Estimating the short-term and long-term wind speeds: implementing hybrid models through coupling machine learning and linear time series models

Abstract. The Station for Measuring Ecosystem Atmosphere Relations (SMEAR) II is a unique station in the world due to the wide range of long-term measurements tracking the Earth-atmosphere interface. In this study, we characterize the composition of organic aerosol (OA) at SMEAR II by quantifying its driving constituents. We utilize a multi-year data set of OA mass spectra measured in situ with an Aerosol Chemical Speciation Monitor (ACSM) at the station. To our knowledge, this mass spectral time series is the longest of its kind published to date, and its detailed analysis required development of a new methodology. To this purpose, we developed an efficient and robust data analysis framework utilizing machine learning tools. These included unsupervised feature extraction and classification stages to manage and process the large amounts of data. The extensive chemometric analysis was conducted with a combination of Positive Matrix Factorization (PMF), rolling window analysis, bootstrapping, K-Means clustering, data weighting and diagnostics based algorithmic choice-making, among others. This combination of statistical tools provided a data driven analysis methodology to achieve robust solutions with minimal subjectivity. Following the extensive statistical analyses, we were able to divide the 2012–2019 SMEAR II OA data (mass concentration interquartile range (IQR): 0.7, 1.3, 2.6 µg m−3) to three sub-categories: low-volatility oxygenated OA (LV-OOA), semi-volatile oxygenated OA (SV-OOA), and primary OA (POA). LV-OOA was the most dominant OA type (organic mass fraction IQR: 49, 62, and 73 %). The seasonal cycle of LV-OOA was bimodal, with peaks both in summer and in February. We associated the wintertime LV-OOA with anthropogenic sources and assumed biogenic influence in LV-OOA formation in summer. Through a brief trajectory analysis, we estimated summertime natural LV-OOA formation of tens of ng m−3 h−1 over the boreal forest. SV-OOA was the second highest contributor to OA mass (organic mass fraction IQR: 19, 31, and 43 %). Due to SV-OOA’s clear peak in summer, we estimate biogenic processes as the main drivers in its formation. Unlike for LV-OOA, the highest SV-OOA concentrations were detected in stable summertime nocturnal surface layers. However, also the nearby sawmills likely played a significant role in SV-OOA production as also exemplified by previous studies at SMEAR II. POA, taken as a mix of two different OA types reported previously, hydrocarbon-like OA (HOA) and biomass burning OA (BBOA), made up a minimal OA mass fraction (IQR: 2, 6, and 13 %). Both POA organic mass fraction and mass concentration peaked in winter. Its appearance at SMEAR II was linked to strong southerly winds. The high wind speeds probably enabled the POA transport to SMEAR II from faraway sources in a relatively fresh state. In case of slower wind speeds, POA likely evaporated or aged into oxidized organic aerosol before detection. The POA organic mass fraction was significantly lower than reported by aerosol mass spectrometer (AMS) measurements two to four years prior to the ACSM measurements. While the co-located long-term measurements of black carbon supported the hypothesis of higher POA loadings prior to year 2012, it is also possible that ACSM was less efficiently capturing short term (POA) pollution plumes. Despite the length of the ACSM data set, we did not focus on quantifying long-term trends of POA (nor other components) due to the high sensitivity of OA composition to meteorological anomalies, the occurrence of which is likely not normally distributed over the eight year measurement period. We hope that our successfully applied methodology encourages also other researchers possessing several-year-long time series of similar data to tackle the data analysis via similar semi- or unsupervised machine learning approaches. This way aerosol chemometric analysis procedures would be further developed into yet more streamlined and autonomous directions.

Download Full-text

Stochastic Investment Modelling: The Case of South Africa

British Actuarial Journal ◽

10.1017/s135732170000355x ◽

1996 ◽

Vol 2 (3) ◽

pp. 765-801 ◽

Cited By ~ 15

Author(s):

R.J. Thomson

Keyword(s):

South Africa ◽

Time Series ◽

Interest Rates ◽

South African ◽

Growth Rates ◽

Time Series Models ◽

Short Term ◽

Model Based ◽

Inflation Rates

ABSTRACTThe purpose of this paper is to describe a methodology for determining an appropriate structure for time-series models of inflation rates, short-term and long-term interest rates, dividend growth rates, dividend yields, rental growth rates and rental yields and to demonstrate the application of that methodology to the development of a model based on South African data. It is suggested that the methodology used in this paper may be applied to other economic environments.

