Fig 2 - uploaded by Doerte Jakob
Content may be subject to copyright.
Source publication
One of the main drivers for this project is the requirement to complement other high-quality surface data-sets with surface wind data for use in climate change detection and attribution studies. The high-quality data may also be used to analyse trends in storminess. Investigations highlighted the following three issues: Over the last two decades Au...
Citations
... where δFFDI is a change in the FFDI and δV is a change in mean windspeed Lucas (2009). Most stations converted from visual estimates of wind force or pressure anemometers [20,21] to instrumental measurements of windspeed using cup anemometers around 1993 [17,21]. A total of 9 station records were available for fire years from July to June 1972-73 to 2009-10. ...
... where δFFDI is a change in the FFDI and δV is a change in mean windspeed Lucas (2009). Most stations converted from visual estimates of wind force or pressure anemometers [20,21] to instrumental measurements of windspeed using cup anemometers around 1993 [17,21]. A total of 9 station records were available for fire years from July to June 1972-73 to 2009-10. ...
... Shifts were also detected in most records of RH and V. Despite adjustments being made to these records over time, some contain inhomogeneities that can be identified due to changes in instrumentation (windspeed) and anomalous changes in particular stations (RH and V). Jakob [21] described the difficulties in extracting a reliable record from Australian windspeed observations. Troccoli et al. [50] analyzed measurements at 2 m and 10 m heights, subjecting records to rigorous quality control. ...
This paper introduces and analyzes fire climate regimes, steady-state conditions that govern the behavior of fire weather. A simple model representing fire climate was constructed by regressing high-quality regional climate averages against the station-averaged annual Forest Fire Danger Index (FFDI) for Victoria, Australia. Four FFD indices for fire years 1957–2021 were produced for 10 regions. Regions with even coverage of station-averaged total annual FFDI (ΣFFDI) from 1971–2016 exceeded Nash–Sutcliffe efficiencies of 0.84, validating its widespread application. Data were analyzed for shifts in mean, revealing regime shifts that occurred between 1996 and 2003 in the southern states and 2012–13 in Queensland. ΣFFDI shifted up by ~25% in SE Australia to 8% in the west; by approximately one-third in the SE to 7% in the west for days above high fire danger; by approximately half in the SE to 11% in the west for days above very high, with a greater increase in Tasmania; and by approximately three-quarters in the SE to 9% in the west for days above severe FFDI. Attribution of the causes identified regime shifts in the fire season maximum temperature and a 3 p.m. relative humidity, with changing drought factor and rainfall patterns shaping the results. The 1:10 fire season between Regimes 1 and 2 saw a three to seven times increase with an average of five. For the 1:20 fire season, there was an increase of 2 to 14 times with an average of 8. Similar timing between shifts in the Australian FFDI and the global fire season length suggests that these changes may be global in extent. A trend analysis will substantially underestimate these changes in risk.
... Aggregate rainfall over the previous 3 and 10 days was calculated as the cumulative total rainfall from BOM daily rainfall data. Although wind has also been investigated as a factor in call probability (Oseen & Wassersug, 2002;Penman et al., 2006), it was excluded because the available data are not accurate enough to associate with daily call activity (Jakob, 2010) as a grid cell scale. We assigned a numerical day of year to each day as an explanatory factor representing day length and harmonic regression (Chatfield & Xing, 2019;Weir et al., 2005) and distinguishing spring and autumn days from one another, unlike photoperiod (see Figure S2 for a plot displaying the correlation between day of year and photoperiod). ...
Aim
Here we investigate the strength of the relationships between meteorological factors and calling behaviour of 100 Australian frog species using continent‐wide citizen science data. First, we use this dataset to quantify the meteorological factors that best predict frog calling. Second, we investigate the strength of interactions among predictor variables. Third, we assess whether frog species cluster into distinct groups based on shared drivers of calling.
Location
Australia.
Method
To assess the relationship between calling and meteorological traits, we used spatio‐temporal subsampling (daily data fitted to 10 km² grid cells) of call and meteorological data as inputs to a boosted regression tree. We scaled the model outputs, which created a descriptive ranking of predictor importance. For strongly day‐driven species, we conducted further analyses to examine the influences of meteorological factors within the breeding season.
