Solar energy services
Introduction
Company
meteoblue is a Swiss specialist company producing high precision weather data for the entire world, using observation data, high-resolution Numerical Weather Predictions (NWP) and specialised data output methods adapted to the needs of different user groups. Based on these simulations, meteoblue produces solar radiation forecast and data validation services. meteoblue produces weather data since 2007, and produces the largest daily available data volume of any private EU weather service. The available weather archives cover more than 30 years in maximum detail, which is important for risk assessment, as well as for verification purposes. Quality verification results are shown on: https://content.meteoblue.com/en/verified-quality/verification.
Distribution
meteoblue offers products, services and project resources to clients worldwide. For representation in certain countries or market segments, meteoblue works with selected distributors, who represent, sell and service meteoblue products, services and /or project resources. meteoblue offers datafeeds specifically designed for the needs of solar power generators, electricity traders, grid or building management. More information is provided in the documents: This document contains the specifications for radiation simulations and comparative measurements, and for related services. More product information is provided in the documents:
- meteoblue_Solar_Controlled_Quality_EN.pdf
- meteoblue_Solar_Forecast_Pricing+Ordering_EN.pdf
- meteoblue_Solar_History_Pricing+Ordering_EN.pdf
Approaches
meteoblue addresses the requirements of the solar energy sector with 4 complementary approaches, which can be combined or individually applied for fulfilment of the purposes:
- System Layout (Storage Management)
Analysis of radiation variability based on climate data of the specific site. Hourly time series of 30 years allow advanced statistical analyses to optimize sizing and management strategy for the solar energy generators and connected storage devices. Validation of expected errors for day-ahead and intraday forecast and delivery of correspondent time series to optimize sizing and management strategy for the storage devices. - Probabilistic Day-Ahead Forecast (Multi Model)
Calculation of a 6-day hourly multi-model ensemble forecast for specified locations, to assess the potential 1% (min), 10%, 25%, 50%, 75%, 90% and 99% (max) percentile distribution of the radiation forecast for determination of uncertainty. - Deterministic Day-Ahead Forecast (MOS)
Calculation of a high precision quarter-hourly radiation (and/or PV production) forecast for each location / substation with known past radiation / production data for refinement of forecast, based on statistical post processing (MOS). - Intraday Forecast (Nowcasting)
Calculation of a precision remote-sensing 5-hour ahead nowcasting with hourly update for radiation based on satellite data and cloud-motion technology for each location / substation. - Real Time Monitoring (Satellite Observation)
Benchmarking of the performance of individual PV systems, based on the satellite derived radiation observation of each day or month. All approaches have been developed and are operated by meteoblue in services or projects worldwide.
Applications
General
Radiation simulations and measurement serve to assess the influence of the sun on the weather, Earth surface, vegetation, buildings, as well as on other human activity. There is a very wide range of applications like:
- Site search and assessment
- System layout
- Long term yield prospection
- System failure detection
- Measurement quality controlling
- Real time monitoring
- Electricity balancing and trading
- Smart home management
Energy Generation
Beside different radiation forecast and parameterisations, meteoblue offers electricity forecast for single Photovoltaic systems and for system aggregations in a defined grid area. These forecasts are based on the same algorithms, using projection on fix inclined surfaces. Diffuse and direct radiation shares are used to calculate the radiation on a specified module surface. Based on position, system size (kWp) and sun exposition, the electricity output is modelled. For further information, on electricity forecasts check the specific product documentation.
Other
Testing in other applications is recommended. ExRAD can be used for calibration of radiation and power measurements, especially to detect time shifts, system configuration differences and calibration problems.
Reference System
For radiation measurements, the following reference systems are recommended:
- Extra-terrestrial radiation: Available worldwide, minimal error source (but not at surface);
- Surface measurements: These are most accurate (when well maintained), but seldom available and subject to large methodical errors.
- Satellite observations: Widely available, consistent source of surface incoming radiation. Most important source for benchmarking surface measurements and used for real time monitoring and short-range forecasting up to 5 hours.
- Other forecasts: Available but not always consistent with reality. Useful to calculate probabilities;
- Other reference systems, such as heating consumption, PV power production. However problems with data collection , normalization and consistency are substantial. In summary, a forecast can only be validated if it is compared to a valid reference system.
Improvements
If a valid reference system exists, simulation and forecast can be improved using statistical post- processing methods like MOS (model output statistics) or neural networking techniques (more speculative). Such improvement are only applied on hourly resolution data. As at least one year of quality controlled measurement data is necessary, such post-processing is only applied for high value projects.
Solar History Products
Applications for System Layout(Site Assessment)
meteoblue simulated data is available in hourly resolution since 1986 with an availability of 100 %. The coverage is worldwide with a spatial resolution 3-30 km. Based on this enormous meteorological database meteoblue offers beside its raw data various data aggregations, tools, images report customized for different applications. For site search and assessment maps and long term time series of data are used:
Maps (Site Search)
Maps of mean annual GHI sums are useful to find the optimum location. Variability maps help to find optimized operation strategies (see Figure below). Other maps are available within weather maps or can be customized on request.
TMY (Typical Meteorological Year)
A TMY is a hourly dataset that contains one year with 8760 data values which is considered as typical for the specific location. Based on a long term time series (10, 20 or 30 years) cumulative distribution function of each month are calculated. With the so called SANDIA method the 12 typical months with minimal Finkelstein-Schäfer distance are chosen and merged to the TMY.
Extreme Year Probabilities
For a sophisticated layout of modern energy systems more than only a TMY is needed. For an optimized sizing of storage devices and energy systems it is necessary to accurately estimate the heating/cooling and electricity demand in extreme years. Time series of the last 30 years allow an appropriate estimate of typical and extreme years. These can be easily accessed and analysed with an probabilistic approach that helps you to choose the right year for your application by variable (radiation, temperature, heating degree days) and risk aversion (percentile distribution). Therefore the 30 years of the reference climate period are sorted as their radiation sum (or heating degree days etc.). Then the extreme year (P01, P99) or other statistically significant confidence intervals can be chosen (P10, P90). This allows the exact assessment of frequency and amplitude of years with temperatures or radiation values, that are extremely high or low. This knowledge is a key advantage to improve risk assessment and project development within the layout of your building management or energy system.
