A Chilean program explores winds over the Atacama Desert region and is producing a public model and observational database in support of the development of wind energy projects.
Renewable energy, especially wind and solar, is an increasingly important field in applied meteorology and climate science. From the initial studies that explore the availability of these energy resources in a given area, to the forecasting models required to optimize the operation of wind or solar power plants, meteorological expertise is needed to design and execute measurement programs, and for the modeling of resource variability. Renewable energy poses special challenges compared to more traditional meteorological applications (Emeis 2013). In the case of wind power, for example, most phenomena of interest occur within the atmospheric boundary layer over horizontal scales ranging from the microscale to the mesoscale. Moreover, the viability of many renewable energy projects depends on the accuracy of the meteorological variables measured and modeled, with even small errors having large financial implications. On the other hand, the emergence of the renewable energy industry has led to a significant increase in the number of measuring sites being deployed worldwide, often in locations previously devoid of meteorological data. The commercial market for meteorological instrumentation has also responded, with the development of sensors tailored to satisfy specific renewable energy requirements. While this enhanced data availability and new instrumentation certainly has great scientific potential, it is true that because of the commercial interests behind most renewable energy projects, the associated meteorological information is usually not fully available to the general scientific community (e.g., Kusiak 2016). The present work describes a measurement and modeling program funded mainly by the Chilean government over the last several years that is aimed at generating public information about the potential of renewable energy resources across the country. As such, it has not only provided information upon which private investors have been able to develop new renewable energy projects, but has also built up a public meteorological database over previously data-void regions. The objective of this paper is to introduce this database to the meteorological community, describing the measurements and model results it contains, the tools available to access the information, and highlighting some interesting climatic features revealed by the observations.
With a latitudinal extent that spans from the tropics to the Southern Ocean (18°S to about 56°S in latitude and roughly along the 70°W meridian), Chile is subject to a large variety of climatic regimes. North of about 30°S, climate is largely conditioned by the southeast Pacific anticyclone, providing for a very dry and stable free troposphere. To the south, the weather is determined by the midlatitude westerlies and the synoptic modulation provided by the successive migration of low and high pressure systems. Zonally, the basic surface contrast provided by the ocean–continent interface to the west is further modulated inland by complex terrain that rises in less than 300 km from sea level up to the Andes Cordillera along the eastern border of Chile, with maximum altitudes above 5,000 m north of 30°S. While this wide variety of climates and topography led naturally to the assumption that there could be many regions in Chile where solar and wind energy potentials were high, the national energy agency (Ministry of Energy) realized in the early 2000s that a lack of reliable measurements and quantitative information about these resources was one of the main barriers for the development of renewable energy projects. A series of projects were then initiated with the collaboration of both national universities and international assistance agencies, dedicated to the compilation of historical databases, measurement campaigns, and numerical modeling of wind and solar resources over the country (Santana et al. 2014). While the spatial extent of some of these activities encompasses all of Chile, the focus of the present contribution is on the northern part of the country, where the Atacama Desert is located (Fig. 1). This area was prioritized by the energy agency because its energy demand is very high as a result of the presence of a large copper mining industry, while the aridity of the region makes hydroelectric energy generation (the principal renewable energy source in southern Chile) impossible. Therefore, in northern Chile solar and wind generation constitutes a very attractive alternative to the fossil fuel–based generation that has historically supplied energy to the region. The focus of the paper is further restricted to the exploration of the wind resource, with the efforts to measure and quantify the solar energy resource being documented in other reports (Rondanelli et al. 2015; Molina et al. 2017).
Location map showing 45 meteorological stations over the Atacama Desert (inset shows location of the region in South America). Colors denote the mean wind speed field simulated with the EE for 2010 at 100 m AGL. Letters and dashed ellipses indicate selected main zones for wind prospection. The bottom layer corresponds to a 30-m Shuttle Radar Topography Mission (SRTM) hillshade relief map.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Location map showing 45 meteorological stations over the Atacama Desert (inset shows location of the region in South America). Colors denote the mean wind speed field simulated with the EE for 2010 at 100 m AGL. Letters and dashed ellipses indicate selected main zones for wind prospection. The bottom layer corresponds to a 30-m Shuttle Radar Topography Mission (SRTM) hillshade relief map.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Location map showing 45 meteorological stations over the Atacama Desert (inset shows location of the region in South America). Colors denote the mean wind speed field simulated with the EE for 2010 at 100 m AGL. Letters and dashed ellipses indicate selected main zones for wind prospection. The bottom layer corresponds to a 30-m Shuttle Radar Topography Mission (SRTM) hillshade relief map.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
EXPLORATION PROGRAM.
Evolution.