Results
We found a strong seasonal signal, with day of year the strongest relationship to calling in 67 out of our 100 species, moderate relationships between temperature and calling, and weak relationships between rainfall and calling. Despite the common narratives, we found that frogs did not group into distinct categories based upon the influence of meteorological factors. For strongly day‐driven species, we found similar patterns within the breeding season.
Main conclusions
We demonstrate the importance of day of year and temperature thresholds in predicting frog calling behaviour in Australia. Understanding how meteorological conditions influence phenological events, such as breeding, will be increasingly important considering the rapid changes in environmental conditions and stability throughout most of the world, and how important breeding is to species survival.
... In many applications continuous daily time step records are required (which are not widely available) and spatial modelling may require high-resolution gridded surfaces which are rarely available on a continental scale. Jakob (2010) discussed the challenges in developing a high-quality surface wind-speed dataset for Australia, noting issues such as changes in instrumentation (e.g., installation of Automatic Weather Stations (AWS)) and the impact of variable time zones (daylight saving). Nevertheless, an up-to-date high-quality wind dataset is required for a range of research and operational activities. ...
Gridded near-surface (2 and 10 m) daily average wind datasets for Australia have been constructed by interpolating observational data collected by the Australian Bureau of Meteorology (BoM). The new datasets span Australia at 0.05 × 0.05° resolution with a daily time step. They are available for the period 1 January 1975 to present with daily updates. The datasets were constructed by blending observational data collected at various heights using local surface roughness information. Error detection techniques were used to identify and remove suspect data. Statistical performances of the spatial interpolations were evaluated using a cross-validation procedure, by sequentially applying interpolations after removing the observed weather station data. The accuracy of the new blended 10 m wind datasets were estimated through comparison with the Reanalysis ERA5-Land 10 m wind datasets. Overall, the blended 10 m wind speed patterns are similar to the ERA5-Land 10 m wind. The new blended 10 m wind datasets outperformed ERA5-Land 10 m wind in terms of spatial correlations and mean absolute errors through validations with BoM 10 m wind weather station data for the period from 1981 to 2020. Average correlation (R2) for blended 10 m wind is 0.68, which is 0.45 for ERA5-Land 10 m wind. The average of the mean absolute error is 1.15 m/s for blended 10 m wind, which is lower than that for ERA5-Land 10 m wind (1.61 m/s). The blending technique substantially improves the spatial correlations for blended 2 m wind speed.
... Most stations converted to instrumental measurements of wind speed using cup anemometers at around 1993 (Jakob, 2010;Lucas, 2009), either from visual estimates of wind force or pressure anemometers (Jakob, 2010;Miller et al., 2013). The preceding observer readings have a larger variance, more frequent calm days and can differ markedly between personnel (Lucas, 2009). ...
... Most stations converted to instrumental measurements of wind speed using cup anemometers at around 1993 (Jakob, 2010;Lucas, 2009), either from visual estimates of wind force or pressure anemometers (Jakob, 2010;Miller et al., 2013). The preceding observer readings have a larger variance, more frequent calm days and can differ markedly between personnel (Lucas, 2009). ...
This report constructs fire climates for Australia from high quality climate data, identifying specific regimes that change abruptly in response to warmer and drier conditions. Fire risk increased abruptly in the late 1990s and early 2000s leading to hotter and more frequent fires. These acute changes in risk are not predicted by the current generation of climate models. This is a global phenomenon related to large-scale shifts in the climate system.
... Because of observational constraints, historical trends in the frequency and intensity of convective winds in Australia are unknown (Walsh et al. 2016;Brown & Dowdy 2019). This is largely due to spatio-temporal inhomogeneities in severe weather reports (Allen et al. 2011) and wind observations (Jakob 2010). It is also noted that convective phenomena occur on small spatial scales which are often missed by the observational network and make the detection of trends difficult. ...
