
The daily SLP fields above the Arctic for 1948-2001 (NCEP/NCAR) was researched with using method of the Empirical Orthogonal Function (EOF). The basis of the investigations was a annual EOF of SLP for 1948-2001 and the climatic variability of the factors of the EOF decomposition of the daily SLP fields.
The next results was received:
It is well known that the initial meteo field computed with help of the OA method. Global forecast models require knowledge of the initial field as above the coast such and above the oceans. The task of the reproduction of the given fields above the oceans is a very difficult for case sparse observation net.
Existed OA uses a average monthly pressure meanings and their standard deviations (SD)
in the grid points and meaning of semiannual spatial autocorrelated function (SPF). Processing of
NCEP/NCAR Reanalyses (1948-2001) allowed to check a correction of their using.
At first the next hypothesis was advanced:
Presence of the Reanalyses database allowed to check advanced hypothesis. At the first stage of the processing the attempt of the season boundary definition was undertook. The EOF's of the SLP above the Arctic Ocean for 1948-2001 were computed. After comparison of the mean values of the decomposed factors the season's boundary were determined (winter semiannual - October-March and summer semiannual - April-September). Given inference was confirmed by the computed monthly repetition of the anticyclon's penetration to the North Pole.
At the second stage the semiannual quantative classification of the daily SLP fields was developed for determination of the natural spatial scales of the synoptic variability. The longitude of the axes of the main and minor antycyclone's penetration (bloking) to the North Pole was a basis of the recognition of the daily SLP variability. In a winter season the 40 types was determined with a average relative error is equal 0.76 and in a summer season - 54 types (0.70).
At the third stage the spatial autocorrelated matrixes were computed for each types for the latitude higher 47.5 degr. Analysis of the obtained matrixes allowed to detect the long-range negative correlated interconnections (from -0.15 to -0.75).
At the forth stage the modified OA method was developed using the typical autocorrelated matrixes. The method was checked on the instance of the reproduction of the SLP meaning to the location of the drifted buoy (Argos ID 9360) on February 1996 in the Arctic Ocean. The advantage of modification estimates by the meaning from 50 to 60% relatively the existed OA method.
The obtained results could be used for:
Cloud properties from the newly extended AVHRR (Advanced Very High Resolution Radiometer) Polar Pathfinder (APP-x) data set were incorporated into the atmospheric component of the Arctic Region Climate System Model (ARCSyM) in order to improve the simulation of the Antarctic surface energy balance. In the experiments, the model cloud fields were altered via the water vapor mixing ratio using cloud properties from the APP-x data set. Significant improvements in monthly mean downwelling longwave radiation at the surface were observed relative to surface measurements. In the austral summer, the use of the APP-x data set resulted in improvements as large as 30 Wm-2 at the South Pole when compared to model results without APP-x clouds. However, only a very small improvement was seen in the turbulent heat fluxes and the surface temperature. It was also found that the satellite data can be used to shorten the model "spin-up" time and may be useful in model initialization for short duration forecasts.
A combination of in situ ship measurements, synthetic aperture radar imagery and high resolution numerical modelling was used to investigate a mesoscale coastal jet radiating out from Hinlopenstretet 14 August 1996. In the meteorological analysis, light breeze and high static stability was found upstream of Svalbard. By studying results of numerical simulations in relation to the topography, upstream stagnation, flow splitting and downstream jets and wakes were identified. In the Hinlopenstretet area this flow pattern was confirmed by in situ ship measurements and the SAR estimated wind. Through investigations of sensitivity simulations, it was found that the low-level jet mainly was a result of stratified flow around dynamically steep, isolated topography, while the channelling effect of Hinlopenstretet had a minor influence. In the core of the jet the velocity was increased by a factor of about three compared to the upstream velocity.