The first efforts of the national energy agency to gather information on wind potential over Chile involved the compilation of existing wind data from private and public sources (DGF 1993; UNTEC 2003; CERE 2005). Numerical modeling and the acquisition of new measurements were only a small part of these efforts. Given that most of the data compiled were not originally intended for wind energy purposes, but for other objectives like air quality or aeronautical safety, it rapidly became apparent that the available information did not provide a true picture of the wind resource across the country. Therefore, subsequent efforts included the installation of new measurement sites and numerical modeling specifically oriented to wind potential evaluation. One line of work was the so-called Program for Rural Electrification (PER) program, a United Nations Development Program–funded measurement effort targeting off-grid rural communities that could use wind energy as a local source of electricity (Canales 2011). Under this program, 31 wind monitoring stations were operated over 1–3-yr periods between 2002 and 2007. A second line of work targeted potential locations for wind farms that could supply energy to the main electrical grids of the country. This work started in 2008 with a project that applied the Weather Research and Forecasting (WRF) Model to produce detailed wind fields for selected regions (UNTEC 2008a,b). The modeling results, together with information on technical, geographical, and land-ownership constraints for possible locations of wind farms, were used to select measurement sites for wind prospection in the northern region of the country, as described in the “Measurement sites” section below). The WRF modeling efforts, on the other hand, were subsequently extended and refined, giving rise to the so-called Wind Energy Explorer described in the following section. A timeline of the modeling and measurement activities carried out over the years is presented in Fig. 2.
Timeline of the national wind and solar exploration program. (top) Description of the observational efforts and measurement periods for the different selected zones, with color intensity representing higher station density. Blue is for 20-m towers, green for 60/80-m towers, yellow for solar monitoring sites, and red for PER stations. (bottom) Description of the modeling initiatives and the evolution of the EE: WF indicates wind farms in operation and LH and LV refer to the Loma del Hueso and Lengua de Vaca stations, respectively, which are the stations with the longest records in the observational database.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Timeline of the national wind and solar exploration program. (top) Description of the observational efforts and measurement periods for the different selected zones, with color intensity representing higher station density. Blue is for 20-m towers, green for 60/80-m towers, yellow for solar monitoring sites, and red for PER stations. (bottom) Description of the modeling initiatives and the evolution of the EE: WF indicates wind farms in operation and LH and LV refer to the Loma del Hueso and Lengua de Vaca stations, respectively, which are the stations with the longest records in the observational database.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Timeline of the national wind and solar exploration program. (top) Description of the observational efforts and measurement periods for the different selected zones, with color intensity representing higher station density. Blue is for 20-m towers, green for 60/80-m towers, yellow for solar monitoring sites, and red for PER stations. (bottom) Description of the modeling initiatives and the evolution of the EE: WF indicates wind farms in operation and LH and LV refer to the Loma del Hueso and Lengua de Vaca stations, respectively, which are the stations with the longest records in the observational database.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Modeling description.
Since 2008 the use of moderate-resolution mesoscale models, in particular the WRF Model, has played an increasing and pivotal role in the Chilean government’s efforts to improve the resource information available to wind energy stakeholders. We use the words “moderate resolution” to distinguish the regional (∼1-km resolution) modeling efforts described here from the very high-resolution (10–100 m) large-eddy simulation (LES) or computational fluid dynamics (CFD) models that have become popular in the industry for micrositing and other wind farm–scale applications (e.g., Uchida and Ohya 2008; Sanderse et al. 2011).
All modeling results presented in this work are based on simulations with the version 3.2 of the WRF mesoscale model (Skamarock et al. 2008). The WRF Model was, and remains today, perhaps the most popular mesoscale modeling system in the scientific community worldwide, and has proven to be a standard tool for wind energy prospecting at the regional scale (e.g., Mattar and Borvarán 2016; Nawri et al. 2014; Carvalho et al. 2014; Storm et al. 2009). As described in the previous section, the WRF modeling system was first applied to the northern region of the country, as part of a two-stage project involving regional model simulations followed by a 20-site prospecting campaign where the site selection was based on the model results. Given the favorable evaluation of this early project, the Chilean government continued to fund the development and application of the modeling system in parallel with the measurement campaigns, extending the spatial domain of the model to eventually cover the entire country.
The model configuration used to simulate the Atacama region underwent a rapid evolution during the early stages of the project. The initial version of the model used an essentially “off the shelf” configuration with low spatial resolution (3 km) and default vertical spacing. Because of limited computational resources, these simulations were restricted to just four months (January, April, July, and September) of a single year (2006). Despite the lack of any real customization, these early experiments were able to successfully guide the initial deployment of the 20-m tower sites that were installed during 2009. As the observations started to come in and a direct model evaluation was made possible, there began a rapid process of refinement and optimization of the original WRF setup. By 2011 a definitive model configuration was established, considered to be of sufficient quality for long-term model runs with public dissemination in mind. This setup was used initially to provide complete simulations of the 2010 calendar year for the central and northern portions of the country at 1-km resolution, and was later extended to cover the southern portion of Chile, along with Easter Island (Chilean territory in the Pacific).