The influence of anthropogenic climate change on extreme temperatures, winds and bushfire weather in Australia is assessed here using a standardised method for projections information. These assessments consider a comprehensive range of factors based on observations, modelling and physical process understanding. Those factors are reviewed using a standardised method to collate lines of evidence and then guide the production of projections data and confidence assessments. Projections are produced based on global climate model data as well as dynamical downscaling data using three regional climate modelling approaches (CCAM, BARPA and NARCliM/WRF), with environmental diagnostics also used for severe convective winds from thunderstorms. The projections data are all calibrated using quantile matching methods trained on observations-based data, with a particular focus on the accurate representation of extremes. The resultant projections data include nationally consistent maps corresponding to the 10-year average recurrence interval (i.e., return period) around the middle of this century, with a focus of the discussion on regions around southern and eastern Australia during summer as needed for some risk assessment applications. The projections data are also available for other seasons and time periods throughout this century, as well as for other metrics of extreme or average conditions. The results for southern and eastern Australia during summer show more extreme temperatures (very high confidence), more severe winds (low confidence) and more dangerous bushfire conditions (high confidence in southern Australia; medium confidence in eastern Australia) attributable to increasing greenhouse gas emissions.
... Una interesante discusión sobre los problemas encontrados en bases de registro de velocidades de viento y formas más complejas de corregirlos puede encontrarse en (Jacob, 2010). ...
The researches related to the return period of extreme winds in Cuba contain a limited number of meteorological stations and consider the annual maximum gusts of wind. In this sense, evaluate the danger caused by maximum winds in Cuba is the goal of this article. For that, maximum daily wind values of the station records were use, which is quantified using the return period, using the technique of "Peaks on the Reference" to set the curve of extreme values through the Generalized Pareto Distribution. The article also presents an efficient computational technique for the separation of wind types and the corresponding calculation of the PR curve for the synoptics and storm with the Reliability Interval for a 95% probability. This is important for the authorities that must prepare risk management plans and the incorporation of the required measures, as well as aid for urban planning, trying to avoid the development of urban centers in high risk regions. In general, it was found that the danger caused by non-hurricane winds in Cuba is relatively low. It was also found that the station records are adequate to calculate the danger caused by wind, although in some station’s errors were found in the characterization of the present weather conditions, so as part of this work a simple technique was implemented to correct these errors.
... Una interesante discusión sobre los problemas encontrados en bases de registro de velocidades de viento y formas más complejas de corregirlos puede encontrarse en (Jacob, 2010). ...
Las investigaciones relacionadas con el período de retorno de vientos extremos en Cuba contienen un número limitado de estaciones meteorológicas y considera las rachas máximas anuales del
viento. En tal sentido, el objetivo de este artículo es evaluar el peligro producido por vientos máximos en Cuba utilizando valores máximos del viento diarios de los registros de estaciones, el cual es cuantificado usando el periodo de retorno, empleando la técnica de “Picos sobre la referencia” para fijar la curva de valores extremos a través de la Distribución Generalizada de Pareto. En el artículo también se presenta una técnica computacional eficiente para la separación de los tipos de viento y el correspondiente cálculo de la curva de PR para los sinópticos y de tormenta con el Intervalo de Confiabilidad para una
probabilidad del 95%. Esto es de suma importancia para las autoridades que deben elaborar planes de gestión de riesgos y la incorporación de las medidas requeridas, así como ayuda a la planeación urbana, tratando de evitar el desarrollo de centros urbanos en regiones de alto peligro. En general se encontró que el peligro producido por vientos no huracanados en Cuba es relativamente bajo. También se comprobó que los registros de las estaciones son adecuados para calcular el peligro producido por viento, aunque en
algunas estaciones se encontraron errores en la caracterización de las condiciones de tiempo presente, por lo que como parte de este trabajo se implementó una técnica sencilla que permite corregir estos errores.
... These results also support the reality of gradual sheltering of the site over time, although the there may also be some quality issues in the Eversleigh wind observations due to a change in observation techniques. This is not overly surprising, as wind data are notoriously sensitive to changes in observing methods and local features (Jakob, 2010). Overall, Table 2 ranks the quality of the measured meteorological data from excellent (A) to unsuitable for climatological analysis in their current form (C). All measured parameters are rated A or B except the wind data, attached thermometer and minimum temperature, which show a trend or step change that is at odds with the behaviour of other related variables. ...
