Monthly Snowfall Research Articles

Statistical Modeling of Monthly and Seasonal Michigan Snowfall Based on Machine Learning: A Multiscale Approach

Abstract Snow is an important component of Earth’s climate system, and snowfall intensity and variation often significantly impact society, the environment, and ecosystems. Understanding monthly and seasonal snowfall intensity and variations is challenging because of multiple controlling mechanisms at different spatial and temporal scales. Using 65 years of in situ snowfall observation, we evaluated seven machine learning algorithms for modeling monthly and seasonal snowfall in the Lower Peninsula of Michigan (LPM) based on selected environmental and climatic variables. Our results show that the Bayesian additive regression tree (BART) has the best fitting (R2 = 0.88) and out-of-sample estimation skills (R2 = 0.58) for the monthly mean snowfall followed by the random forest model. The BART also demonstrates strong estimation skills for large monthly snowfall amounts. Both BART and the random forest models suggest that topography, local/regional environmental factors, and teleconnection indices can significantly improve the estimation of monthly and seasonal snowfall amounts in the LPM. These statistical models based on machine learning algorithms can incorporate variables at multiple scales and address nonlinear responses of snowfall variations to environmental/climatic changes. It demonstrated that the multiscale machine learning techniques provide a reliable and computationally efficient approach to modeling snowfall intensity and variability.

Intra‐annual snowfall variability in the central United States

AbstractSnowfall in the central United States has been shown to be highly variable at interannual timescales and subjected to long‐term linear trends. However, limited understanding is available about the intra‐annual variations and trends in snowfall within this region. Using a selection of GHCN daily snowfall observations and statistical analyses, this study develops a monthly climatology of snowfall characteristics in the central United States from 1948 to 2021. Here we find larger magnitude snowfall events have a greater relative frequency to seasonal snowfall late in the snow season, have increased in frequency in the western and northern regions of the basin, and contribute to a bimodal distribution of monthly snowfall in these sub‐regions. While snowfall has significantly increased over time within the northern and western regions of the domain by as much as 83%, snowfall has significantly decreased by much as 46% in the southern and eastern regions. Further, March snowfall totals and event frequencies have declined across the vast majority of the study domain, with maximum decreases in eastern IA of approximately 50%. When coupled with increases in snowfall and snowfall events in February, this suggests a shift in the snowfall season with time in this portion of the study region. Such results may be of particular interest to snow and water resource managers where integration of the nonstationarity of snowfall events and totals with time into management plans may yield better solutions.

Correlation Analysis and Evaluation for Snowfall Data of ECMWF and Honam Region in Korea

As the temperature of the Arctic rises due to global warming, the polar vortex, which is the cold air of the Arctic, has moved southward, which causes the occurrence of unusual cold waves and heavy snowfall in the mid-latitude regions (North America, Europe, and Asia). Since this phenomenon is known to induce damage to people and properties, it is very important to analyze the process and cause of snowfall as well as its predictability in order to lessen if not prevent further damage. While there are many studies that utilize climate models in relation to rainfall, snowfall studies are lacking. This study evaluated the domestic applicability of snowfall data of the European Centre for Medium-Range Weather Forecasts (ECMWF) seasonal forecast data before performing the snowfall generation process and forecasting research. We analyzed the correlation between monthly fresh snowfall of the 13 stations in the Honam region, which is one of the heaviest snowfall areas in Korea and of monthly snowfall of the ECMWF. After clustering 13 stations using hierarchical cluster analysis, correlation analysis was performed and its accuracy was determined using a confusion matrix. As a result, it was concluded that the applicability of the ECMWF seasonal forecast data is sufficient based on the high correlation and accuracy results between the monthly fresh snowfall data and the monthly snowfall data of the ECMWF. The results of this study are expected to be utilized as basic data for predicting future heavy snowfall to verify its impact and assess the vulnerability to heavy snowfall damage.