Solar Report - Climatology and Variability
Version: 0.1 Created: 2016-05-10 / MB Last edited: 2017-07-07 / MB Based on hourly long term radiation data the variability of irradiation of a certain location is analysed on different time scale (Inter-annual, seasonal and intra-daily variability). The results are visualized with comprehensible graphics and summarized in a report. An example of typical intra-daily profiles is shown in Figure below:
Aggregated Time Series
meteoblue offers daily and monthly data aggregations of radiation variables and reference yield simulations. As all aggregation are based on hourly data variability information like standard deviation, confidence intervals or min-max values can also be delivered. An example of a monthly aggregated TMY of PV reference yield simulation is the Meteogram solar_PVsimple which is available via API (see Figure below).
Meteogram Solar_PVsimple
Apply: &type=solar_PVsimple
Special parameters: ¶ms=kWp,facing angle,slope angle
URL example:
Solar Seport-Pv System Layout
Version: 0.1 Created: 2016-05-10 / MB Last edited: 2017-07-07 / MB This report compares the seasonal distribution of monthly PV reference yield for your specific site for 20 different system layouts (PV panel inclination and orientation) based on the meteogram solar_PVsimple (see Figure above).
Transmission
The data can be accessed via API which is for customers that need automatic integration in their web applications. For customers with single request we deliver the data via email or web download in meteoblue standard JSON or CSV format. Furthermore daily or monthly reports can be transmitted via email or FTP.
API Datapackage: History-Solar
Variable | Unit | Description | Intervals in minutes | Intervals in hours | Daily aggregations |
---|---|---|---|---|---|
GHI (solar radiation) | W/m$^2$ | Global Horizontal Radiation | 5, 10, 15, | 1, 24 | Total |
DIF | W/m$^2$ | Diffuse Radiation | 5, 10, 15, | Total | |
DNI | W/m$^2$ | Direct Normalized Irradiance (Radiation) | 5, 10, 15, | 1, 24 | Total |
GNI | W/m$^2$ | Global Normalized Irradiance (Radiation) | 5, 10, 15, | 1, 24 | Total |
ExRAD | W/m$^2$ | Extraterrestrial solar radiation | 5, 10, 15, | 1, 24 | Total |
Example API-URL:
Example API output (&format=csv):
{
"metadata":
{
"name": "Basel",
"latitude": 47.56,
"longitude": 7.57,
"height": 279,
"timezone_abbrevation": "CEST",
"utc_timeoffset": 2.00,
"modelrun_utc": "4716-04-07 12:00"
},
"units":
{
"time": "YYYY-MM-DD hh:mm",
"radiation": "Wm-2"
},
"history_1h":
{
"time": ["2015-01-01 00:00", "2015-01-01 01:00", "2015-01-01 02:00", "2015-01-01
03:00", "2015-01-01 04:00", "2015-01-01 05:00", "2015-01-01 06:00", "2015-01-01
07:00", "2015-01-01 08:00", "2015-01-01 09:00", "2015-01-01 10:00", "2015-01-01
11:00", "2015-01-01 12:00", "2015-01-01 13:00", "2015-01-01 14:00", "2015-01-01
15:00", "2015-01-01 16:00", "2015-01-01 17:00", "2015-01-01 18:00", "2015-01-01
19:00", "2015-01-01 20:00", "2015-01-01 21:00", "2015-01-01 22:00", "2015-01-01
23:00", "2015-01-02 00:00", "2015-01-02 01:00", "2015-01-02 02:00", "2015-01-02
03:00", "2015-01-02 04:00", "2015-01-02 05:00", "2015-01-02 06:00", "2015-01-02
07:00", "2015-01-02 08:00", "2015-01-02 09:00", "2015-01-02 10:00", "2015-01-02
11:00", "2015-01-02 12:00", "2015-01-02 13:00", "2015-01-02 14:00", "2015-01-02
15:00", "2015-01-02 16:00", "2015-01-02 17:00", "2015-01-02 18:00", "2015-01-02
19:00", "2015-01-02 20:00", "2015-01-02 21:00", "2015-01-02 22:00", "2015-01-02
23:00", "2015-01-03 00:00", "2015-01-03 01:00", "2015-01-03 02:00", "2015-01-03
03:00", "2015-01-03 04:00", "2015-01-03 05:00", "2015-01-03 06:00", "2015-01-03
07:00", "2015-01-03 08:00", "2015-01-03 09:00", "2015-01-03 10:00", "2015-01-03
11:00", "2015-01-03 12:00", "2015-01-03 13:00", "2015-01-03 14:00", "2015-01-03
....
],
...
}
}
History+
The easiest way to access and analyse meteoblue solar history data is to buy history+ for the desired location. It enables analyses and unlimited downloads of 30 years hourly time series for 100€ only. history+ includes 12 weather variables like solar radiation, temperature, cloud cover, precipitation, snowfall, wind speed and some more. For more information visit: www.meteoblue.com/historyplus
Solar Data Management and Evaluation
Reference Datasets
Time series of satellite observed radiation and model simulations for any specific time range are used as reference to benchmark measurement data, detect system failure or asset power plant production. Furthermore comparison of PV production data and simulations are helpful to calibrate forecast services and for validation purposes.
Solar Report - Quality Control
Figure: Plausibility Verification of Measurement Data from Bari, Italy.
Based on reference systems, meteoblue offer quality controlling of radiation and PV production data (see Figure 3.1). These reports help to optimize measurement layout and find errors of temporal allocation and data irregularities. Quality controlling is inevitable, when measurement are used for post processing purposes.
Solar Report - Model Validation
A validation report compares simulation data to on site measurement. It supports your system operation and power management decision with forecast accuracy assessments. Every Validation Report contains:
- Monthly graphic comparison detects seasonal malfunction of your system or microclimatic forecast mismatch (see Figure 3.2).