The distribution of column ozone generally exhibits a minimum near the longitude of South America and is often referred to as a "croissant" pattern.There is also a zonal asymmetry in the occurrence of polar stratospheric clouds. At the same time (during May-November) southwestward outflow from the Tibetan High near the tropopause across the Indian Ocean deforms the base of the south polar vortex. This outflow occurs in pulses on quasi-weekly time scales. We employ the University of Wisconsin Nonhydrostatic Modeling System (UWNMS) together with global meteorological and satellite data, to investigate the possibility that the Tibetan High exerts a significant influence on the long-wave structure and constituent distributions near the southern polar vortex. A case study during August 2001 is chosen for comparison with UWNMS. Further studies include simulation of the sudden warming that occurred in late September 2002, the vertical structure of this phenomenon, as well as modeled influence of synoptic waves on ozone transport in the dynamically important layer 8-20km. Quantitative estimates of air mass mixing over the Indian Ocean using particle trajectories, will be presented.
Two case studies are presented in which the ALAPS model resolution has been increased from 27.5 km to 10 km, to assess the impact on model performance of a high resolution topography. In each case the higher resolution domains better modeled the 2 dimensional structure of the near surface wind flow. However, single station verification suggested that the lower resolution model better predicted the onset of adverse weather. A single case study will also be presented to highlight the significant impact on forecasts of raising the ALAPS upper boundary from a sigma value of 0.05 (~18 km) to 0.015 (~ 25 km). By +36 hours significant forecast differences were apparent in the mid- troposphere, with the differences having propogated to the surface by +48 hours.
Examples of the use of a neural network unsupervised learning algorithim know as self-organizing maps (SOMs) will be presented. The SOM technique will be applied to model validation studies of the Polar MM5 and ARCSyM for results from both the Arctic Regional Climate Model Intercomparison Project (ARCMIP) and using the forecast archive from the Antarctic Mesoscale Prediction System (AMPS). The presentation will focus on model validation under a variety of synoptic situations using in-situ observational data and gridded atmospheric analyses.
Organic soils (moss, peat, lichen), permafrost, as well as freezing and thawing of the active layer are common features of the physical processes in high latitudes that are frequently not included in numerical weather prediction models. The influence of organic soils, soil frost, and cross-effects between soil moisture and temperature states on the near-surface atmospheric conditions will be elucidated. It will be demonstrated that cross-effects like the Ludwig-Soret and Dufour effects that are commonly neglected may become relevant during the melting season and along the freezing line. The changes in soil temperatures and moisture caused by these cross-effects affect the exchange of heat and moisture at the atmosphere-soil interface. Organic soils allow for less super-cooled water, i.e. they freeze more quickly than mineral soils. They have a large porosity and heat capacity, and a low thermal conductivity. Ignoring these features also has impact on the heat and moisture exchange at the atmosphere-soil interface as organic soils are important surface features in high-latitude regions. It is shown that the inclusion of organic soils and the cross-effects in the land-surface models of numerical weather prediction models could provide some potential for model improvement. Unfortunately inclusion of organic soils has to be postponed until suitable 3D-data sets of organic and mineral soils are available in the resolution at least required by numerical weather prediction models.
The combination of low Froude number flow and severe topography in polar regions provides a challenge to numerical weather prediction. Special problems arise in simulating hard to resolve fine scale flow jets over the Antarctic surface resulting as flow is forced around even shallow topographical barriers. The University of Wisconsin-Nonhydrostatic Modeling System (UW-NMS) employs a variable step topography system designed to improve simulations of these flow systems. Unlike the "eta" stepped topography system, the NMS system is based on a surface coordinate step of variable depth, chosen to exactly match surface elevation. This allows the NMS to represent slopes as steep as 90 degrees while also being capable of representing even the most subtle topography. Finite differencing advection cast in vorticity/kinetic energy form and directly specifying vorticity and kinetic energy at topographical boundaries ensures the NMS of a vectorally consistent numerical treatment of flow interacting with topographical or structural barriers. As a result, competent simulations of flow around topographical obstacles are possible even in the severe Antarctic flow regimes. Simulations of classic barrier flow problems and examples of the application of NMS to the Antarctic flow system will be presented.