New insights into the world's longest series of monthly snowfall (Parma, Northern Italy, 1777–2018)

AbstractThe emergence of decreasing trends in snowfall frequency and snow depth highlights the challenges arising from shifts in snow regimes. In particular, snow‐dependent southern Europe regions may be negatively impacted by such changes. Snow regimes strongly influence the water availability in reservoirs and groundwater. This study presents the world's longest series of monthly snowfall (fresh snow depth, snow days per year, days with snow on the ground) for the Parma Observatory, northern Italy (44°48′N and 10°19′E) between 1777 and 2018. These datasets were collected over the centuries using early rain gauges and recently, SIAP tipping bucket models. Sources of inhomogeneity were identified and analysed through the study of metadata and of nonparametric tests on datasets. Homogenized time series were obtained after correcting the observed snowfall data for the instruments response and localization characteristics, for the observing practices with the help of the standard normal homogeneity tests. The Buishand and the Mann–Kendall tests were further applied to check the correction and detect possible monotonic trends. Over the study period, days per year of snow decreased after the change point detected in 1897, in association with a rise in surface air temperature (including a distinct urban warming trend). In this study high numbers of snowy days per year (hereafter referred as snowfall frequency) and snow depth values during the latest phase of the LIA at Parma are consistent with a cycle of minimum solar activity, suggesting that enhanced, solar‐induced blocking activity dominated, with the arrival into the Mediterranean of large cold air masses developing over Siberia and northern Europe. Over the last century, snow appears to have resonance‐like characteristics as similar trends were observed over the Northern Hemisphere, where the extent of the snow cover has been reported to markedly decline in the transition spring and autumn seasons.

Spatiotemporal variability of snowfall and its concentration in northern Xinjiang, Northwest China

This study investigated the spatial and temporal variability of snowfall and its concentration across northern Xinjiang, Northwest China based on the corrected dataset (derived from 26 stations) using a wet-bulb temperature method during the snowfall hydrological years from 1961 to 2017. The snowfall concentration was analyzed by means of three indices: the snowfall concentration index (CI), snowfall concentration degree (SCD), and snowfall concentration period (SCP). The results demonstrated that the snowfall and temperature increased by 5.69 mm and 0.35 °C per decade, respectively, in northern Xinjiang during the time period from 1961 to 2017. Maximum snowfall occurred at a critical temperature ranging from − 2 to 1 °C. The increased snowfall mainly took place in northern Xinjiang during winter. The distribution of the high snowfall CI value indicated that the daily snowfall showed a high irregularity in the northern slope of the Tianshan Mountains and the northern Altay Mountains, in which more than 77% snowfall was contributed by the 25% snowfall days in a year. The results of SCP revealed that the snowfall concentrated in northern Xinjiang during December, with an earlier arrival in the north and east of northern Xinjiang than in the western part. The SCD results proved that the monthly snowfall was more concentrated in the northern Altay Mountains, Ili Valley, and Boertala Valley than over the eastern part of northern Xinjiang during a year. The snowfall in almost all stations increased in concentration throughout the year. The obtained study results could provide scientific reference for future water resource management and snow disaster prevention under a warming climate.

Causes of Weather-Related Crashes in Salt Lake County, Utah

In 2010, the World Health Organization declared 2011 to 2020 the Decade of Action for Road Safety in response to the enormous toll that roadway crashes take on individuals, communities, and national economies. More than 6.2 million vehicle crashes occurred in the United States in 2015, resulting in around 2.4 million injuries and more than 35,000 fatalities. Motor vehicle crashes in the United States in 2015 had an economic cost of $836 billion, equal to nearly 5 percent of the U.S. gross domestic product. For this study, more than 122,000 crashes occurring in Salt Lake County, Utah, from 2012 to 2016 were analyzed to see how weather conditions affected the number and type of crashes observed. Approximately 12 percent of crashes occurred under “adverse weather conditions.” Weather-related crashes were most likely in the winter season and were very highly correlated with monthly snowfall. Although the greatest number of crashes occurred during evening rush hours, adverse weather-related crashes made up a larger percentage of crashes from 5 to 7 a.m. Although excessive speed was much more likely to be a factor in crashes during adverse weather, such crashes, on average, were less severe. Roadway slope also played a significant role, with slight increases causing more crashes.