- Daily analysis show every day forecast performance and system outages while the intraday graphics can detect time of day shadings.
- Precision of forecast and possible measures for improvement.
Solar Forecasting Products
Description
meteoblue solar forecast is delivered via API, FTP or Email and offers all common radiation variables and localized PV reference yield forecast based on the forecast of multiple weather models and satellite based nowcasting. It reproduces local radiation development for 144 hours ahead in 1 hour intervals, which can be further refined into 15-10 minute intervals. Solar radiation variables and power output are calculated on an hourly basis. It may be downscaled to any minutely intervals. During daytime the forecast API is updated every 15 minutes, based on the latest satellite image. Thus it integrates real time monitoring, hour ahead and day ahead forecasting in one product. The spatial resolution depends on the model coverage. The current status of model coverage is described in chapter 5.2, continuously updated information can be found on the following page: https://content.meteoblue.com/research-development/data-sources/nmm-modelling/model-domain The process of solar power forecasting model is determined by a simple algorithm, which includes the capacity of the photovoltaic systems (kWp), the performance ratio (default= 85 %) and the solar irradiance on tilted surfaces. The main modelling procedure is the transformation of global radiation to irradiance on a tilted surface. First, the diffuse fraction of the global radiation must be estimated. If the diffuse and direct fractions of the global radiation are known, they can be projected on inclined module surfaces. For single locations, the orientation and the inclination angle are needed. This information can be automatically detected based on time series of measured PV yield.
Applications for Real Time Monitoring(Nowcast)
The nowcasting function of all solar APIs offers an excellent reference for real time monitoring of measurement instrument or power plant production. As the delay of satellite data availability is not constant it is recommended to retrieve monitoring data once a day just after sunset. Then the API contains satellite observation for the past day and forecast values for the next days. The observation data of the actual day can be used for monitoring purposes like:
- Performance ratio analyses
- Detect system anomalies and irregularities with measuring equipment
- Snow warning
- Detect system failures and maximize production All meteoblue solar forecasts benefit from nowcasting technique. For further information, see chapter Nowcasting.
Applications for Forecasting
The Intraday and Day Ahead forecast of the solar forecast API is offered worldwide. It contains various variables in customized time resolution for current day to 7 days ahead. It is used for many applications like:
- Electricity Trading
- Grid Management
- Maintenance scheduling
- Reporting to grid operator
- Optimisation of own consumption
- Building management
- Heating and Cooling control.
For more information read chapter Solar forecasting products & chapter Radiation forecasting by Numerical modelling
Forecast Variables
The available solar forecast variables and range/intervals are:
- GHI = global horizontal irradiation (W/m$^2$): 144 h ahead, 24 updates per day.
- GNI = global normal irradiation (W/m$^2$): 144 h ahead, 24 updates per day.
- DNI = direct normal irradiation (W/m$^2$): 144 h ahead, 24 updates per day.
- DIF = diffuse irradiation (W/m$^2$): 144 h ahead, 24 updates per day.
- ExRAD = Extraterrestrial Irradiation (W/m$^2$): 144 h ahead.
- PVpp = Photovoltaic reference yield (kW): 144 h ahead, 24 updates per day.
- GTI = global tilted irradiation (W/m$^2$): 144 h ahead, 24 updates per day.
- mT = module temperature (°C): 144 h ahead, 24 updates per day.
- IAM = incidence angle modifier: 144 h ahead, 2 updates per day.
- SC = snow cover (mm) 144 h ahead, 2 updates per day.
- PR = performance ratio 144 h ahead, 24 updates per day.
Many other variables are available or can be supplied on request.
API Data Package: Solar
Variable | Unit | Description | Intervals in minutes | Intervals in hours | Daily aggregations |
---|---|---|---|---|---|
GHI (solar radiation) | W/m$^2$ | Global Horizontal Radiation | 5, 10, 15, | 1, 24 | Total |
DIF | W/m$^2$ | Diffuse Radiation | 5, 10, 15, | Total | |
DNI | W/m$^2$ | Direct Normalized Irradiance (Radiation) | 5, 10, 15, | 1, 24 | Total |
GNI | W/m$^2$ | Global Normalized Irradiance (Radiation) | 5, 10, 15, | 1, 24 | Total |
ExRAD | W/m$^2$ | Extraterrestrial solar radiation | 5, 10, 15, | 1, 24 | Total |
Example API-URL:
Example API output (&format=csv):
time gni_instant gni_backwards dni_instant dni_backwards dif_instant dif_backwards ghi_instant ghi_backwards extraterrestrialradiation_instant extraterrestrialradiation_backwards
2019-06-28 03:00 0 0 0 0 0 0 0 0 0 0
2019-06-28 04:00 0 0 0 0 0 0 0 0 0 0
2019-06-28 05:00 0 0 0 0 0 0 0 0 0 0
2019-06-28 06:00 133.32 26.56 121.68 24.2 9.68 1.95 20.29 4.06 63.15 10.78
2019-06-28 07:00 481.85 324.78 409.18 286.09 67.84 34.13 151.09 74.39 269.11 166.58
2019-06-28 08:00 700.42 610.24 545.78 494.04 144.07 108.74 343.4 251.63 483.03 377.8
2019-06-28 09:00 825.16 766.28 604.75 577.16 202.67 174.85 518.34 432.9 6 90.35 589.57
....