The Moderate Resolution Infrared Spectroradiometer (MODIS) is a 36 channel infrared imager with a horizontal resolution of up to 250 meters. It flies onboard the Earth Observing System (EOS) AM and PM platforms in a polar orbit that converge over the South Pole to provide excellent spatial coverage. Although the MODIS instrument was not specifically designed for sounding applications, scientists at the Cooperative Institute for Meteorological Studies (CIMSS) are testing MODIS radiances in retrieval algorithms. Currently, MODIS radiances are being used to retrieve total precipitable water (TPW) in clear 5 by 5 fields-of-view (FOV) yielding a horizontal resolution of 5 kilometers. Cloud detection is also possible at this resolution. Cloud-top pressure (CTP) can be retrieved using a standard CO2 intercept approach. These retrievals have been evaluated against co-located measurements at the Atmospheric Radiation Measurement Program's Cloud and Radiation Testbed (ARM/CART) site in Oklahoma for a limited number of cases with snow cover. They have also been assimilated into a 60-kilometer, polar stereographic version of the CIMSS Regional Assimilation System (CRAS) for one case commencing December 6, 2000. Boundary conditions were provided by the National Center for Environmental Prediction's (NCEP) Global Forecast System (GFS). The retrievals were compared to TPW and CTP calculated from the forecast model during a 24 hour forecast initialization period. Comparison statistics indicate that the MODIS retrievals tend to dry the forecast model in the coastal regions surrounding Antarctica and reduce cloudiness throughout the model domain. Smaller differences were present as far north as latitude 45 S. The retrievals had little effect over the central continent where the atmosphere is typically very dry.
A 48-kilometer CRAS forecast grid has been set up to generate real-time forecasts for Antarctica. CTP and TPW from the AM and PM satellites will be inserted into a 12-hour forecast spin-up cycle to provide initial water vapor, cloud and precipitation fields. Forecasts will be compared to a control forecast without MODIS information. Cloud-track and water vapor winds from MODIS will be added in the future. In addition, the CRAS model physics package is being evaluated for use at high latitudes.
Greenland, the largest island in the Northern Hemisphere, and 90% covered by ice, has highly dynamic weather patterns, the forecasting of which are critical to human activities which are increasingly occurring on the inland ice sheet. Remote field camps would benefit greatly from a tailored weather forecasting capability in contrast to the general forecasts currently available. The Greenland Weather Prediction IniTiativE using a Regional AtmospheriC model (PITERAC) is a significant step in this direction ( http://polarmet.mps.ohio-state.edu/PolarMet/grldnwp.html). PITERAC employs the state-of-the-art Polar MM5 model that has a sophisticated description of physical processes in the atmosphere over polar regions to provide weather prediction for human activities (e.g., scientific, recreation) on the Ice Sheet. The model runs in two nested domains, with the coarse domain covering most of the Arctic region (including Alaska) at 60km and the fine domain spanning Greenland at 20km resolution. A third finer domain covering part of the Ice Sheet can be implemented if required. PITERAC's initial fields and lateral boundary conditions come from the global NCEP aviation model (now Global Forecast System) output. PITERAC now runs once a day at 00UTC, and will run twice a day, to produce a 72 hour forecast. Comparisons between PITERAC output and Greenland Climate Network automatic weather station (GC-Net AWS) observations are presented to verify PITERAC's performance. A real-time forecast verification system is planned.