Application of random forest time series, support vector regression and multivariate adaptive regression splines models in prediction of snowfall (a case study of Alvand in the middle Zagros, Iran)

We have conducted a case study to investigate the performance of support vector machine, multivariate adaptive regression splines, and random forest time series methods in snowfall modeling. These models were applied to a data set of monthly snowfall collected during six cold months at Hamadan Airport sample station located in the Zagros Mountain Range in Iran. We considered monthly data of snowfall from 1981 to 2008 during the period from October/November to April/May as the training set and the data from 2009 to 2015 as the testing set. The root mean square errors (RMSE), mean absolute errors (MAE), determination coefficient (R 2), coefficient of efficiency (E%), and intra-class correlation coefficient (ICC) statistics were used as evaluation criteria. Our results indicated that the random forest time series model outperformed the support vector machine and multivariate adaptive regression splines models in predicting monthly snowfall in terms of several criteria. The RMSE, MAE, R 2, E, and ICC for the testing set were 7.84, 5.52, 0.92, 0.89, and 0.93, respectively. The overall results indicated that the random forest time series model could be successfully used to estimate monthly snowfall values. Moreover, the support vector machine model showed substantial performance as well, suggesting it may also be applied to forecast snowfall in this area.

Using GRACE and climate model simulations to predict mass loss of Alaskan glaciers through 2100

ABSTRACTGlaciers in Alaska are currently losing mass at a rate of ~−50 Gt a−1, one of the largest ice loss rates of any regional collection of mountain glaciers on Earth. Existing projections of Alaska's future sea-level contributions tend to be divergent and are not tied directly to regional observations. Here we develop a simple, regional observation-based projection of Alaska's future sea-level contribution. We compute a time series of recent Alaska glacier mass variability using monthly GRACE gravity fields from August 2002 through December 2014. We also construct a three-parameter model of Alaska glacier mass variability based on monthly ERA-Interim snowfall and temperature fields. When these three model parameters are fitted to the GRACE time series, the model explains 94% of the variance of the GRACE data. Using these parameter values, we then apply the model to simulated fields of monthly temperature and snowfall from the Community Earth System Model, to obtain predictions of mass variations through 2100. We conclude that mass loss rates may increase between −80 and −110 Gt a−1 by 2100, with a total sea-level rise contribution of 19 ± 4 mm during the 21st century.

Creation and Validation of a Comprehensive 1° by 1° Daily Gridded North American Dataset for 1900–2009: Snowfall

AbstractThis study details the creation of a gridded snowfall dataset for North America, with focus on the quality of the interpolated product. Daily snowfall amounts from National Weather Service Cooperative Observer Program stations and Meteorological Service of Canada surface stations are interpolated to 1° by 1° grids from 1900 to 2009 and examined for data record length and quality. The interpolation is validated spatially and temporally through the use of stratified sampling and k-fold cross-validation analyses. Interpolation errors average around 0.5 cm and range from less than 0.01 to greater than 2.5 cm. For most locations, this is within the measurement sensitivity. Grid cells with large variations in elevation experience higher errors and should be used with caution. A new gridded snowfall climatology is presented based on in situ observations that capture seasonal and interannual variability in monthly snowfall over most of the North American land area from 1949 to 2009. The Community Collaborative Rain, Hail and Snow Network is used as an independent set of point data that is compared to the gridded product. Errors are mainly in the form of the gridded data underestimating snowfall compared to the point data. The spatial extent, temporal length, and resolution of the dataset are unprecedented with regard to observational snowfall products and will present new opportunities for examining snowfall across North America.

Seasonal Forecast of Local Lake-Effect Snowfall: The Case of Buffalo, U.S.A.

The climate and weather patterns of Buffalo (New York, U.S.A.) are strongly influenced by thecity’s proximity to Lake Erie. Total monthly snowfall in Buffalo is forecasted using neural network techniques(Multi-Layer Perceptron = MLP) and a multiple linear regression (LR) model. The period of analysis comprises 28 years from January 1982 to December 2009. Input data include: zonal wind speed (u-wind), meridional wind speed (v-wind), air temperature, the geopotential height (GPH) over Lake Erie and the surrounding regions at the 1000 mb -, 925 mb -, 850 mb -, and 700 mb - levels as well as the surface pressure and air temperature, mean water temperature, lake surface water temperatures (LSWT) and the amount of ice coverage of Lake Erie; the 500 mb GPH over James Bay, Canada; and the surface pressure over the North-Central Great Plains. Different lead times of the input variables are tested for their suitability. The most accurate result is obtained by using the MLP with an optimum lead time approach (lead times vary for the different input variables between one and six months). The results of the MLP with six months lead time are in good agreement with observed precipitation records over the study period.