Historical weather observations on the daily scale are vital for the improvement of reanalysis products and the analysis of long‐term variability of extreme events. While daily datasets extend for several centuries in parts of the Northern Hemisphere, the majority of historical data for the Southern Hemisphere are monthly averages or totals. In this paper, we describe a newly recovered dataset of ten daily meteorological variables for 1877 to 1922 from Eversleigh, a property in the New England region of New South Wales in Australia. Here, we present the full process of data rescue, from digitization to quality control and an assessment of homogeneity. We show that the majority of variables were recorded to a high standard and that the data are of general use for climate analysis. Forty years of daily temperature, cloud cover, wind, and pressure observations are now available for the New England Plateau, offering data that are more complete than any other records for the region in the Australian Bureau of Meteorology dataset. The Eversleigh dataset now provides an opportunity to gain more insight into the 19th century weather and climate of eastern Australia during a time of large interannual climate variability before the dominant impact of an anthropogenic warming signal.
... The first of these was due to the installation of Automatic Weather Stations (AWS) throughout the network in the 1990s, with an accompanying change from the Dines anemometer (which records wind gusts every 2-3 seconds without averaging) to a rotating cup anemometer (using a 3-second moving average to record wind gust speeds consistent with WMO standards). The resulting wind gusts recorded by the rotating cup anemometer within the AWS system are about 6-19% lower compared to the Dines equivalent gust (depending on wind characteristics and type of Dines anemometer), representing a notable inhomogeneity in the long-term gust records at BoM stations ( Jakob 2010, Miller et al. 2013), noting some calibrations to help compensate for such differences having been applied for one of the datasets used for this study (the JDH dataset, as described in Section 2.3 as well as Appendix Section 5.1). Secondly, there was a change in the type of AWS system used in some parts of the network occurred during 2010/11, discussed by Azorin- Molina et al. (2018) as causing an inhomogeneity point within wind gust data at 56 stations in the BoM network. ...
The spatial and temporal characteristics of extreme convective wind gusts in South Australia are analysed over the period 1979 to 2017, using station data in combination with wind gust and environmental parameter data from atmospheric reanalyses. The reanalyses used are the European Center for Medium-range Weather Forecasts ERA-Interim, and the Bureau of Meteorology High-resolution Atmospheric Regional Reanalysis for Australia (BARRA). The wind gust data obtained from the reanalyses generally provide a good representation of the observed wind gusts, apart from extreme events above 30 m.s-1 which are a focus of this study. The use of previously established environmental parameters associated with thunderstorm activity is found to be better than using wind gust data from the reanalyses for indicating the occurrence of observed extreme wind gusts. A conditional parameter is also tested here, based on two different sets of environmental conditions, to account for extreme wind events that can sometimes occur in low/zero large scale CAPE environments. Results include a systematic analysis of extreme wind environments based on all days from 1979 to 2017 and an examination of two extreme wind events: November 1979, as well as September 2016 which included the occurrence of several tornadoes. Regional variations in extreme wind environments are indicated, including being more strongly associated with CAPE for inland locations than near-coastal locations where wind shear plays a more important role. Long-term changes in extreme wind environments are examined over the 1979-2017 period, indicating a potential area of decreased risk in the far north of South Australia in recent decades, and a potential area of increased risk around the southern Flinders Ranges and Yorke Peninsula region. Extreme convective wind gust environments are found to have a strong relationship with the Southern Annular Mode, particularly during winter and spring in some regions, but not with the El Niño-Southern Oscillation or Indian Ocean Dipole.