The model configuration used for the definitive simulations is summarized in Table 1. Given the large extent of the region to be simulated at high horizontal resolution, it was necessary to divide the entire spatial domain into 22 subdomains (5 over the Atacama region) that were run independently. Each subdomain consisted of four nested computational grids that scale down from 27- to 1-km resolution. The average size of the innermost (1 km) grids is about 250 km × 250 km and together they cover practically all of continental Chile. The simulation period chosen was the year 2010. In general, the specific model settings were chosen based on sensitivity testing and evaluation against the newly available field data, or to reduce computational burden. Although many aspects of the model configuration do not deviate considerably from WRF standard practice, there are some important departures that deserve mention. First, a dense vertical spacing of 10 m was applied below the first 100 m above the model surface. This was found to be very important to correctly resolving the often-complex vertical structure of the nocturnal flow regime over the Atacama Desert characterized by the presence of very strong low-level jets [see the “Nocturnal drainage flows along valleys” sidebar and Muñoz et al. (2013)]. Needless to say, a high vertical resolution in the lowest levels is also of great practical importance for wind prospecting applications. After considerable sensitivity testing, the use of the quasi-normal scale elimination (QNSE; Sukoriansky et al. 2005, 2006) turbulence scheme was chosen based on its generally better performance for the nighttime regime. Given the extreme topographic gradients over the Atacama region, a small time step was required on the outermost domain to ensure numerical stability, and a smoothing filter was applied to the terrain elevation in regions of particularly steep terrain. The reduction of both the roughness length and default moisture content of the desert surface also proved helpful to better model the near-surface wind, temperature, and humidity. This is most probably due to the particularly extreme dryness and absence of vegetation over much of the Atacama compared to other desert regions worldwide. Interestingly, lessons learned in configuring the WRF Model for the drainage winds over the Atacama Desert aided subsequently the modeling of surface wind fields over Antarctica, where katabatic winds are common (Falvey and Rojo 2016).
Configuration of the WRF Model used for the EE.
Measurement sites.
Guided by the modeling results described above, a series of wind measurement campaigns in northern Chile started in 2009, originally funded by the German International Cooperation Agency (GIZ). The most common installation configuration consists of a 20-m mast erected on a concrete foundation and guyed to three lateral supporting points. Wind speed measurements are performed at 10 and 20 m AGL, while a wind vane registers wind direction at 10 m AGL. Depending on the site, additional variables like temperature, relative humidity, and atmospheric pressure are also measured. A datalogger records 10-min averages of the variables and 3-monthly visits allow for data collection and station checkup.
Figure 1 shows the locations of the measurement sites along with the mean velocity field at 100 m based on the currently available 1-km-resolution simulation for 2010. This wind map differs from that available for the initial prospecting campaign, but the same salient features are present in both, and it is easy to appreciate how the selected measurement sites cluster around “hot spots” of higher mean speed simulated by the model.
Over time, five main zones of interest have been investigated: zone A close to the mouth of the Loa River, zone B along the Loa River valley near the mining towns of Calama and Sierra Gorda, zone C to the east of the city of Antofagasta, zone D on an elevated plain to the northwest of the coastal town of Taltal, and the high-altitude zone E in the Chilean Altiplano, at nearly 4,500 m MSL (Fig. 1). Motivated by the promising winds found in some of the zones, six high instrumented towers (60/80 m) were erected during 2010–12 in zones B and D. Although all sites have minimum measuring periods of one year, many have been discontinued or relocated based upon budget availability, analysis of the wind data, and changes in energy policies. Nonetheless, as will be shown later, several measuring sites have completed more than five years of operation. In parallel, seven solar radiation monitoring sites were also maintained over the same period, each having wind measurements typically at 5 m AGL. Currently, there are eight zone D stations and four solar monitoring sites active in the Atacama Desert region. Together with the modeling results, all these observation sites are providing a more complete picture of the wind climatology over the region (see sidebar “Examples of circulation features over the Atacama Desert”).
EXAMPLES OF CIRCULATION FEATURES OVER THE ATACAMA DESERT
Based on the data gathered by these projects, a picture of the near-surface wind regime over several points in the Atacama Desert is beginning to emerge. An overriding characteristic of this regime is a strong diurnal variability in wind speed and direction, hardly surprising given the extreme insolation over the desert surface and the general lack of synoptic variability. The top panels in Fig. SB1 show the mean spatial pattern of mean diurnal (left) and nocturnal (right) wind speed at 80 m simulated by the WRF Model. The model shows the diurnal wind regime to be spatially homogenous, characterized by westerly oriented upslope flows that align with prevailing northwesterly winds at higher altitudes. In contrast, the nocturnal wind regime shows a more complex spatial structure characterized by a diverse range of flows, including stagnation zones, strong drainage systems (which in some locations exceed the diurnal maximum), and synoptically driven regimes at higher-elevation locations.
Wind regimes at different locations over the Atacama Desert. (top) The mean wind speed maps correspond to EE simulations for the year 2010 at 80 m AGL, for diurnal (1200–1800 UTC − 4 h) and nocturnal (0000–0600 UTC − 4 h) conditions. (bottom) The near-surface (10 m AGL) wind regimes at six observing sites (WD and WS correspond to wind direction and wind speed, respectively).