Variable | Unit | Description | Intervals in minutes | Intervals in hours | Daily aggregations |
---|---|---|---|---|---|
PV power | kWh | Photovoltaic power | 5, 10, 15, | 1, 24 | Total |
GTI | W/m$^2$ | Global Tilted Irradiance (Radiation) | 5, 10, 15, | Total | |
PR | % | Performance ratio | 5, 10, 15, | 1 | Mean |
mT | °C | Module temperature | 5, 10, 15, | 1, 24 | Mean |
IAM | % | Incidence Angle Modifier | 5, 10, 15, | 1 | Mean |
Snow cover | cm | On the PV modules | 5, 10, 15, | 1, 24 | Mean |
Special parameters: ¶ms=kWp, facing angle, slope angle
Example API-URL:
time pvpower_backwards pvpower_instant gti_backwards gti_instant moduletemperature_backwards moduletemperature_instant iam_backwards iam_instant snowcover performanceratio
2019-06-28 03:00 -999 -999 0 0 23.85 23.61 78 0.87 0 0.86
2019-06-28 04:00 -999 -999 0 0 22.85 22.02 90 0.93 0 0.93
2019-06-28 05:00 -999 -999 0 0 21.16 20.48 93 0.94 0 0.93
2019-06-28 06:00 -999 -999 24.21 122.85 20.71 21.44 94 0.95 0 0.83
2019-06-28 07:00 -999 -999 311.75 471.44 23.11 25.53 94 0.95 0 0.65
2019-06-28 08:00 -999 -999 600.4 687.22 29.3 32.87 94 0.95 0 0.65
2019-06-28 09:00 -999 -999 738.71 777.8 36.08 39.06 94 0.95 0 0.66
2019-06-28 10:00 -999 -999 795.02 801.19 41.73 44.19 94 0.94 0 0.66
2019-06-28 11:00 -999 -999 784.4 758.41 46.59 48.71 93 0.93 0 0.66
2019-06-28 12:00 -999 -999 711.83 661.7 50.3 51.69 90 0.87 0 0.65
2019-06-28 13:00 -999 -999 586.53 500.52 52.22 51.95 78 0.66 0 0.56
2019-06-28 14:00 -999 -999 410.94 347.2 50.95 50.33 34 0 0 0.43
2019-06-28 15:00 -999 -999 344.43 338.76 49.96 49.44 0 0 0 0.43
2019-06-28 16:00 -999 -999 327.23 313.86 8.73 47.84 0 0 0 0.43
....
API Data Package: PVmulti
Specifications can be provided on request.
General API Configuration
This is an overview of the API-URL. Blue and bold means that this is fix and can't be changed by the customer. Italic and light blue means that can be changed by the customer. This example API-URL is an invalid demo URL. Description:
- “http://my.meteoblue.com/packages/solar-1h?” address and packages
- “&lat=47.5584” coordinates
- “&lon=7.57327” coordinates
- “&asl=279” altitude
- “&apikey=personalAPIkey” Personal APIkey
Optional settings
- “&tz=Europe%2FZurich” time zone
- “&name=Basel” location name, label your forecast, has no effect on data “&timeformat=Y-M-D” Time format
- “&format=json” Output format
For correct daily aggregations it is best to omit the “&tz=” parameter, as then everything will be in local time including daylight saving (however over on the ocean everything will be in UTC if you omit the tz).
Geographic Coordinates
Position is defined by the geographic location (coordinate) and the elevation (altitude) of the point considered, be on the surface or in the atmosphere. Coordinates are latitude (North, South) and longitude (East, West) given in degrees and decimals. Formats are:
- Latitude: from -90.0000° (South) to 90.0000° (North) - Apply: &lat=...
- Longitude: from -180.0000° (West) to 180.0000° (East) - Apply: &lon=... A wrong configuration of the position of your system will lead to a substantial loss of accuracy. To verify the position of your site you can use the <What's here> function of Google maps as shown in Figure below:
The correct decimal coordinates of your site can be detected using the <What's here> function in Google maps. The coordinates are displayed in grey below the address.
Time Zone (tz)
The time zone is used to provide data in local time. For autonomous systems we recommend to use UTC. Daylight saving time might otherwise cause problems as in a data shift of one hour. For user interfaces data in local time is desired. You can provide “tz=Europe%2FZurich” to get data in CET or CEST timezone. “%2F” is an URL encoded slash “/”. If the time zone is not provided to an API request, a time-zone database is used to get the time zone. For coastal areas UTC might be selected incorrectly. We recommend to provide the time zone, if available. For fixed time offsets use for example GTM+2 for -02:00 UTC offset or GTM+ 02:30 for -02:30 offset. A complete list of possible time zones can be found at Wikipedia.
Time Format
You can choose between the following time formats: YYYY-MM-DD hh:mm (default), YYYYMMDD hh:mm, UTC timestamp (seconds or milliseconds), ISO 8601
Examples:
- YYYY-MM-DD hh:mm: 2016-02-10 03:00
- YYYYMMDD hh:mm: 20160210 03:00
- Time stamp_utc: 1455069600
- Time stamp_ms_utc: 1455069600000
- iso8601: 2016-02-10T03:00+01:00
note
Time stamps are always returned in UTC time zone per definition. To get local time you have to apply the time zone-offset manually.
Output Format (Format)
Currently only json and csv are supported. Additional formats might follow as per customer requirements. csv output format only supports a single time-resolution. In order to request “daily” and “3-hourly” data, you would have to make two API requests.
PVpro: Characterization of Your PVforecast
To characterize your PV system within the solar forecast API you need to verify the following parameters, which are specified by keywords (kwp, slope, facing). Example for a PV system with 125 kWp facing 27° towards South: &kWp=125&facing=180&slope=27
Capacity (kWp)
The peak value of your system referring to the maximum AC output. The implemented kWp (kiloWatt peak) value is used as upper limit of your forecast.
Apply: &kWp=...
Exposition
The simulation (red&blue) underestimates the midday peak of the actual PV-production (black), because the configuration of the slope angle is too low.
The midday peak of the simulation (red&blue) is late compared to the actual PV- production (black), because the configuration of the orientation angle is too high.
For the projection on tilted surfaces the exposition of the plane and the position of the sun need to be defined by two angles:
- Slope angle / Sun height (0°=horizontal - 90°=vertical) - Apply: &slope=...
- Orientation angle / Solar azimuth (0°=facing north, 90°=E, 180°=S, 270°=W) - Apply: &facing=...