In May 2003 the upper boundary condition (UBC) in the Antarctic Mesoscale Prediction System (AMPS) was changed from a radiative type to a nudging type (developed by Helin Wei of the PMG), and the model top was raised from 100 hPa to 50 hPa. Previous studies have shown that severe internal gravity wave reflection is found at the model top near steep topography such as the Antarctic coastal regions when the radiative UBC is employed. This spurious signal can cause temperature biases as large as 20C near the model top, and can propagate to the surface resulting in large distortions of the mean sea level pressure field (up to 15 hPa too high at times). The nudging UBC nudges the temperature and perturbation pressure in the 8 highest model sigma levels toward the AVN forecast fields, and coupled with the higher model top, is expected to significantly reduce internal gravity wave reflection and associated problems. In this study, June 2002 (before the change) and June 2003 (after the change) are compared in order to quantify the impact of the new UBC. Specifically, vertical motion fields are compared and upper-tropospheric temperatures are assessed with respect to radiosondes and the ECMWF operational analysis.
Numerical modeling studies of the polar regions are subject to errors derived from the application of cloud and cloud radiative schemes that are appropriate for the tropics and middle latitudes but not designed for high-latitude meteorology. Only limited in-situ observations have been taken of the optical properties of clouds over Antarctica. Previous efforts to detail the effectiveness of global and regional atmospheric models at simulating the Antarctic hydrologic cycle and the accompanying radiative effects often find serious deficiencies within the parameterizations. Fortunately, improved satellite retrieved cloud properties have recently become available for Antarctica. We are using satellite-retrieved radiation and cloud properties from the Moderate Resolution Imaging Spectroradiometer (MODIS) and Advanced Very-High Resolution Radiometer (AVHRR) in conjunction with surface observations to verify and improve the cloud parameterization within the Polar MM5 and its eventual successor, the polar version of the Weather Research and Forecasting (WRF) model. The Polar MM5, developed by the Polar Meteorology Group, uses the Reisner explicit microphysics scheme to prognostically determine cloud water, cloud ice and precipitation. Polar MM5 has been previously found to undersimulate cloud fraction over interior Antarctica. The satellite-derived cloud properties that are being considered include cloud optical depth, particle effective radius, fraction, and height. Existing MODIS cloud products are utilized whenever possible. These will be combined with radiative fluxes and surface observations to diagnose the accuracy and biases of the Polar MM5 simulated clouds and radiation. Based upon the results, improvements to the parameterizations will be devised and tested. Furthermore, a parameterization for clear-sky precipitation will be included in Polar MM5. In addition to supporting the real-time weather forecasts with Polar MM5, via the Antarctic Mesoscale Prediction System, the expected improvements will be included in a new multi-year climatological simulation to detail the response across Antarctica of clouds and radiation to the cycles of the El Nino-Southern Oscillation.
To gain a better understanding of Arctic climate change over recent decades, a high-resolution comprehensive Arctic System Reanalysis (ASR) of the atmosphere, ocean and land surface is planned. To best understand the processes and feedbacks impacting climate change, we require high-quality representations over a long time series of temporally and dynamically consistent fields. Reanalyses combine a short-term model forecast with all available observations (from the ground, rawinsondes, aircraft, satellites, etc.) to provide optimum analyses of directly measured fields. Short-term forecasts also produce fields that are incompletely monitored or unmeasured (precipitation, evaporation, clouds, etc.) The planned ASR allows a synthesis of several Arctic field programs (SHEBA, LAII/ATLAS, ARM, ...) in a physically consistent framework. The high-resolution ASR benefits from the lessons learned during the earlier global reanalyses [National Centers for Environmental Prediction (NCEP)/National Center for Atmospheric Research (NCAR) and ECMWF]. The ASR will eventually encompass a highly detailed treatment of the coupled atmosphere-land-ocean system. The work toward this goal will begin with the atmosphere, with the land and ocean components to follow. To begin the groundwork for the ASR, the Weather Research and Forecasting (WRF) model is being specially adapted for the polar regions. The WRF is currently in the later stages of development by NCAR and NCEP, and early versions of the model are already available. The WRF is the successor to the Penn State/NCAR Fifth Generation Mesoscale Model (MM5). The polar-optimized version of the new model is the natural choice as the algorithm for the upcoming Arctic Reanalysis. NCAR is developing a robust three-dimensional variational assimilation for WRF to best incorporate the modern Arctic in-situ and remote sensing data. Moreover, a high-latitude physics package has been previously developed for MM5 by the Polar Meteorology Group of the Byrd Polar Research Center. Based upon this earlier work, improvements to the simulation of cold cloud and the implementation of mixed sea ice and open water grid points are among the adaptations being applied in Polar WRF. The current model, Polar MM5 is used for a variety of applications including real-time Antarctic numerical weather prediction, climatic studies of the El Nino-Southern Oscillation (ENSO) linkage to Antarctica, and studies of the hydrology of Arctic river basins. Verification of Polar MM5 versus observations indicates that the model performs well for the high latitude regions of both hemispheres.