... Due to the variety of factors affecting the measurement of wind speed, that is, its high natural short-term variance (Balling and Cerveny, 2005;Jakob, 2010), long-term trends (Vautard et al., 2010;McVicar et al., 2012), and the relatively high spatial variability (Azorin-Molina et al., 2014), as well as the high sensitivity of wind to local site conditions (WMO, 2017), implementing quality control and homogenization procedures on wind series has been challenging. To summarize, only a few approaches have been developed for mean wind speed so far in recent years: (a) Wang (2008) and Wan et al. (2010) used the RHtestV2 data homogenization package (Wang and Feng, 2007) for Canadian monthly mean wind speed data, and Si et al. (2018) applied the RHtestV4 for Tianjin (China) monthly mean wind speed data; (b) Petrovi c et al. (2008), Li et al. (2011), andPéliné-Németh et al. (2014) applied the Multiple Analyses of Series for Homogenization (MASH; Szentimrey, 1999Szentimrey, , 2008 to homogenize daily wind speed series for Ireland, for the greater Beijing area (China) and Hungary, respectively; (c) Štěpánek et al. (2013), Azorin-Molina et al. (2014), and Minola et al. (2016) used the AnClim package (Štěpánek, 2004) to detect sudden break points in monthly wind speed series for the Czech Republic, Spain and Portugal, and Sweden, respectively; (d) Guijarro (2015) and Azorin-Molina et al. (2018b) applied the Climatol package to detect artificial change points and adjust inhomogeneities in monthly wind speed time series in Spain and Portugal and Saudi Arabia, respectively; and (e) Laapas and Venäläinen (2017) Alexandersson, 1986) and the Maronna-Yohai test to monthly, seasonal, and annual aggregates. ...
... Observed DPWG data were recorded and supplied by the Australian Bureau of Meteorology (BoM; http://www.bom.gov.au/; last accessed 1 December 2018) at heights of~10 m above the land surface. DPWG data were measured by different types of anemometers, with Dines pressure tube anemometers used until the~1980s and replaced with Synchrotac 706 rotating cup anemometers in the Automatic Weather Stations (AWS) installed in the last two to three decades (for details of instruments, see Jakob, 2010;Cechet and Sanabria, 2012). This general replacement is the major cause of inhomogeneities in the observed series (Jakob, 2010), with Synchrotac anemometers showing a tendency towards to "overspeed" compared to the Dines ones; i.e., the sensor´s inertia causes the anemometer to continue spinning after the wind speed has decreased (Gorman, 2004). ...
... DPWG data were measured by different types of anemometers, with Dines pressure tube anemometers used until the~1980s and replaced with Synchrotac 706 rotating cup anemometers in the Automatic Weather Stations (AWS) installed in the last two to three decades (for details of instruments, see Jakob, 2010;Cechet and Sanabria, 2012). This general replacement is the major cause of inhomogeneities in the observed series (Jakob, 2010), with Synchrotac anemometers showing a tendency towards to "overspeed" compared to the Dines ones; i.e., the sensor´s inertia causes the anemometer to continue spinning after the wind speed has decreased (Gorman, 2004). ...
Daily Peak Wind Gust (DPWG) time series are important for the evaluation of wind‐related hazard risks to different socioeconomic and environmental sectors. Yet, wind time series analyses can be impacted by several artefacts, both temporally and spatially, which may introduce inhomogeneities that mislead the study of their decadal variability and trends. The aim of this study is to present a strategy in the homogenization of a challenging climate extreme such as the DPWG using 548 time series across Australia for 1941–2016. This automatic homogenization of DPWG is implemented in the recently developed Version 3.1 of the R package Climatol. This approach is an advance in homogenization of climate records as it identifies 353 break points based on monthly data, splits the daily series into homogeneous subperiods, and homogenizes them without needing the monthly corrections. The major advantages of this homogenization strategy are its ability to: (a) automatically homogenize a large number of DPWG series, including short‐term ones and without needing site metadata (e.g., the change in observational equipment in 2010/2011 was correctly identified); (b) use the closest reference series even not sharing a common period with candidate series or presenting missing data; and (c) supply homogenized series, correcting anomalous data (quality control by spatial coherence), and filling in all the missing data. The NCEP/NCAR reanalysis wind speed data were also trialled in aiding homogenization given the station density was very low during the early decades of the record; however, reanalysis data did not improve the homogenization. Application of this approach found a reduced range of DPWG trends based on site data, and an increased negative regional trend of this climate extreme, compared to raw data and homogenized data using NCEP/NCAR. The analysis produced the first homogenized DPWG dataset to assess and attribute long‐term variability of extreme winds across Australia.