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Wind regimes at different locations over the Atacama Desert. (top) The mean wind speed maps correspond to EE simulations for the year 2010 at 80 m AGL, for diurnal (1200–1800 UTC − 4 h) and nocturnal (0000–0600 UTC − 4 h) conditions. (bottom) The near-surface (10 m AGL) wind regimes at six observing sites (WD and WS correspond to wind direction and wind speed, respectively).
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Wind regimes at different locations over the Atacama Desert. (top) The mean wind speed maps correspond to EE simulations for the year 2010 at 80 m AGL, for diurnal (1200–1800 UTC − 4 h) and nocturnal (0000–0600 UTC − 4 h) conditions. (bottom) The near-surface (10 m AGL) wind regimes at six observing sites (WD and WS correspond to wind direction and wind speed, respectively).
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
The bottom panels in Fig. SB1 present data for each of the five measuring zones A–E. The two panels for each station describe the diurnal variation of the wind direction frequencies and the association between wind speeds and wind directions. The stations in zone B (b21 and T80CN in Fig. SB1) are located along broad valleys running from high elevations to the west down to the lowlands in the east. Both show a conspicuous nocturnal wind regime characterized by very persistent down-valley wind directions (easterly in the case of b21 and northerly in the case of T80CN) and relatively high wind speeds, traits that are more marked during winter (not shown). Furthermore, the 80-m towers in this zone show that the vertical profiles of these nocturnal winds have a nose-like shape, with maximum speeds occurring between 20 and 60 m AGL (see the “Nocturnal drainage flows along valleys” sidebar). Muñoz et al. (2013) classified these winds as drainage winds and provided a more complete observational characterization of them. Stations in zones A and C are located along coastal valleys. As such, the wind regimes of stations a7 and c71 in Fig. SB1 show a strong and well-defined diurnal phase with a marked westerly wind direction and wind speeds that can reach up to 20 m s−1. A possible superposition of coastal and valley breeze forcings may explain these high diurnal winds, considering that stations closer to the coast have lower speeds (not shown). Station d02 in zone D illustrates the wind regime in the Taltal area, where a significant wind energy development is projected. While a diurnal cycle in wind direction is noticeable in the annual climatology of this station, the diurnal phase during winter is much reduced and strong northerly winds dominate, except for a brief weakening during the evening transition (not shown). The location of these sites near the top of a zonally oriented mountain chain around 2,000 m MSL suggests that the topography may be interacting here with a northerly barrier jet existing in this region at this altitude (Rutllant et al. 2013), a hypothesis that could eventually be tested with numerical model diagnostics of these winds. Finally, as predicted by the EE modeling results, the highest wind speeds are found in the high-elevation zone E (station e02 in Fig. SB1). The dominant wind direction is northwesterly, with weak diurnal and annual modulation. Wind speeds can exceed 25 m s−1 at 10 m AGL, being larger in the afternoon during the summer (not shown). While knowledge of these features of the near-surface wind climatology is essential for wind energy projects, the database produced by these projects should be of interest to scientists in diverse disciplines and at this time its use in geomorphological and biological studies of the Atacama Desert is already under way (M. Reyers 2018, personal communication).
Validation of model results.
The previous sections have alluded to the good performance of the WRF Model for site selection in northern Chile. In this section we provide a brief quantitative description of model performance, focusing on both spatial and temporal variability. The model results are those obtained from the definitive model configuration described in the “Modeling description” section. The model skill in predicting the spatial pattern of the wind field is assessed by comparing the annual wind speed means obtained by the model and the observations, as shown in Fig. 3. With a small bias, a determination coefficient of 0.85, and a root-mean-square error of 0.71 m s−1, the model is clearly able to detect regions with higher and lower mean wind speeds. The evaluation of the time variability of the WRF wind speed simulation is assessed by the correlation coefficients presented in Table 2, which were computed for concurrent models and observations in terms of daily, diurnal, and nocturnal averages, and for the full 2010 year, as well as distinguishing between the warm and cold seasons. Stations closer to the coast in zones A and C, which have strong diurnal and annual cycles in wind speed, show larger correlation coefficients for the diurnal phase at the annual scale. In contrast, zone B stations subject to a strong drainage wind regime (see the “Nocturnal drainage flows along valleys” sidebar ) tend to show the largest correlations in the nocturnal phase during the cold season, when these winds are stronger. Finally, stations in zone D show relatively high correlations that change little between the diurnal and nocturnal phases.
Scatterplot of annual mean wind speeds between the measured and modeled data. Modeled data correspond to EE simulations for 2010. Measured data correspond to averages of the year 2010 when available (red dots), or to averages over the maximum number of complete years when not available (blue dots). The dashed line represents a perfect fit and R2 is the determination coefficient.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Scatterplot of annual mean wind speeds between the measured and modeled data. Modeled data correspond to EE simulations for 2010. Measured data correspond to averages of the year 2010 when available (red dots), or to averages over the maximum number of complete years when not available (blue dots). The dashed line represents a perfect fit and R2 is the determination coefficient.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Scatterplot of annual mean wind speeds between the measured and modeled data. Modeled data correspond to EE simulations for 2010. Measured data correspond to averages of the year 2010 when available (red dots), or to averages over the maximum number of complete years when not available (blue dots). The dashed line represents a perfect fit and R2 is the determination coefficient.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Correlation coefficient R of wind speed (WS) averages between measured and modeled values. Nov–Feb (NDJF) are summer months and May–Aug (MJJA) are winter months. Diurnal hours are 0800–1600 UTC – 4 h and nocturnal hours are 1600–0800 UTC – 4 h. Note asterisk means R is not statistically significant (p value: 0.05).