The Antarctic Meteorological Research Center (AMRC) of the University of Wisconsin-Madison has a part of its grant from the National Science Foundation (NSF), United States Antarctic Program (USAP), to collect and disseminate weather data for atmospheric and related scientific studies. Data collections include, but are not limited to, historical logs and records of manned and unmanned surface observations, composite satellite images, as well as images of various icebergs, which have been tracked since 2000. These data sets are of great value for historical record, scientific study, and the daily lives of those who travel to and from Antarctica. The various land, sea, and air rescues by the USAP over the past few years have also been reliant on the AMRC's available data. While many of the AMRC's data collections are used by those pursuing numerical weather prediction, there are many more observations that may not be included in these efforts or known about to numerical modelers when this data may be quite valuable. Adding additional data sets to the models will also aid Aviation Technical Services (ATS) and other forecasters when they predict the upcoming weather by providing a more accurate description of the current situation. While there will be some unique issues when dealing with these observations, such as irregularities with respect to time, communication delays, non-traditional observational sources, and a need to consider for which uses (research or science) the data would be most valuable, this data should be made aware of, and considered for inclusion into various numerical weather prediction outlets. This presentation will introduce Antarctic meteorological data sources, outline some example applications demonstrating the critical value of the observations, and discuss some of the issues faced when utilizing Antarctic meteorological observations.
Polar MM5 was developed by the Polar Meteorology Group at the Byrd Polar Research Center ( http://www-bprc.mps.ohio-state.edu/PolarMet/pmm5.html), and is a modified version of the PSU/NCAR fifth-generation Mesoscale Model. Polar MM5 is used to simulate the atmospheric state over the Arctic river basins ( http://polarmet.mps.ohio-state.edu/PolarMet/arcticnwp.html). The NCAR LSM is implemented into the Polar MM5 to describe the detailed land surface characteristics. Two domains are used with a resolution of 60 km covering North America and Eurasia and including the North Pole. The vertical discretization consists of 28 irregularly spaced levels in s-coordinates from the surface up to 10 hPa and the rigid upper boundary condition is used. The model physics options are: mixed phase explicit moisture; Grell cumulus scheme; CCM2 atmospheric radiation scheme; and the ETA planetary boundary layer scheme. Parallel 10-day simulations of the Polar MM5 and the original MM5 reveal that the Polar MM5 produces better near-surface variable forecasts than the original MM5 over Arctic land areas, which confirms that the modified physical parameterization schemes are appropriate for these regions. The real-time Polar MM5 coupled with NCAR LSM forecasts can capture much of the near-surface variability, especially for the near-surface winds that are usually hard to predict over complex land surfaces, but there is a slight cold bias near the surface. Possible reasons for the cold bias of the near-surface temperature are that there is too little downward longwave radiation, the ETA PBL scheme is deficient or the coupled model overestimates the surface albedo.