Download Full-text

Hybrid Time Series Models for Forecasting Maize Production in India

Current Journal of Applied Science and Technology ◽

10.9734/cjast/2021/v40i2331492 ◽

2021 ◽

pp. 49-57

Author(s):

Pramit Pandit ◽

Bishvajit Bakshi ◽

Varun Gangadhar

Keyword(s):

Time Series ◽

Model Building ◽

Linear Models ◽

Linear Time ◽

Arima Model ◽

Hybrid Models ◽

Time Series Models ◽

Real World Data ◽

Maize Production ◽

Non Linear

In spite of the immense success of different linear and non-linear time series models in their respective domains, real-world data are rarely pure linear or non-linear in nature. Hence, a hybrid modelling framework with the capability of handling both linear and non-linear patterns can substantially improve the forecasting accuracy. With this backdrop, an effort has been made in this investigation to evaluate the suitability of hybrid models in compassion to single linear or non-linear models for forecasting maize production in India. Data from 1949-50 to 2016-17 have been utilised for the model building purpose while retaining the data from 2017-18 to 2019-20 for the post-sample accuracy assessment. Outcomes emanated from this investigation clearly reveals that the ARIMA-NLSVR model has outperformed all other candidate models employed in this study. It is noteworthy to mention that both the hybrid models have performed better than their individual counterparts. The superior forecasting ability of both the non-linear models over the linear ARIMA model has also been evident.

Download Full-text

Long-Term Data Traffic Forecasting for Network Dimensioning in LTE with Short Time Series

Electronics ◽

10.3390/electronics10101151 ◽

2021 ◽

Vol 10 (10) ◽

pp. 1151

Author(s):

Carolina Gijón ◽

Matías Toril ◽

Salvador Luna-Ramírez ◽

María Luisa Marí-Altozano ◽

José María Ruiz-Avilés

Keyword(s):

Time Series ◽

Deep Learning ◽

Time Series Analysis ◽

Data Traffic ◽

Medium Term ◽

Short Term ◽

Traffic Forecasting ◽

Series Analysis ◽

Radio Planning

Network dimensioning is a critical task in current mobile networks, as any failure in this process leads to degraded user experience or unnecessary upgrades of network resources. For this purpose, radio planning tools often predict monthly busy-hour data traffic to detect capacity bottlenecks in advance. Supervised Learning (SL) arises as a promising solution to improve predictions obtained with legacy approaches. Previous works have shown that deep learning outperforms classical time series analysis when predicting data traffic in cellular networks in the short term (seconds/minutes) and medium term (hours/days) from long historical data series. However, long-term forecasting (several months horizon) performed in radio planning tools relies on short and noisy time series, thus requiring a separate analysis. In this work, we present the first study comparing SL and time series analysis approaches to predict monthly busy-hour data traffic on a cell basis in a live LTE network. To this end, an extensive dataset is collected, comprising data traffic per cell for a whole country during 30 months. The considered methods include Random Forest, different Neural Networks, Support Vector Regression, Seasonal Auto Regressive Integrated Moving Average and Additive Holt–Winters. Results show that SL models outperform time series approaches, while reducing data storage capacity requirements. More importantly, unlike in short-term and medium-term traffic forecasting, non-deep SL approaches are competitive with deep learning while being more computationally efficient.

Download Full-text

An Efficient Prediction Model for Water Discharge in Schoharie Creek, NY

Journal of Climatology ◽

10.1155/2014/284137 ◽

2014 ◽

Vol 2014 ◽

pp. 1-10 ◽

Cited By ~ 6

Author(s):

Katerina G. Tsakiri ◽

Antonios E. Marsellos ◽

Igor G. Zurbenko

Keyword(s):

New York ◽

Time Series ◽

Water Discharge ◽

Winter Period ◽

Statistical Methodology ◽

Summer Period ◽

Short Term ◽

Efficient Prediction ◽

Climatic Time Series

Flooding normally occurs during periods of excessive precipitation or thawing in the winter period (ice jam). Flooding is typically accompanied by an increase in river discharge. This paper presents a statistical model for the prediction and explanation of the water discharge time series using an example from the Schoharie Creek, New York (one of the principal tributaries of the Mohawk River). It is developed with a view to wider application in similar water basins. In this study a statistical methodology for the decomposition of the time series is used. The Kolmogorov-Zurbenko filter is used for the decomposition of the hydrological and climatic time series into the seasonal and the long and the short term component. We analyze the time series of the water discharge by using a summer and a winter model. The explanation of the water discharge has been improved up to 81%. The results show that as water discharge increases in the long term then the water table replenishes, and in the seasonal term it depletes. In the short term, the groundwater drops during the winter period, and it rises during the summer period. This methodology can be applied for the prediction of the water discharge at multiple sites.

Download Full-text