Angles of incidence are measured relative to the Earth's surface. Aslope Angle of 0° is parallel to the surface, 90° is perpendicular to the surface. Orientation Angles are measured from facing North (0°) clockwise. Angles measurements are used for position of the sun, inclination of a solar collector (Photovoltaic system), or a measurement device. The wrong definition of the exposition will lead to a systematically wrong daily curve as shown in Figure a) & Figure b)
Keyword: snow
Since 2015 meteoblue offers the option to detect snow and automatically reduce the resulting forecast of PV-pp. It is assumed, that a snow cover of 40mm or more results is completely intransparent for any light. Malfunction of the meltdown function can lead to substantial errors especially in mountainous regions. Default: disabled Apply: &snow=1
Keyword: power_efficiency
As the efficiency of a specific PV systems is depending on many factors like layout, age, shading etc. we offer its manual implementation. The power efficiency factor applied is a mean value and will still be varied by module temperature, IAM and others to simulate the actual PR. All values are allowed but values above 1 are not logical, but might be useful for systems with capped production curves as the simulation will not exceed the peak capacity (see chapter Capacity (kWp)) Default: 0.85 Apply: &power:efficiency=...
Radiation Forecasting by Numerical Modelling
NEMS and NMM Modelling.
meteoblue domain coverage: NEMS (black), NMM (red). Global domain (NEMS30) not shown.
These models have been adapted to mesoscale, and calibrated for various variables since 2004. The process is described in several publications (see Janjic 2003). meteoblue has put these models in large- scale operations since 2007. The NEMS model is being calculated globally since 2013, as well as in high- resolution domains (3-12 km). meteoblue is also the first worldwide provider to daily produce hourly weather forecasts in spatial resolution of less than 10 km for entire South America, India, East Asia and Central America. Model calculations are done for "domains" - large areas covering parts of or the entire continents, for which a complete forecast is calculated. Domains are embedded in larger scale Global models, which provide the "boundary" conditions (e.g. the air currents coming into or moving out of the domain area) for the days of the forecast. Since 2013, meteoblue computes its own global weather forecast and thus can generate the boundary conditions on its own. Each domain is divided into grid cells, which are rectangularly arranged and evenly spaced between each other. The average distance between the grid centers (grid points) is the "spatial resolution", which varies from 25 km down to 1 km. Domains typically have 55 atmospheric levels, which range from the surface to approximately 14 kilometers height. Each grid cell has its own position, altitude, exposition, land surface type and boundary conditions. The altitude is defined using the median altitude of the grid cell area, as taken from high-resolution land surface models (~100 m). This approach allows calculation of the weather specifically for each area, with uniform, high quality from a city centre over airports to distant mountains areas, and provides an unique weather forecast quality and consistency. NMM technology has been developed and used for weather forecasting in North America by NOAA and for hurricane prediction. More recent developments include the NEMS framework (NOAA Environmental Modeling System) which has been adapted and further developed to allow global and regional high resolution modeling, seamless nesting for regional domains and improved cloud and precipitation schemes. meteoblue does not use WRF and other open source models, as they produce less accurate results. Main improvements of NMM and NEMS models relative to WRF are cloud and precipitation schemes, radiation simulation, domain nesting and global scalability and many adaptations to high resolution.
Other Weather Models Used
Beside the 15 NMM and NEMS model domains that meteoblue runs in house, 18 third party models are accessed. To offer the best possible solar forecast within our API a weighed multi model average is applied. An overview of all models used is given in Figure below.
Overview of numerical weather model domains operated by meteoblue (NEMS & NMM) and third party domains available.
Solar Service Quality Standards
Hourly mean absolute errors (rMAE) of meteoblue day-ahead (MOS, NEMS30 & multimodel) and intraday (SAT-OBS & hour-ahead) radiation data averaged for all stations in a specific region.
Solar forecasting services have been developed and tested for quality on many sites over 8 years and 4 continents. All data and products are constantly validated on measurement station and satellite data worldwide. The achievable accuracy, depends on the climatic region and the forecast horizon. An overview is shown in Figure 5.3. More information is provided in the document: >>meteoblue_Solar_Controlled_Quality_EN.pdf<<
Statistical Post Processing
Description
MOS (Model output statistics) is a localized power forecast based on the radiation forecast of multiple weather models and historical power measurements. Thereby, the model output is corrected statistically by the function developed from the measurement data (Model Output Statistics), using the multi-model radiation as a basis. The forecast is adjusted to the local station, using quality controlled measurements from the site, available for minimum 1 year, in hourly intervals. https://content.meteoblue.com/research-development/data-sources/nmm-modelling/mos
General Process
Radiation and power forecasts can be optimized by using statistical post-processing algorithms. The improvement of the forecast is depending on several site specific conditions. In average an improvement of 10% (range of MAE: 5%-25%) is realistic. To assure the functionality of the post-processing algorithms a Quality-Control report of the measurement data is obligatory. Furthermore it is recommended to purchase a Validation report which includes the quality of standard and optimized forecast and a product recommendation. meteoblue recommends the following procedure:
- Delivery of measurement data in standard format
- Preparation of QC-Report
- Preparation of Validation-Report
- Preparation of API due to product recommendation
Methodology
meteoblue algorithms use a conditional MOS, which combines a simple neural network with improved meteoblue MOS technology. The algorithms differentiate between multiple sky conditions to find the best fit of linear regression for more than 30 weather variables from multiple weather models. The resulting forecast algorithm is individually designed for the specific site and provides the best possible forecast (state of the art) via a special API set for the specific application and location.
MOS Forecast Variables
The possible forecast variables and range/intervals are:
- V_MOS: Energy_kWh;
(Availability variable is necessary to calculate kWh/kWp) - GHI_MOS: GHI
(ExRad for quality control is processed afterwards)
Other variables can be supplied on request.
Configuration Process
The MOS power forecasting model is determined based on a radiation forecast. This radiation forecast is adjusted to the local power plant by a statistical function, derived from past measurements, as follows:
- Customer provides location information: position, altitude (exposition).
- Customer provides historic radiation and/or production data: radiation and/or electricity (power) output for at least 1 hour, in 10 -60 minute intervals: as .csv or .txt. files. Data should be provided with comma separator “;” (not “,”) to avoid formatting mistakes.
- meteoblue conducts data quality control, to ensure a good return on investment .
- meteoblue extracts radiation archive data and matches to radiation / production.