To investigate the cold bias of the near-surface temperature the Polar MM5 version 3.6.1 is used to make sensitivity runs for eleven cases from 00 UTC 01 December 2000 to 00 UTC 12 December 2002 over the North American domain. The 1o horizontal resolution NCEP Global Final Analysis surface and upper air operational analyses are used to provide the initial and boundary conditions for the model. The MM5 is used to produce a sequence of short duration (48 h) simulations with the first 24h of each simulation being discarded for spin-up reasons. Eleven runs are made to evaluate the contributions of NOAH LSM, NCAR LSM, PBL schemes, albedo, polar physics and shallow convection over the Arctic river basins. New features of the NOAH LSM are frozen-soil physics and snowpack physical processes. The simulated results are compared with the surface observations and with each other.
Atmospheric motion vectors (AMVs) from geostationary satellite data have been used in the Canadian Meteorological Centre (CMC) global and regional numerical weather prediction systems for many years. They do not provide winds information at high latitudes because of geometry aspects. Previous studies have shown the potential of polar orbiting satellites derived winds in the high latitudes. An impact trial is underway for the 2003 summer period. This study will evaluate the impact of MODIS winds in the CMC 3Dvar 6-hourly assimilation and NWP system in preparation for an operational implementation. The presentation will focus on the observations selection procedure, the MODIS winds characteristics compared to the analysis program first guess field and the evaluation of the forecast against radiosondes and against verifying analyses.
Precipitation in Iceland during a period of 12 months is
simulated with the PSU/NCAR MM5 model. The results are
compared with precipitation estimated by a statistical model
based on observations and a number of topographic and
geographic predictors.
The simulated precipitation pattern agrees with the
statistical model in areas where data is available and
gives a credible precipitation pattern in data-sparse
mountainous regions.
The simulation is however in general overestimating the
precipitation, but the magnitude and the seasonal and
geographical distribution of the overestimation indicate that
the overestimation is to a large extent associated with
observation errors, mainly wind-loss of solid precipitation.
Indications of errors due to coarse horizontal resolution
were also found.
Keywords:
Precipitation mapping, statistical model, data-sparse
terrain, MM5, Iceland, wind-loss.
A multi-year record of low and high daily surface temperatures for selected points in interior Alaska were compared to the associated daily predictions provided by the NWS/NCEP Medium Range Forecast Model Output Statistics (MRF MOS) to test the null hypothesis that the differences between observed minimum and maximum temperatures and the MRF MOS simulation are no greater than chance. Consistent with previous work, MRF MOS showed statistically discernible reduction in forecast skill over climatology for 3 days or more in advance. Although the temporal distribution of the forecast error pattern is mixed, the analysis shows a general warm bias in the winter, and a somewhat cooler bias in the summer. We suggest several possible sources of error which include shallow inversions and the radiative impact of cloud cover. An analysis of value-added subjective forecasts compared to direct use of MRF MOS simulation results is also presented.
The NWS is rapidly moving into the era in which meteorologists will be making weather forecasts by manipulating 2-D digital grids of sensible weather elements. The system of technologies making this possible is known as the Interactive Forecast Preparation System (IFPS). Forecasters have long utilized a forecast process consisting of meteorological analysis of observations, NWP, and knowledge of associated relationships (e.g. terrain, antecedent environmental conditions) to derive a mental picture of the forecast for their areas of responsibility. Once conceived, this mental picture is generally turned into words by way of simple word processing. IFPS will radically change the input, mechanics, and output of this forecast process. And by its very nature, the underlying gridded database will require resolution sufficient to describe local effects, terrain, etc. Initializing this high-resolution database in a meteorologically consistent manner is proving to be one of the most challenging aspects of the IFPS implementation in Alaska. This presentation will discuss the issues, potential solutions, and implications of high resolution forecast grids in the IFPS era.