Neural Network Based Seasonal Predictions of Lake-Effect Snowfall

In the case of the city of Buffalo (New York, United States), located on the eastern shore of Lake Erie and, therefore, strongly influenced by the lake-effect, total monthly snowfall was predicted one to six months in advance. For this, neural network (NN) techniques, specifically a multi-layer perceptron, as well as a multiple linear regression (LR) model were applied. The period of analysis comprised 28 years from January 1982 to December 2009. Input data included surface air temperature; the temperature difference between the lake surface water temperature (LSWT) and the 850 hPa air temperature; the u-component of the wind (u-wind) and the v-component of the wind (v-wind), geopotential height (GPH) over Lake Erie and the surrounding regions at the 1000, 925, 850 and 700 hPa levels as well as the surface pressure; the 500 hPa GPH over James Bay, Canada; the surface pressure over the Great Plains; and the mean water temperature and LSWT of Lake Erie, as well as the amount of ice cover. Moreover, several teleconnection indices were implemented: the North Atlantic Oscillation (NAO), the Arctic Oscillation (AO), the Pacific/North American (PNA) pattern, the Southern Oscillation Index (SOI) and the Pacific Decadal Oscillation (PDO). Different lead times for the input variables were tested for their suitability. The most accurate result was obtained using the NN with an optimized one-month lead time approach (lead times varied between one and six months for the different input variables). R ésumé [Traduit par la rédaction] Dans le cas de la ville de Buffalo (New York, États–Unis), située sur la rive est du lac Érié et donc fortement influencée par l'effet de lac, nous avons prévu la chute de neige mensuelle totale de un à six mois à l'avance. À cette fin, nous avons appliqué des techniques de réseau neuronal, plus précisément un perceptron multicouche, ainsi qu'un modèle de régression linéaire multiple. La période d'analyse s’étendait sur 28 années, de janvier 1982 à décembre 2009. Les données d'entrée consistaient en : la température de l'air à la surface; la différence entre la température de l'eau à la surface du lac et celle de l'air à 850 hPa; la composante u du vent (vent–u) et la composante v du vent (vent-v), la hauteur géopotentielle au-dessus du lac Érié et de la région environnante aux niveaux 1000, 925, 850 et 700 hPa ainsi que la pression à la surface; la hauteur géopotentielle à 500 hPa au-dessus de la baie James au Canada; la pression à la surface dans les Grandes Plaines; et la température moyenne de l'eau ainsi que la température de l'eau à la surface du lac Érié de même que l’étendue de la couverture de glace. De plus, nous nous sommes servis de plusieurs indices de téléconnexion : l'oscillation de l'Atlantique Nord, l'oscillation de l'Arctique, la téléconnexion Pacifique-Amérique du Nord, l'indice d'oscillation australe et l'oscillation décennale du Pacifique. Nous avons testé la convenance de différents délais de démarrage pour les variables d'entrée. Le résultat le plus précis a été obtenu en utilisant le réseau neuronal avec une approche de délais de démarrage optimisé d'un mois (les délais de démarrage variaient entre un et six mois pour les différentes variables d'entrée).

Relationship between winter severity and survival of mule deer fawns in the Peace Region of British Columbia

This extension note summarizes results of an ongoing study to measure the survival of mule deer ( Odocoileus hemionus) fawns through their first winter in the Peace Region of British Columbia. Each spring since 1991, mule deer were counted and classified by sex and age by driving along roads that go through areas known for providing good winter range. Seven transects were driven for a total distance of 205.3 km. The number of fawns observed is expressed as a ratio of fawns per 100 does. Average monthly air temperature and total monthly snowfall data from November to the following April from 1991 through 2008 were obtained from Environment Canada’s website. A winter severity index (WSI) integrating temperature and snowfall data was calculated and correlated to the observed fawn-to-doe ratio. A statistically significant relationship between this ratio and the WSI was obtained through regression analysis. Resulting data confirms previous research that showed survival of fawns through their first winter is higher in milder winters. For the same time period, we also compared WSI values for the Peace Region with those from three other areas in British Columbia. Results indicate that, on average, the Peace Region experienced harsher winter conditions than the other regions. The variation of the WSI was also much greater in the Peace Region. These results have implications for mule deer management in this region.