NOCTURNAL DRAINAGE FLOWS ALONG VALLEYS
One of the more interesting features of the low-level circulation over the Atacama Desert is the presence of particularly strong nocturnal drainage flows with hourly averaged down-valley speeds reaching up to about 20 m s−1 during the cold season. The strongest of the jet systems are found in broad, gently sloping valleys that extend from the Andes to the lower plains of the Atacama Desert in the vicinity of the mining town of Calama, and are of great interest both as potential sites for wind farms and also for the transport and dispersion of air pollutants produced by mining operations.
Figure SB2 shows representative examples taken from the period 16–23 April 2010, during which a large amount of observational and model data were available, including those taken at the 80-m prospecting tower at Sierra Gorda and vertical wind profiles from a sodar instrument that was deployed specifically to verify the existence of the nocturnal jets. Both observations and models show that the flows are concentrated along the central axis of the valley with the nighttime flow direction coinciding with the terrain slope. The vertical wind profile within the jet region shows a nose-shaped structure with a mean nocturnal maximum close to 10 m s−1, although wind speeds may reach up to 20 m s−1 on occasion. The observed characteristics of these wind regimes and their predictability have been studied by Muñoz et al. (2013) and Jacques-Coper et al. (2015). The WRF Model simulations have also proven quite successful at capturing the salient features of the jets, including their spatial pattern, jet height, and temporal variability. Indeed, the presence of the nocturnal flows was largely unknown before the wind energy prospecting initiative began, and the placement of measurement sites at the primary jet locations was a direct result of the guidance provided by the WRF Model simulations. An animation of surface winds model results for a 4-day case of drainage flows in the same region is provided in the online supplemental material (https://doi.org/10.1175/BAMS-D-17-0019.2).
Observed and simulated data in the Sierra Gorda valley (SGORD site in Fig. 1). All three plots are based on observed and modeled data from 15 to 24 Apr 2010, a period during which several meteorological stations were deployed in the valley, including the SGORD 80-m mast and a sodar vertical wind profiler that was installed near the SGORD site specifically to verify the presence of the low-level nocturnal wind maximum. (top left) The 10-min observed (red circles) and hourly modeled winds at 40-m height at the SGORD station. The gray bands highlight the nighttime periods (0000–0800 LT). (right) The mean nocturnal vertical wind profiles derived from the 80-m meteorological tower (red dots), the sodar (green line, measurements at 10-m intervals), and from a high-vertical-resolution WRF Model simulation that was run for this observing period (blue line). There was significant missing data above the jet region in the sodar dataset and the shaded area denotes the fraction of missing sodar measurements as a function of altitude. (bottom left) A 3D visualization of the mean nocturnal wind velocity (40 m) overlaid onto the 3D shaded topography of the Sierra Gorda valley. The wind velocity is represented by arrows pointing in the mean flow direction with length and colors associated with the wind speed. An animation of the model results for a 4-day case of drainage flows in the same region is provided in the online supplemental material.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Observed and simulated data in the Sierra Gorda valley (SGORD site in Fig. 1). All three plots are based on observed and modeled data from 15 to 24 Apr 2010, a period during which several meteorological stations were deployed in the valley, including the SGORD 80-m mast and a sodar vertical wind profiler that was installed near the SGORD site specifically to verify the presence of the low-level nocturnal wind maximum. (top left) The 10-min observed (red circles) and hourly modeled winds at 40-m height at the SGORD station. The gray bands highlight the nighttime periods (0000–0800 LT). (right) The mean nocturnal vertical wind profiles derived from the 80-m meteorological tower (red dots), the sodar (green line, measurements at 10-m intervals), and from a high-vertical-resolution WRF Model simulation that was run for this observing period (blue line). There was significant missing data above the jet region in the sodar dataset and the shaded area denotes the fraction of missing sodar measurements as a function of altitude. (bottom left) A 3D visualization of the mean nocturnal wind velocity (40 m) overlaid onto the 3D shaded topography of the Sierra Gorda valley. The wind velocity is represented by arrows pointing in the mean flow direction with length and colors associated with the wind speed. An animation of the model results for a 4-day case of drainage flows in the same region is provided in the online supplemental material.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Observed and simulated data in the Sierra Gorda valley (SGORD site in Fig. 1). All three plots are based on observed and modeled data from 15 to 24 Apr 2010, a period during which several meteorological stations were deployed in the valley, including the SGORD 80-m mast and a sodar vertical wind profiler that was installed near the SGORD site specifically to verify the presence of the low-level nocturnal wind maximum. (top left) The 10-min observed (red circles) and hourly modeled winds at 40-m height at the SGORD station. The gray bands highlight the nighttime periods (0000–0800 LT). (right) The mean nocturnal vertical wind profiles derived from the 80-m meteorological tower (red dots), the sodar (green line, measurements at 10-m intervals), and from a high-vertical-resolution WRF Model simulation that was run for this observing period (blue line). There was significant missing data above the jet region in the sodar dataset and the shaded area denotes the fraction of missing sodar measurements as a function of altitude. (bottom left) A 3D visualization of the mean nocturnal wind velocity (40 m) overlaid onto the 3D shaded topography of the Sierra Gorda valley. The wind velocity is represented by arrows pointing in the mean flow direction with length and colors associated with the wind speed. An animation of the model results for a 4-day case of drainage flows in the same region is provided in the online supplemental material.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
PRODUCTS DESCRIPTION.