In meteorological model evaluation for environmental sciences, it has been common practice to compare model output at observational sites with point measurements, and then to derive statistics from the comparison to indicate the performance of the meteorological model. In this presentation, we present two case studies in which the point-to-point model-observation comparison is combined with the comparison between the model simulated and the observed meteorological processes to provide an evaluation of a mesoscale model for air-quality and Arctic applications. The observations used in the two case studies are from the Texas Air Quality Study 2000 field experiment and SHEBA 1998 field experiment. We wish to illustrate with the case studies that although point-to-point comparisons between model output and observations are an important step in the model evaluation procedure, a comparison of the forecasted processes with the observations is as, if not more, important as a point-to-point comparison. Such a comparison is very helpful in developing conceptual models of the local processes that affect the observed atmospheric variables and parameters beyond what can be inferred only from the observations which have their own limitations. Moreover, the comparison between the model simulated and the observed processes can provide needed information on the strategic improvement of future observational experiments for the purpose of meteorological model improvement.
The influence of blowing snow on the intensity of strong katabatic winds is examined in the mesoscale model MAR (Modèle Atmosphérique Régional). Indeed an increase of the katabatic air density is expected in case of erosion of snow by the wind over a slope. This is due to the additional weight of the blown snow particles embedded in the katabatic flow and to the cooling caused by an increase of snow sublimation. Previous studies have already mentioned that impact, especially for katabatic wind speeds larger than 25 m/s (Gosink, 1989).
MAR has been forced at its lateral boundaries by the European Re-Analyses (ERA-15) and strong wind events have been simulated. The model domain include a substantial part of the antarctic continent, with the Amery Ice Shelf, Wilkes Land, Adélie Land, Victoria Land and the Ross Ice Shelf. The horizontal resolution is 40 km. A significant amplification of strong katabatic wind events is simulated when the physical processes accompanying blowing snow are included in the model. Sensitivity tests to each process have been performed. The main contribution to the wind speed increase is the additonal weight due to blown snow particles.
As part of an observational program to estimate freshwater fluxes through the Canadian Archipelago, a multiply-nested mesoscale model is used to estimate wind stress in the Nares Strait and Smith Sound channels west of Greenland. The high-resolution model fields will be compared, where possible, to other available model products and observations. Preliminary results are presented and discussed.
Short-to-medium range weather prediction and climate forecasts are often based on rather crude representation of lower atmosphere boundary conditions in polar regions. Especially in the Arctic Ocean covered by the multiyear Ice Pack, the lower atmosphere model boundary forcing does not reflect the dynamic sea ice and ocean environ-ments. It is generally accepted that ocean - sea ice - atmosphere interactions and feedbacks are critical to arctic climate and its variability. At the same time, both global and regional climate models commonly do not account properly (or at all) for such feedbacks, mainly because they do not include fully predictive sea ice and ocean mod-els in their polar weather / climate simulations. Finally, fully coupled global models are typically not ready to address the above issues due to both their inadequate representation of the Arctic region and often their non-arctic focus.
In this talk, we argue that a regional arctic climate model, consisting of state-of-the-art atmosphere, ocean, sea ice, and land components, utilizing the available modern computer technology, and coordinated with the ongoing and planned observations is needed to address some of the outstanding problems in arctic weather and climate predic-tion. We report on the progress in modeling the Arctic Ocean and sea ice, outline an approach for the development of a regional climate model, and discuss possible benefits to research, commercial and defense activities associ-ated with the Arctic Ocean.
Model developments over the last few years at the European Centre for Medium-Range Weather Forecasts (ECMWF) of relevance to high latitudes will be presented. The presentation will follow a general pattern of error diagnostics, attribution, identification of missing or incorrectly described physical processes, development of an improved model, testing and validation. Comparison of results of the two ECMWF reanalysis, ERA15 (1979-1993) and ER40 (195709-200208), summarizes the progress made over the last ten years.
Polar areas also represent a challenge for data assimilation. The relative scarcity of conventional data gives an added importance to proper handling of remote sensing observations, in particular the microwave and infrared sounders from polar orbit platforms. However, in such extreme cold and stable conditions, the quality of the background field, the handling of model and observation biases, and the accuracy of the linear forward models used to go from model space to satellite radiance space, have to be scrutinised carefully and lead often to problems in the analysis. Those will be illustrated using examples from ERA40.
