Land-based Weather Stations Research Articles

Extreme Gradient Boosting Model for Rain Retrieval using Radar Reflectivity from Various Elevation Angles

The purpose of this study was to develop an optimal estimation model for rainfall rate retrievals using radar reflectivity, thereby gaining an effective grasp of rainfall information for disaster prevention uses. A process was designed for evaluating the optimal retrieval models using various dataset combinations with radar reflectivity and ground meteorological attributes. Various ground meteorological attributes (such as relative humidity, wind speed, precipitation, etc.) were obtained using the land-based weather stations affiliated with Taiwan’s Central Weather Bureau (CWB). This study used nine radar reflectivity provided by the Hualien weather surveillance radar station’s Volume Cover Pattern 21 system. The developed models are built using multiple machine learning algorithms, including linear regression (REG), support vector regression (SVR), and extreme gradient boosting (XGBoost), in addition to the Marshall–Palmer formula (MP). The study examined 14 typhoons that occurred from 2008 to 2017 at Chenggong station in southeast Taiwan, and Lanyu station in the outlying islands, and the top four major rainfall events were designated as test typhoons—Nanmadol (2011), Tembin (2012), Matmo (2014), and Nepartak (2016). The results indicated that for rainfall retrievals, radar reflectivity at a scanning (elevation) angle of 6.0° combined with ground meteorological attributes were the optimal input variables for the Chenggong station, whereas radar reflectivity at an elevation angle of 4.3° combined with ground meteorological attributes were optimal for the Lanyu station. In terms of model performance, XGBoost models had the lowest error index at Chenggong and Lanyu stations compared with MP, REG, and SVR models. XGBoost models at Lanyu station had the highest efficiency coefficient (0.903), and those at Chenggong station had the second highest (0.885). As a result, pairing the combination of optimal radar reflectivity and ground meteorological attributes, as verified by the evaluation process, with a high-efficiency algorithm (XGBoost) can effectively increase the accuracy of rainfall retrieval during typhoons.

Uncertainty of hydrologic processes caused by bias-corrected CMIP5 climate change projections with alternative historical data sources

Uncertainty in simulating hydrologic response to future climate is generally assumed to result from the combined uncertainties of the General Circulation Model (GCM), representative concentration pathway (RCP), downscaling method, and hydrologic model used. However, another source of uncertainty, the observed climate data source used to statistically downscale and bias-correct GCM projections, has largely been overlooked. This study assessed the shifts, variability, and uncertainty in streamflow simulation from three downscaling data sources (NCDC land-based weather stations, NEXRAD spatial grid, and PRISM spatial grid) relative to those introduced by six GCMs and three RCPs in west-central Kansas, U.S. Streamflow simulated by the Soil and Water Assessment Tool (SWAT) hydrologic model was found to be more sensitive to future precipitation than to maximum and minimum temperatures. The greatest uncertainty in simulated streamflow was associated with selection of the GCM. Uncertainty in simulated streamflow associated with the observed bias-correction data source (NCDC, PRISM, NEXRAD) was greater than with RCPs and was primarily related to uncertainty in precipitation. This study highlighted the importance of recognizing uncertainty from bias-correction data sources in representing future climate scenarios in hydrologic simulations.

Spatial analysis of the effective coverage of land-based weather stations for traffic crashes

This paper investigates the effective spatial coverage of nationwide land-based weather stations of the National Oceanic Atmospheric Administration (NOAA) for traffic crash analysis. The weather data were collected from the Quality Controlled Local Climatological Data (QCLCD) and the fatal crashes were obtained from the Fatality Analysis Reporting Systems (FARS) during the year of 2007–2014. Both QCLCD and FARS contain geographic coordinates for locations and weather condition information as a categorical variable. The spatial coverage of weather stations for the analysis was made by geoprocessing, which uses multiple buffers (i.e. radii 5, 10, 15, and 20 miles), and then was evaluated via Cohen's κ statistics, which is used to determine an agreement of weather between QCLCD and FARS within the buffer. The applicability of the weather station's data by nine climate regions was assessed by developing a series of negative binomial models. According to the estimated Cohen's κ statistics, the rain and snow weather conditions have a moderate agreement up to 20 miles. However, in the case of fog weather condition, it has a slight agreement. The statistical modeling results showed that weather stations data can be a good exposure measure for weather-related fatal crashes along with the vehicle-miles-traveled. Considering one geographical feature that approximately more than 75% of all fatal crashes are located within 20-miles radius of the weather stations in the USA, it is evident that the data from the existing weather stations can be cost-effective to develop geospatial crash risk analysis model.

Influenza Herd-Level Prevalence and Seasonality in Breed-to-Wean Pig Farms in the Midwestern United States.