Observations database.
The observational database compiled in the projects described earlier is publically available at http://walker.dgf.uchile.cl/Mediciones/ (Fig. 4), and includes data for a total of 83 stations over the entire country. Figure 5 and Table 3 show the measuring periods and variables registered at the 45 stations located over the Atacama Desert, as shown in Fig. 1 (31 towers of 20-m height, six high towers, and eight solar radiation stations). As of July 2017, the program maintains eight active 20-m towers in zone D, two 20-m towers at the coast around 30° and 31°S (operating since 2006), and four solar sites. For every station, the website provides access to the original raw data as well as to consolidated files in Excel and ASCII formats. Metadata are also available in the form of installation reports, visit logs, instrument descriptions, and calibration certificates when available. Current developments involving the database include 1) a detailed quality control revision of the data, eliminating and documenting periods with suspicious data and 2) the generation of figures for each site, providing a rapid depiction of the climatological regime revealed by these measurements (diurnal and annual cycles, wind and direction distributions, etc.).
Screenshot of the website hosting the observational database named Campaña de medición del recurso Eólico y Solar (Measurement campaign of solar and wind energy resources). The site includes (top left) a list of the stations, (top right) a location map, (bottom left) location information for the stations, and (bottom right) an instrument register table. Additionally, all metadata are available for each station, and the original raw data and consolidated files are available via download links (arrowed boldface blue words).
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Screenshot of the website hosting the observational database named Campaña de medición del recurso Eólico y Solar (Measurement campaign of solar and wind energy resources). The site includes (top left) a list of the stations, (top right) a location map, (bottom left) location information for the stations, and (bottom right) an instrument register table. Additionally, all metadata are available for each station, and the original raw data and consolidated files are available via download links (arrowed boldface blue words).
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Screenshot of the website hosting the observational database named Campaña de medición del recurso Eólico y Solar (Measurement campaign of solar and wind energy resources). The site includes (top left) a list of the stations, (top right) a location map, (bottom left) location information for the stations, and (bottom right) an instrument register table. Additionally, all metadata are available for each station, and the original raw data and consolidated files are available via download links (arrowed boldface blue words).
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Time periods with wind data availability for selected sites in the observational database. When 20-m data were nonexistent, other heights were considered, such as 12, 10, and 4 m AGL.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Time periods with wind data availability for selected sites in the observational database. When 20-m data were nonexistent, other heights were considered, such as 12, 10, and 4 m AGL.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Time periods with wind data availability for selected sites in the observational database. When 20-m data were nonexistent, other heights were considered, such as 12, 10, and 4 m AGL.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Summary of the meteorological variables measured and their measurement heights (values in meters) available for 45 observation sites in the Atacama Desert.
Modeling products.
Although the initial intention of the WRF simulations was to guide site selection for wind prospecting in the northern region of Chile, an important by-product of these efforts was the development of the so-called Explorador Eólico (EE; Wind Energy Explorer), which has since proved to be a valuable tool for wind energy prospection, resource assessment, policy making, and outreach.
The EE is an Internet-based tool that provides rapid and free access to the results of the WRF simulations described in the “Modeling description” section above (Fig. 6). Unlike many wind atlas products, the EE aimed to provide not just static wind maps, but full and dynamic access to the underlying time series of wind data. Such data are necessary to provide a genuine understanding of the behavior of the wind resource at sites of interest.
Screenshot of the EE. This example shows the main capabilities of the site. An interactive map allows the display of mean wind fields for user-defined time periods and altitudes. In the example shown, the mean 45-m wind field at 0400 UTC – 4 h during Aug 2010 is shown and the zones where strong katabatic low-level jets form may be clearly distinguished. By clicking on the map, users may request quick-look plots (the VISOR) showing climatological profiles (in the example, monthly mean vertical profiles at a nocturnal jet location have been selected) or download a wind resource assessment report along with data files containing the complete hourly data.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Screenshot of the EE. This example shows the main capabilities of the site. An interactive map allows the display of mean wind fields for user-defined time periods and altitudes. In the example shown, the mean 45-m wind field at 0400 UTC – 4 h during Aug 2010 is shown and the zones where strong katabatic low-level jets form may be clearly distinguished. By clicking on the map, users may request quick-look plots (the VISOR) showing climatological profiles (in the example, monthly mean vertical profiles at a nocturnal jet location have been selected) or download a wind resource assessment report along with data files containing the complete hourly data.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
Screenshot of the EE. This example shows the main capabilities of the site. An interactive map allows the display of mean wind fields for user-defined time periods and altitudes. In the example shown, the mean 45-m wind field at 0400 UTC – 4 h during Aug 2010 is shown and the zones where strong katabatic low-level jets form may be clearly distinguished. By clicking on the map, users may request quick-look plots (the VISOR) showing climatological profiles (in the example, monthly mean vertical profiles at a nocturnal jet location have been selected) or download a wind resource assessment report along with data files containing the complete hourly data.