Multidimensional Analysis of Snow Cover Data in South Dakota Diversity of Landscapes

Snow distribution in SD was studied with Factor Analysis (FA) of monthly total snowfall [in]. The long-term data obtained from the High Plains Regional Climate Center were used for the territory of South Dakota. The perspective for creating an Atlas of Climate and Water Resources for SD directed this study of total monthly snowfall with connection to landscape diversity.The initial matrix {Xn*p} where n is number of stations and p is number of variables of monthly average for period of observations. The maximum number of stations n with mutual interval of observations for SD is equal 93 (n=93). These stations have mutual time interval of 18 years observations (1952-53 – 1969-70). Total monthly snowfall has data for p=10 is number of months with observation and p=11 is number of months with observations with total snowfall for the winter season. The second matrix contains proportions of total monthly snowfall to total annual (proportion is the monthly total snowfall divided on total seasonal snowfall); the number of rows n and the number of variable are the same as in the first case: n=93 and p=10, 11.The average annual sum of total monthly snowfall (September-June) for SD 34.64 in obtained on 93 stations for 1952-1970, the median is 31.84 from the same data; ranged from 10.21 to 152.27 from the same data. The most variable month is April with average4.89 [in], median 3.67, min 0.75 and max 33.18; the average proportion for April is 0.13, min 0.03 and max 0.24. The averages for November to April grow as sequence: 3.60, 5.70, 4.88, 7.25, 7.46 and 4.89 the variability of those months has sequence: 2.49 2.52 2.25 3.33 3.53 and 4.87. The Pearson coefficient of correlation for the monthly snowfall averages from September to June with annual sum has sequence: 0.83, 0.92, 0.96, 0.92, 0.93, 0.93, 0.94, 0.92, 0.90, and 0.82; the correlation for the monthly proportions from September to June with annual sum has sequence: 0.38, 0.45, -0.01, -0.32, -0.33, -0.21, -0.15, 0.27, 0.36, and 0.35.The FA of both initial matrixes had extracted two factors for monthly observations with incorporation of 93 % of total variability of the data in the model; and three factors model with 70% variability of data for monthly proportions. The two factor groups in model of observed snowfall contain winter months (Dec – Mar) and all other. The model of monthly proportions with tree factor groups include winter months in two group and combination of spring and fall months in third. The factor scores presented as maps for SD allow trace distribution of factor groups of monthly snowfall patterns for two models through the territory of SD. The combination of factor scores distribution for two models is in a good agreement with main landscape regions and subregions of SD.

Multidimensional Analysis of Snow Cover Data in South Dakota Diversity of Landscapes

AbstractSnow distribution in SD was studied with Factor Analysis (FA) of monthly total snowfall [in]. The long-term data obtained from the High Plains Regional Climate Center were used for the territory of South Dakota. The perspective for creating an Atlas of Climate and Water Resources for SD directed this study of total monthly snowfall with connection to landscape diversity.The initial matrix {Xn*p} where n is number of stations and p is number of variables of monthly average for period of observations. The maximum number of stations n with mutual interval of observations for SD is equal 93 (n=93). These stations have mutual time interval of 18 years observations (1952-53 – 1969-70). Total monthly snowfall has data for p=10 is number of months with observation and p=11 is number of months with observations with total snowfall for the winter season. The second matrix contains proportions of total monthly snowfall to total annual (proportion is the monthly total snowfall divided on total seasonal snowfall); the number of rows n and the number of variable are the same as in the first case: n=93 and p=10, 11.The average annual sum of total monthly snowfall (September-June) for SD 34.64 in obtained on 93 stations for 1952-1970, the median is 31.84 from the same data; ranged from 10.21 to 152.27 from the same data. The most variable month is April with average4.89 [in], median 3.67, min 0.75 and max 33.18; the average proportion for April is 0.13, min 0.03 and max 0.24. The averages for November to April grow as sequence: 3.60, 5.70, 4.88, 7.25, 7.46 and 4.89 the variability of those months has sequence: 2.49 2.52 2.25 3.33 3.53 and 4.87. The Pearson coefficient of correlation for the monthly snowfall averages from September to June with annual sum has sequence: 0.83, 0.92, 0.96, 0.92, 0.93, 0.93, 0.94, 0.92, 0.90, and 0.82; the correlation for the monthly proportions from September to June with annual sum has sequence: 0.38, 0.45, -0.01, -0.32, -0.33, -0.21, -0.15, 0.27, 0.36, and 0.35.The FA of both initial matrixes had extracted two factors for monthly observations with incorporation of 93 % of total variability of the data in the model; and three factors model with 70% variability of data for monthly proportions. The two factor groups in model of observed snowfall contain winter months (Dec – Mar) and all other. The model of monthly proportions with tree factor groups include winter months in two group and combination of spring and fall months in third. The factor scores presented as maps for SD allow trace distribution of factor groups of monthly snowfall patterns for two models through the territory of SD. The combination of factor scores distribution for two models is in a good agreement with main landscape regions and subregions of SD.