Influenza is a costly disease for pig producers and understanding its epidemiology is critical to control it. In this study, we aimed to estimate the herd-level prevalence and seasonality of influenza in breed-to-wean pig farms, evaluate the correlation between influenza herd-level prevalence and meteorological conditions, and characterize influenza genetic diversity over time. A cohort of 34 breed-to-wean farms with monthly influenza status obtained over a 5-year period in piglets prior to wean was selected. A farm was considered positive in a given month if at least one oral fluid tested influenza positive by reverse transcriptase polymerase chain reaction. Influenza seasonality was assessed combining autoregressive integrated moving average (ARIMA) models with trigonometric functions as covariates. Meteorological conditions were gathered from local land-based weather stations, monthly aggregated and correlated with influenza herd-level prevalence. Influenza herd-level prevalence had a median of 28% with a range from 7 to 57% and followed a cyclical pattern with levels increasing during fall, peaking in both early winter (December) and late spring (May), and decreasing in summer. Influenza herd-level prevalence was correlated with mean outdoor air absolute humidity (AH) and temperature. Influenza genetic diversity was substantial over time with influenza isolates belonging to 10 distinct clades from which H1 delta 1 and H1 gamma 1 were the most common. Twenty-one percent of farms had three different clades co-circulating over time, 18% of farms had two clades, and 41% of farms had one clade. In summary, our study showed that influenza had a cyclical pattern explained in part by air AH and temperature changes over time, and highlighted the importance of active surveillance to identify high-risk periods when strategic control measures for influenza could be implemented.

Emergence of white pine needle damage in the northeastern United States is associated with changes in pathogen pressure in response to climate change.

The defoliation of the eastern white pine (Pinus strobus) across the northeastern United States is an escalating concern threatening the ecological health of northern forests and economic vitality of the region's lumber industry. First documented in the spring of 2010 affecting 24328 hectares in the state of Maine, white pine needle damage (WPND) has continued to spread and is now well established in all New England states. While causal agents of WPND are known, current research is lacking in both sampling distribution and the specific environmental factor(s) that affect the development and spread of this disease complex. This study aims to construct a more detailed distribution map of the four primary causal agents within the region, as well as utilize long-term WPND monitoring plots and data collected from land-based weather stations to develop a climatic model to predict the severity of defoliation events in the proceeding year. Sampling results showed a greater distribution of WPND than previously reported. WPND was generally found in forest stands that compromised >50% eastern white pine by basal area. No single species, nor a specific combination of species had a dominating presence in particular states or regions, thus supporting the disease complex theory that WPND is neither caused by an individual species nor by a specific combination of species. In addition, regional weather data confirmed the trend of increasing temperature and precipitation observed in this region with the previous year's May, June, and July rainfall being the best predictor of defoliation events in the following year. Climatic models were developed to aid land managers in predicting disease severity and accordingly adjust their management decisions. Our results clearly demonstrate the role changing climate patterns have on the health of eastern white pine in the northeastern United States.

A comparison of SNOTEL and AMSR-E snow water equivalent data sets in western US watersheds

It is a consensus among earth scientists that climate change will result in an increased frequency of extreme events (e.g. floods, droughts). Streamflow forecasts and flood/drought analyses, given this high variability in the climatic driver (snowpack), are vital in the western USA. However, the ability to produce accurate forecasts and analyses is dependent upon the quality of these predictors. Run-off and stream volume analysis in the region is currently based upon in situ telemetry snow data products. Recent satellite deployments offer an alternative data source of regional snowpack. The proposed research investigates and compares remotely sensed snow water equivalent (SWE) data sets in western US watersheds in which snowpack is the primary driver of streamflow. Watersheds investigated include the North Platte, Upper Green and Upper Colorado. SWE data sets incorporated are in situ snowpack telemetry (SNOTEL) sites and the advanced microwave scanning radiometer-earth observing system (AMSR-E) aboard NASA's Aqua satellite. The time period analysed is 2003-2008, coincident with the deployment of the NASA Aqua satellite. Bivariate techniques between data sets are performed to provide valuable information on the time series of the snow products. Multivariate techniques including principal component analysis (PCA) and singular value decomposition (SVD) are also applied to determine similarities and differences between the data sets and investigate regional snowpack behaviours. Given the challenges (including costs, operation and maintenance) of deploying SNOTEL stations, the objective of the research is to determine whether remotely sensed SWE data provide a comparable option to in situ data sets. Correlation analysis resulted in only 11 of the 84 SNOTEL sites investigated being significant at 90% or greater with a corresponding AMSR-E cell. Agreement between SWE products was found to increase in lower elevation areas and later in the snowpack season. Two distinct snow regions were found to behave similarly between both data sets using a rotated PCA approach. Additionally, SVD linked both data products with streamflow in the region and found similar behaviour among data sets. However, when comparing SNOTEL data with the corresponding satellite cell, there was a consistent bias in the absolute magnitude (SWE) of the data sets. The streamflow forecasting results conclude regions that have few (or zero) land-based weather stations can incorporate the AMSR-E SWE product into a streamflow forecast model and obtain accurate values.