We will present a summary of an extensive validation effort of ERA40 precipitation, evaporation and surface temperature over the Mackenzie River Basin. In addition, we will present point validation of surface radiative, latent and sensible heat fluxes against multiyear tower measurements located in the Arctic region. The validation of ERA40 (the entire period will be available on-line to the research community in autumn 2003) will be an on-going community effort in the years to come. Its results will help to characterize biases in the ERA40 products, which in turn will lead to a better usage of that data. Additionally, validation will help identify model deficiencies, data assimilation problems, or errors in some of the data sources used in ERA40.
Finally, the performance of operational and ERA40 medium-range forecasts in polar regions will be compared with mid-latitude results.
After reports from both ECMWF and the DAO of successful initial assimilation of winds derived from MODIS imagery for a test period in March 2001, MODIS winds have been generated on a quasi-operational basis at CIMSS since July 2002. We report results from more extensive assimilation experiments carried out during the first two months of near-real time data dissemination. The results confirm the initial hypothesis that there would be a modest positive impact of the data on the skill in the northern hemisphere, and a substantial impact in the southern hemisphere. The sign of the southern hemisphere impact depends on the choice of verifying analysis, the verification region, and to a lesser extent on the configuration of the experiment. Paradoxically, there is a robust positive impact in the southern mid-latitudes which are not covered by the observations, while the impact in the observed high-latitude region seems to be predominantly negative. This indicates that while the information in the MODIS winds is real and of value to the assimilation, the verification over the high southern latitudes remains problematic.
Polar wind analyses for numercial weather prediction have long been hampered by the lack of wind observations in these regions. Modis Polar winds (Atmospheric Motion Vectors AMV) that are derived at the University of Wisconsin (Madison) by tracking structures in successive swaths from the polar-orbiting Moderate Resolution Imaging Spectroradiometer (MODIS) are therefore a promising candidate to fill this gap due to their capabiliby of providing unprecedented coverage of thepPolar regions. Longterm data assimilation experiments demonstrate that Modis Polar winds alter the mean polar wind analysis considerably and that Modis Polar winds have a positive impact on weather forecasts over the whole Northern Hemisphere. These results will be presented together with studies, which show that key analysis errors were reduced. The problem of the observed slow speed bias which is often associated with AMV observations will be addressed in a special case study. It can be shown that few observations, that are biased against the model first guess, have a dedicated impact on the analysed vertical movement and therefore on the tendency of a developing pressure system.
Over the past year, we have developed the so-called 'Arctic MM5' real-time operational mesoscale modeling system. The system ingests Alaskan surface, satellite and upper air observations plus NCEP Eta-model analyses several times daily in real-time and incorporates this data into a quintuple-grid hierarchy of numerical forecasts. The forecast duration during the period in question (May 2002-June 2003) is a maximum of 39 hours and the system utilizes a minimum grid resolution of 5 km over local domains encompassing Fairbanks, Anchorage, and Barrow-ARM/CART. Post-processed products include an integrated aircraft icing forecast algorithm that also incorporates observed data of numerous types, and other aviation impact variables.
The real-time system provides an appropriate platform for in-depth evaluation of the adequacy of the Arctic MM5 as well as changes to the model configuration implemented over time. For the purpose of such evaluation, we have implemented verification routines utilizing standard domain-averaged and point specific skill score statistics, including RMSE, bias, S1 skill score and the equitable threat score.
In our presentation, we will summarize the characteristics of the real-time system in, including improvements/changes implemented through the May 2002-June 2003 period. The primary focus of the presentation, however, will be an evaluation of the forecast performance of the modeling system on the various scales, based on the aforementioned skill score statistics. Implications of the scores for high latitude mesoscale modeling and eventual use of such models in reanalysis systems will be briefly discussed.