Snow Cover in South Dakota: Statistical analysis of spatiotemporal diversity

AbstractSnow distribution and accumulation influence many human activities and the dynamic sustainability of ecological systems. Snow cover distribution analysis is a second research step towards creating a Atlas of Climate and Water Resources for SD (temperature analysis was presented last year). We analyzed the regional diversity of monthly total snowfall based on long-term data obtained from the High Plains Regional Climate Center for South Dakota. Multidimensional statistical methods were used, and the results presented for the State of South Dakota.The sets of initial matrixes were compiled with snow observations for the state. The first type of initial matrix of time series {Xt*n}, where t = number of years (67) and n = number of meteorological stations (25), contains 25 stations with mutual observational interval (1931-1998) of 67 winters. The second type of initial matrix {Xt*m}, where t = number of years (67) and m = number of months in a year, included seven initial matrixes.Statistical analysis allowed us to differentiate weather stations by temporal trends and spatial distribution for the time intervals 1931-1998. The average annual sum of total monthly snowfall (October-May) ranged from 25.5 to 53.5 inches for 25 stations trough this time interval. The most variable stations (Bowman Court House, Alexandria, Colony, Gordon, Clark, hot Springs and Eureka) were determined; their seasonality was described (the most variable months and correlation among months during period of snowfall) and their seasonal regime determined.

Kawasaki Disease Hospitalizations in a Predominantly African-American Population

This is a descriptive study of the occurrence of Kawasaki disease (KD) in an urban population that was a majority of African Americans. Records of 189 children (mean age, 2.9 +/- 2.2 years [range: 2 months to 11.1 years]) hospitalized for KD over 8 years (January 1, 1992 to December 31, 1999) were reviewed and data analyzed. One hundred thirty-six (72%) were African American (AA), 43 (23%) were white, and 9 (5%) children were "others.'' The annual frequency was 15 for AA and 7.7 for white per 100,000 5-year-old children. Coronary artery abnormalities (CAA) were reported in 21 (11%) children (18 [13.2%] of 136 AA, and 3 [4.7%] of 43 whites [p=0.095]). AA children with CAA were older than their white counterparts (26 to 24 vs. 5 to 2.8 months, p=0.03). There was a higher occurrence in winter and spring (110 cases [58%] vs. 79 cases [42%]) compared to summer and fall. KD occurrence was positively associated with average monthly snowfall (r=0.35, p=0.004) and inversely associated with average monthly temperature (r = - 0.2, p=0.048). African-American children were more likely to be hospitalized for KD compared to white children. The association of KD with temperature and precipitation suggest that it is influenced by environmental factors.

A 750-yr record of autumn snowfall and temperature variability and winter storminess recorded in the varved sediments of Bear Lake, Devon Island, Arctic Canada

The varve record from High Arctic, proglacial Bear Lake reveals a regionally coherent hydroclimatic signal as well as complexities due to changing hydroclimatic and limnologic conditions. Varve formation is strongly dependent on underflows that exhibit variability in strength during the past 750 yr. Periods with reduced underflow sedimentation and accumulation rates fail to produce varves in the distal part of the lake. Isolated coarse silt and sand grains occur in 80% of the varves and are interpreted to be niveo-aeolian in origin. Coarse (>500 μm) sand grains deposited on the lake ice by strong winter winds are notably less common since A.D. 1850, likely due to reduced storminess. Regression of the varve thickness record with meteorological records indicates high correlations with autumn (September and October) temperatures and total monthly snowfall. These correlations are best at times when underflow activity is sufficiently strong to produce varves throughout the lake. The close association with warmer temperatures and snow-bearing synoptic systems moving north in Baffin Bay suggests that the primary climate signal in the varves is varying autumn snow pack that controls nival discharge in the following year. The similarity between the other records of melt season temperature and sea-ice cover and the Bear Lake record suggests that summer and autumn conditions were generally similar across the Baffin Bay region through much of the last millennium.