Citation: Bulletin of the American Meteorological Society 99, 10; 10.1175/BAMS-D-17-0019.1
The EE interface is centered around an interactive map over which wind data may be displayed. Tools are provided that allow users to select locations on the map and subsequently bring up rapid visualizations of key wind parameters, extract detailed site reports, and download data files with complete hourly data. Additional capabilities include calculating wind turbine generation from a public domain database of over 600 turbine types and an experimental statistical reconstruction method for estimating long-term variability based on correlating the WRF results for 2010 with large-scale variables from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis, and then using the relationships to reconstruct winds over an extended time period (1980–2016). Since its inception in 2012, the EE has supported a growing user base (since January 2016, an average of 46 sessions each day) and has generated more than 70,000 site resource assessment reports. Work is currently under way to implement a new, more modern interface for the EE and to add new WRF simulations that have recently been performed for the 2015 calendar year.
CONCLUSIONS.
The program described herein has demonstrated synergistic interactions along several axes. Considering a modeling/measurement axis, for example, the fact that the location of most of the prospecting sites was initially decided based mainly on numerical model results, and later the observed data helped in validating and improving the model configuration, closes a virtuous circle that is not frequently seen at this scale in South America. The projects have also demonstrated clear synergies between applied and scientific research. While the projects were driven by a very concrete and applied objective related to fostering wind energy development, it has produced a public database that is already being used in academic pursuits (Rondanelli et al. 2015; Watts et al. 2016). The scientific use of these observational and modeling results has led to the documentation of strong drainage winds along valleys traversing the Atacama Desert (Muñoz et al. 2013; Jacques-Coper et al. 2015), while the interesting wind regime in the Taltal region may deserve a more detailed study in the near future as well. Recent wind energy developments in northern Chile, on the other hand, provide a clear demonstration that the applied objective of the projects is being accomplished: based on the model results and the observations in zone D, the Chilean government has reserved close to 300,000 Ha of public land for wind farm developments, the first of which, with 99-MW nominal capacity, has been supplying energy to the national electric grid since 2014 and has become one of the wind farms with the best performance indices in the country. Other wind energy projects have also been developed in the Calama and Sierra Gorda areas (zone B), and several more are in the planning stages elsewhere. Finally, the projects also provide a compelling example of a fruitful interinstitutional collaboration because they were led by governmental agencies, received support for the planning and execution of the observations and modeling by university units, and benefited from financial and technical assistance from international agencies. While coordinating the interaction of all these different organizations has not always been easy, the experience has been replicated with different nuances to explore Chile`s solar, hydroelectric, and tidal renewable energies (Santana et al. 2014). The publicly available products of all these efforts, including measurements and model results (accessible online at www.energia.gob.cl/energias-renovables), have been in part responsible for the increasing share of nonconventional renewable energy sources in Chile’s energy matrix.
ACKNOWLEDGMENTS
This article was supported by the Chilean Ministry of Energy under Exempt Decrees 348 of 2015 and 645 of 2017. The measurement campaigns in northern Chile and the Wind Energy Explorer were partially funded by the German International Cooperation (GIZ), the Chilean National Energy Commission (pre-2010), and the Chilean Ministry of Energy (post-2010). The authors appreciate the comments and suggestions provided by two anonymous reviewers.
REFERENCES
Canales, C. C., 2011: Barriers elimination for the use of renewable energies as electricity sources in rural areas (in Spanish). Ministry of Energy Final Rep., Project GEF-Ministerio de Energía-PNUD CHI/00/G32 (11799), 151 pp. [Available from the authors.]
Carvalho, D., A. Rocha, M. Gomez, and C. Santos, 2014: WRF wind simulation and wind energy production estimates forced by different reanalyses: Comparison with observed data for Portugal. Appl. Energy, 117, 116–126, https://doi.org/10.1016/j.apenergy.2013.12.001.
CERE, 2005: Improvement of knowledge and administration of wind information in Chile: Second stage (in Spanish). Centro de Estudio de los Recursos Energéticos Tech. Rep., University of Magallanes, Punta Arenas, Chile, 74 pp., www.cne.cl/archivos_bajar/INFORME_FINAL_CERE_para_CNE_2005_after_DMC2.pdf.
DGF, 1993: EOLO Project: Evaluation of the national wind potential (in Spanish). Dept. of Geophysics Tech. Rep., University of Chile, Santiago, Chile, 159 pp., www.jimic.cl/appv/Congreso/Proyecto_EOLO/INFORME/.