- meteoblue implements correction factors in system. meteoblue provides corrected MOS power forecast with 24 updates per day.
Observation Data Format
It is recommended to provide at least one year of data in hourly time resolution to train the MOS. The data should be provided in a standard format using comma separated files (.csv). Especially designed routines will be applied to detect completeness, time-shift and plausibility of the time series. Apart from radiation and/or production time series, the minimal metadata (header information) have to be supplied. The observation data for MOS configuration should support following requirements:
- standard text format: .csv oder .txt
- one file for each site, that contains full available timerange
- FILENAME: Site-ID_LAT_LON_timerange_type_variable_aggregation_generationdate.csv
- e.g.: MB-4412_42.75_12.86_20130523-20141218_OBS_PV_HOUR_20141218z1200.csv
- The file contains SITE_ID, date and time (in UTC) and the target variables. PVpp is always delivered with system availability (=available capacity (kWp) at a given time stamp).
Sample:
ID_MB;date;UTC;Energy_kWh;kWp
1557_65_10;23.05.2013;12:00:00;0.12;87.98
1557_65_10;23.05.2013;13:00:00;0.63;87.98
1557_65_10;23.05.2013;14:00:00;0.04;87.98
1557_65_10;23.05.2013;15:00:00;4.92;87.98
1557_65_10;23.05.2013;16:00:00;11.63;87.98
Accuracy Improvements
The expected precision of radiation or power forecast with the MOS forecast is between 5% and 25% rMAE on an hourly basis for 2-24 hour forecast, depending on location and year. The value of statistical post processing differs depending on location and input data (measurements). To assess the actual benefit of statistical post processing, we recommend our Validation-Report. The highest value of these methods can be observed for the exact configuration of PV yield forecast. Differences in site results are shown in figure below.
Hourly mean absolute errors (rMAE) of meteoblue day-ahead (multimodel & PV-mos) and intraday (3h-nowcast) PV yield forecast validated on 38 PV systems in Central Europe.
Nowcasting
Description
Nowcasting is the combination of localized power forecast based on the radiation forecast of different weather models with real time satellite observations collected every 15 minutes. Thereby, the model radiation output of the present day is corrected in real time with the satellite derived radiation. Based on the actual cloud pattern and wind vectors a so called cloud motion forecast is processed which optimizes the forecast accuracy of the next 5 hours. Depending on the actual weather situation persistence, cloud motion and the model output are weighted differently for the time horizons of the next hours. The combination of these techniques is summarized with the term: nowcasting.
The solar forecast retrieved before sunrise (blue), in the morning (red) at noon (green) and after sunset (violet) of the same day.
The instant of time that are already past contain the actual satellite observation. Thus the solar forecast API contains only observation values after sunset and is therefore combines forecast and monitoring within the same data feed. The effects of nowcasting are depending on the accuracy of the weather model. If the day ahead forecast is correct, nowcasting has almost no impact (see Figure above), otherwise the nowcast is strongly changing the content of the API (see Figure below).
The solar forecast retrieved before sunrise (blue), in the morning (red) at noon (green) and after sunset (violet) of the same day.
Nowcasting Availability
Presently the nowcasting technology is automatically implemented in the API for all locations within sight of the Meteosat Second Generation (MSG). Thus it covers Europe, the Arabian peninsula, Africa, Brazil, and parts of its neighbouring countries (see Figure below).
The full MSG disc of Global Horizontal Radiation (GHI) at 2016-03-16 15:00 UTC (Source: KNMI).
Nowcasting Variables
The nowcasting is effected on all radiation variables and PV power simulations.
Used Variables
The used observation variables, accessed on from satellite observations are:
- Global Horizontal Irradiation (GHI)
- Clear Sky Global Horizontal Irradiation (GHI_CS)
- Cloud top-height (CTH) The used simulation variables, from NEMS are:
- Global Horizontal Irradiation (GHI)
- Clear Sky Global Horizontal Irradiation (GHI_CS) Wind vectors u & v on different levels
Cloud Motion Vectors (CMV)
Based on the image of the cloud-top level the wind vectors (see Figure below) of the next 6 hours is extracted for the specific cloud height of each point. A cloud motion vector (wind trajectory) is calculated for each time horizon of 1-6 hours.
Solar Radiation Nowcast Based on Cloud Motion Vectors(CMV)
Based on the six wind trajectories, the referring pixel layer is chosen for each time horizon (1-6 hours). This pixel is used to calculate the expected radiation. The result is a radiation forecast for the next 6 hours. Cloud motion nowcasting works only after sunrise, when the referring satellite data is available.
Wind speed and trajectories at 600 hPa level on 2016-03-16 15:00 UTC.
Persistence Solar Radiation Nowcast
For the persistence solar radiation nowcast the radiation of the last 4 hours is used (without any cloud motion) and weighted.
Combined Solar Radiation Nowcast
For different time horizons and weather situation the weighing of persistence, cloud motion and numerical weather prediction is changing. It is useful to combine the different approaches to minimize errors. Weather variability could be classified with the clearness index of the surrounding satellite pixels.
Expected Precision
The expected precision of radiation or power forecast with nowcasting is between 5 and 20% NMAE on an hourly basis for 0-5 hour forecast, depending on location and satellite availability. The accuracy of nowcast is declining with forecast horizon (see Figure 7.5). After 5 hours the multi-model is performing better than the nowcast.
Nowcasting (red: 1h, blue:2h, orange:3h) improves performance of solar forecast service (black line). Vertical axis shows MAE (in %) of hourly forecast over a full year. Stations sorted in order of ascending error.
Appendix I: Definitions
Radiation
Solar radiation in this document is defined as the amount of energy reaching the Earth surface, and measured in Watt per square meter (W/m$^2$). It contains several wavelengths and it reaches the Top of the Atmosphere (ToA) as “Extra-Terrestrial” radiation (ExRAD) and the ground as Global Horizontal Irradiation (GHI). It can be divided into Direct (Beam) and Diffuse radiation (DIF). Beside this radiation measures on horizontal surfaces, meteoblue offers variables on inclined or normal (to sun angle) surfaces like Global Tilted Irradiation (GTI), Global Normal Irrradiation (GNI) and Direct Normal Irradiation (DNI). Maximum radiation intensity reaching the Earth atmosphere is the extraterrestrial Solar constant of 1368 W/m$^2$ (1321 to 1413 W/m$^2$ depending on distance of the Earth to the sun).