Marginal Impacts of Design, Traffic, Weather, and Related Interactions on Roadside Crashes

A multivariate model that incorporates the effects of design, traffic, weather, and related interactions with design variables on reported roadside crashes is presented. By providing for a framework that accounts for all measurable effects, the model minimizes the impact of omitted variable effects. Furthermore, the presented framework accounts for partial observability effects that stem from fluctuations in environmental conditions as well as unobserved effects that contribute to heterogeneity in the traffic safety network. A sample of 318 sections 1 mi long was used for the study. These sections represent the state highway network in the state of Washington on the basis of environmental and road classification factors and therefore were used for the collection of detailed precipitation, snowfall, and temperature data in addition to roadway and roadside design and traffic parameters. The resulting model suggests that the marginal impact of weather is both in main effects and interactive form, and that even after controlling for unobserved heterogeneity and partial observability, weather effects play a statistically significant role in roadside crash occurrence. In particular, it was found that in addition to precipitation, average monthly snowfall exceeding 4 in. and interactions between snow depths and horizontal curves were found to have a statistically significant effect on roadside crash frequency probabilities. The marginal effects of these variables were also statistically significant; furthermore, the contribution of weather and related interactions to the likelihood of roadside crash frequencies was approximately 19%, design main effects contributed to 33%, and traffic and design interactions contributed to 6%. Weather interactions with design contributed to approximately 6% of the overall likelihood. Traffic as a main effect contributed 36% to the overall roadside crash likelihood.

A climatology of snowfall‐temperature relationships in Canada

A better understanding of potential climate change impacts on the global hydrological cycle requires knowledge of the interaction between air temperature and water in its various forms. One important example is the effect of air temperature on snowfall. Proper parameterization of the snowfall‐temperature relationship in climate models is essential for accurate prediction of future snowfall changes that might arise from high‐latitude warming. On a climatological basis, at any location, air temperature and snowfall can be correlated positively (higher temperatures increase atmospheric moisture and snowfall through the Clausius‐Clapeyron relationship) or negatively (precipitation falls as rain instead of snow). Examination of 50 years of monthly snowfall water equivalent and mean temperature data indicates that the snowfall‐temperature relationship is positive in the high latitudes and negative in southern Canada, along both coasts, and east of the Rockies. The “zero line” (the transition zone north of which warmer months receive more snowfall than colder months) migrates southward from autumn to winter so that by January most of eastern and northwestern Canada has a positive snowfall‐temperature slope. The primary exception to a straightforward relationship between slope and latitude occurs east of the Rockies, where anomalous negative slopes extend far to the north. In this region, dry, adiabatically warmed air from downslope or circulation effects may reduce the number of snow events and modify the slope of the snowfall‐temperature curve. Based on first principles and Canadian observations, we develop a function relating temperature to snowfall that attempts to account for the complex spatial and seasonal nature of the snowfall‐temperature relationship. Given the importance of snowfall in the global radiation balance, this analysis could be used to improve climate change predictions.

Precipitation Amount Separated by Precipitation Type Using Discriminating Temperature During Winter Season

The discriminating temperature, which is applied to determine the precipitation type (snowfall or rainfall) corresponding to 50% occurrence probability, is obtained from climatic data (SDP). The monthly discriminating temperatures were determined for 15 climatic regions of Japan using station daily data. The values range from 1.1°C to 3.0°C. By the values and digital maps of daily temperature, the precipitaion types were judged daily for whole grid map of daily precipitaion. Finally were obtained the digital maps of monthly snowfall and rainfall. As a result, precipitation is observed mostly as the type of rainfall along the coast of Japan Sea more than generally expected. While in Hokkaido and in the mountain area of northern part of Japan, precipitation is observed mostly as snowfall in winter.