Emeis, S., 2013: Wind Energy Meteorology. Springer, 196 pp.
Falvey, M., and P. Rojo, 2016: Application of a regional model to astronomical site testing in western Antarctica. Theor. Appl. Climatol., 125, 841–862, https://doi.org/10.1007/s00704-016-1794-x.
Jacques-Coper, M., M. Falvey, and R. C. Muñoz, 2015: Inter-daily variability of a strong thermally-driven wind system over the Atacama Desert of South America: Synoptic forcing and short-term predictability using the GFS global model. Theor. Appl. Climatol., 121, 211–223, https://doi.org/10.1007/s00704-014-1231-y.
Kusiak, A., 2016: Renewables: Share data on wind energy. Nature, 529, 19–21, https://doi.org/10.1038/529019a.
Mattar, C., and D. Borvarán, 2016: Offshore wind power simulation by using WRF in the central coast of Chile. Renewable Energy, 94, 22–31, https://doi.org/10.1016/j.renene.2016.03.005.
Molina, A., M. Falvey, and R. Rondanelli, 2017: A solar radiation database for Chile. Sci. Rep., 7, 14283, https://doi.org/10.1038/s41598-017-13761-x.
Muñoz, R. C., M. J. Falvey, M. Araya, and M. Jacques-Coper, 2013: Strong down-valley low-level jets over the Atacama Desert: Observational characterization. J. Appl. Meteor. Climatol., 52, 2735–2752, https://doi.org/10.1175/JAMC-D-13-063.1.
Nawri, N., G. Nína, H. Bjornsson, A. Hahmann, K. Jonasson, C. Hasager, and N. Clausen, 2014: The wind energy potential of Iceland. Renewable Energy, 69, 290–299, https://doi.org/10.1016/j.renene.2014.03.040.
Rondanelli, R., A. Molina, and M. Falvey, 2015: The Atacama surface solar maximum. Bull. Amer. Meteor. Soc., 96, 405–418, https://doi.org/10.1175/BAMS-D-13-00175.1.
Rutllant, J. A., R. C. Muñoz, and R. D. Garreaud, 2013: Meteorological observations on the northern Chilean coast during VOCALS-Rex. Atmos. Chem. Phys., 13, 3409–3422, https://doi.org/10.5194/acp-13-3409-2013.
Sanderse, B., S. P. Van der Pijl, and B. Koren, 2011: Review of computational fluid dynamics for wind turbine wake aerodynamics. Wind Energy, 14, 799–819, https://doi.org/10.1002/we.458.
Santana, C., M. Falvey, M. Ibarra, and M. García, 2014: Renewable energies in Chile: Wind, solar, and hydroelectric potentials from Arica to Chiloé (in Spanish). Ministry of Energy Rep., Santiago, Chile, 158 pp., www.minenergia.cl/archivos_bajar/Estudios/Potencial_ER_en_Chile_AC.pdf.
Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., https://dx.doi.org/10.5065/D68S4MVH.
Storm, B., J. Dudhia, S. Basu, A. Swift, and I. Giammanco, 2009: Evaluation of the Weather Research and Forecasting Model on forecasting low-level jets: Implications for wind energy. Wind Energy, 12, 81–90, https://doi.org/10.1002/we.288.
Sukoriansky, S., B. Galperin, and I. Staroselsky, 2005: A quasinormal scale elimination model of turbulent flows with stable stratification. Phys. Fluids, 17, 085107, https://doi.org/10.1063/1.2009010.
Sukoriansky, S., B. Galperin, and V. Perov, 2006: A quasi-normal scale elimination model of turbulence and its application to stably stratified flows. Nonlinear Processes Geophys., 13, 9–22, https://doi.org/10.5194/npg-13-9-2006.
Uchida, T., and Y. Ohya, 2008: Micro-siting technique for wind turbine generators by using large-eddy simulation. J. Wind Eng. Ind. Aerodyn., 96, 2121–2138, https://doi.org/10.1016/j.jweia.2008.02.047.
UNTEC, 2003: Improvement of the knowledge on wind resources in northern and central Chile (in Spanish). Foundation for Technological Transfer (UNTEC) Tech. Rep., University of Chile, Santiago, Chile, 131 pp., www.cne.cl/archivos_bajar/InformeFinalEolo2003.pdf.
UNTEC, 2008a: Selection of sites of interest for wind prospection in the northern electric grid region: Stage 1 (in Spanish). Foundation for Technological Transfer Tech. Rep., University of Chile, Santiago, Chile, 98 pp. [Available from the authors.]
UNTEC, 2008b: Selection of sites of interest for wind prospection in the northern electric grid region: Stage 2 (in Spanish). Foundation for Technological Transfer Tech. Rep., University of Chile, Santiago, Chile, 155 pp. [Available from the authors.]
Watts, D., N. Osses, and R. Pérez, 2016: Assessment of wind energy potential in Chile: A project-based regional wind supply function approach. Renewable Energy, 96, 738–755, https://doi.org/10.1016/j.renene.2016.05.038.