Parameterisation
Solar radiation is measured on horizontal surfaces by the following variables.
- GHI = Global Horizontal Irradiance (total short-wave radiation energy received).
- DIF = Diffuse Horizontal Irradiance (model acc. to Reindl).
- IR = Direct Irradiance = GHI-DIF
Projections on normal surfaces (perpendicular to the sun) are defined as - GNI = Global Normal Irradiation: Global irradiance on normal surfaces (tracking the sun);
- DNI = Direct Normal Irradiance: Direct irradiance on normal surfaces (tracking the sun), of interest mainly for solar thermal and concentrated Photovoltaic power generation. Therefore, DNI can be (much) larger than GHI, if the Sun height (at a particular time) is low (see 2.1.), because the same area of horizontal surface receives only a fraction of the sunlight, compared to a normal surface.
Radiation Measurements
Solar radiation is measured using pyranometers (see Figure below), except DNI which is measured with a pyrheliometer. Measurements are given either in W/m$^2$ for a particular time, or as the average W/m$^2$ for the preceding time interval. Measurement is a very sensitive process, since the intensity is affected by several processes:
- Solar angle: energy incidence changes with the height (“angle”) and direction (“azimuth”) of the sun (see 8.5. and 8.19.).
- Inclination of measurement device towards the sun: typically, measurements are made in horizontal position. “Normal” measurements are always oriented towards the sun.
- Time interval: radiation can change rapidly within seconds, so the interval and integration of individual measurements has an influence on the result;
- Air pollution: turbidity and aerosols influence the amount of energy reaching the surface;
- Instrument pollution: Particle and aerosol deposits may reduce instrument sensitivity;
- Instrument age: deterioration of instrument components may change sensitivity.
In summary, radiation measurements are amongst the most difficult to calibrate amongst all meteorological variables and the quality of measurement data usually varies substantially.
Schematic Design of Pyranometer (QUASCHNING 2009 in Bührer, 2010).
Power
Power in this document is defined as the electricity output of a given device, at a specified point of measure (PoM) in kilo-Watt hours (kWh).
Solar Exposition Angles
For the projection on tilted surfaces the exposition of the plane and the position of the sun need to be defined by two angles:
- Slope angle / Sun height (0°=horizontal - 90°=vertical)
- Orientation angle / Solar azimuth (0°=facing north, 90°=E, 180°=S, 270°=W)
Angles of incidence are measured relative to the Earth's surface. A slope Angle of 0° is parallel to the surface, 90° is perpendicular to the surface. Orientation Angles are measured from facing North (0°) clockwise. Angles measurements are used for position of the sun, inclination of a solar collector (Photovoltaic system), or a measurement device.
$ SP = {\eta_{prakt}\over \eta_{MMP}} $
Performance Ratio(PR)
The performance ratio describes the efficiency of a power plant. It is calculated with the quotient of the effective efficiency and the theoretical efficiency of a power plant: The performance ratio varies significantly within different weather conditions. meteoblue solar forecast simulates the PR dynamically for crystalline modules including:
- Module Temperature based on ambient temperature, wind speed and radiation
- Spectral sensitivity
- Reflection losses on the module surface
The sensitivity of non-crystalline modules to spectral composition and module temperature can differ from the applied algorithms. To implement their behaviour in your forecast meteoblue recommends statistical optimization (MOS). The default mean PR is 85 % but can also be adjusted using a specific keyword.
Horizon
Often the horizon is not flat but elevated because ground surface is complex or obstacles are in the sight. This can affect the sunshine intensity and thus the typical daily curves. The exact definition of the horizon is used to optimize simulation accuracy. The horizon is defined with 12 numbers that turn clockwise (compare orientation angle) from North 0° and its value defines the horizon height in degrees (from 0°- 60°):
- horizon= [N,NNE,NEE,E,ESE,SSE,S,SSW,SWW,W,WNW,NNW)
- e.g. horizon=[0,0,0,0,20,30,40,30,20,0,0,0]
Snow Cover
Losses due to snow cover can be implemented in the forecast with a specific keyword. The PVpro feed contains snow cover and therefore allows the customer to adjust the forecast himself. Snow cover can be erroneous in mountainous regions.
Position
Position is defined by the geographic location (coordinate) and the elevation (altitude) of the point considered, be on the surface or in the atmosphere. Coordinates are latitude (North, South) and longitude (East, West) given in degrees and decimals. Formats are:
- Latitude: from -90.0000° (South) to 90.0000° (North);
- Longitude: from -180.0000° (West) to 180.0000° (East); Altitude is defined in meters above sea level (m.asl). More information is available under: https://content.meteoblue.com/en/help/standards/position
Data API
Data API is an Application Programming Interface, which enables access to a defined datafeed in a regular automated way.
Datafeed or Package
Datafeed or package is a determined set of variables available through an API in a regular automated format.
Variables
Variable is a measurable factor that helps in defining a particular system. It is standardised and varies over time, during the course of observation . Radiation, cloud cover, humidity, wind, and others are weather related variables to define relevant physical information.
Time Stamps
Accurate time stamps are essential for correct measurement and forecast of radiation. Variable values are usually displayed in UTC (Coordinated Universal Time). Local time is used if required by specification, and if precise conversion tables from local into UTC are available. Information on conversion of UTC in local time can be found under www.timeanddate.com//worldclock/. Local forecasts are adjusted to the local time zone using the selected place or location. In some cases, local timezone management and switches of standard to summer time may lead to inaccuracies of +/-1 hour (see also timezone explanation under http://en.wikipedia.org/wiki/Timezone).
Time Stamps: Important Notice
Radiation observations might be instantaneous values (inaccurate) or averages over a certain time period (interval). If the observations are averages, it is very important to know if the 10:00 value represents a previous average from 9:00 to 10:00, a centred average (from 9:30 – 10:30) or a forward average (from 10:00-11:00). Also note that depending on the datafeed type, the time might be in UTC or in local time. The feed header will indicate the time offset to UTC.
Time Averaging
There is two common methods of time averaging:
- Backward Averaging: time stamp represents preceding time range In hourly resolution 12:00 represents 11:00-11:59. The standard time convention of meteoblue archive data is: Hourly Backward Averages in UTC
- Instantaneous Averaging: timestamp represents time range before and after the time stamp; values are better to be compared with instantaneous observations then Backward Averages. In hourly resolution 12:00 represents 11:30-12:29. The standard time convention of meteoblue radiation datafeed provides both conventions so the data can be compared either with the archived data or instant measurement values.
meteoblue Datafeeds: Time Stamps
All solar radiation and power forecasts are average values centred on the indicated time. Thus: the 11:00 value represents the mean value observed between 10:30 and 11:30 if the time resolution of the feed is one hour (dt=1), or the mean value between 10:55 and 11:05 if the time resolution is 10-minutes (dt=0.166666). This centred averages allow to directly compare meteoblue forecast values with instantaneous values valid at the time indicated. Within hourly datafeeds, backward averages are available also. Note, that for dt=0.5 or smaller, the forecast time steps between the full hour will be interpolated, and therefore have no higher accuracy than the original hourly resolution.
meteoblue Archives: Time Stamps
The meteoblue radiation archive data are accumulated values of the previous hour. Thus: the 11:00 value represents the average value observed between 10:00 and 11:00. This form is used to make radiation archives consistent with the principle that measurements can be taken only after the conclusion (end) of the measurement period. All archive time stamps refer to UTC.
Customer Observation Sata: Time Stamps
For observation data provided by our customer the same time convention than for meteoblue archives are valid, as statistical post processing routines are trained on our archive data. If possible all observation data should be provided in as hourly backward averages with UTC time stamp. Inconsistent time stamps and temporal offsets will be detected by our QC routines.
Solar Position Definitions
Solar position changes as a function of the Earth rotation and orbit around the sun, relative to the Earth surface and (measurement) modules, and is described by the following variables (see Figure below):
Symbol | Description | Unit |
---|---|---|
rlt | real local time. | hh:mm:ss |
UTC | Universal time convention | hh:mm:ss |
$α_E$ | Azimuth of (measurement) module. | ° |
$α_S$ | Solar azimuth | ° |
$δ_S$ | Solar declination | ° (Geographical Coordinate) |
$Φ_T$ | Inclination of the Earth axis (relative to the sun) | ° |
$Θ_{gen}$ | Incidence angle on (measurement) module | ° |
n | Normal vector of (measurement) module. | n.a. |
$\gamma_E$ | Inclination of (measurement) module. | ° |
$\gamma_{S(t)}$ | Sun height (at a particular time) | ° |
$\gamma_{Z(t)}$ | Zenith angle of the Sun (90°-$\gamma_{S(t)}$) | ° |
Definitions for solar position (QUASCHNING 2009 in Bührer, 2010).
Sun Height or Solar Angle
The sun height or Solar angle $\gamma_{S(t)}$ is defined as the angle between the horizon and the sun position at a specific point of time. Examples:
- At sunrise, the Sun height is 0° on a flat surface (like the sea). The sun position is exactly parallel to the Earth Surface.
- At equinox noontime, the Sun height is 90°. The sun position is exactly vertical to the Earth Surface.
Sun Zenith Angle
Zenith angle of the Sun $\gamma_{Z(t)}$ is defined as (90°-$\gamma_{S(t)}$) at perfect noontime (in “real local time”). Examples:
- At sunrise on the sea, the zenith angle is 90-0 = 90°. The sun position is parallel to the Earth Surface.
- At equinox noontime, the zenith angle is 90-90 = 0°. The sun position is vertical to the Earth Surface.
Sun Azimuth
meteoblue defines Solar clockwise from North. Thus North is 0°, East 90°, South 180° and West 270°. Examples:
- At astronomical (perfect) noontime, azimuth is always 180° North of the tropic of Cancer.
- At astronomical (perfect) noontime, azimuth is always 0° South of the tropic of Capricorn.
- On the equator, sunrise during equinox occurs with azimuth of 90° and sunset with 270°. To calculate the local sun azimuth in UTC, the correction from “real local time” to UTC has to be made. Some systems use South as the reference (0°), with Eastern azimuths being positive (1 to 180°) and Western azimuths being negative (-1 to -180°). Also used is a convention that defines the azimuth as the difference from the noontime azimuth (0°). If such a system is used, the degrees provided have to be converted accordingly.
Solar Declination
The Solar declination is the geographical latitude, in which the sun reaches the 90° Zenit on a given day. Solar declination can therefore vary between 23°(North) and -23° (South). At the beginning of spring and fall (Solar equinox), the sun declination is zero (latitude 0° = equator), since the sun is placed exactly over the equator (STULL 2000) and changes as a function of the Earth rotation and transladation around the sun.
Formula for calculation of solar declination:
$\delta=\varPhi_r.cos[\frac{360^o.(d_y -d_r)}{365}]$
Usually, Solar declination is used for calculations of the sun position during the year, to describe the changes in highest angle between days.
Appendix II: Sources
A publication list is shown in Table below
Title$^1)$ | Autor$^2)$ | Year$^3)$ | Publisher$^4)$ |
---|---|---|---|
Solarstrom – Vorhersage und Markteinbindung | Michael Bührer | 2010 | Masterarbeit, 69 pp. Universität Basel |
A new simplified Version of the Perez diffuse irradiance model for tilted surfaces | Perez, R. et al. | 1987 | Solar Energy, Vol. 39, 1987; p. 221-231 |
Modeling daylight availability and irradiance components from direct and global irradiance | Perez, R. et al. | 1990 | Solar Energy, Vol. 44, 1990; p. 271-289 |
Regenerative Energiesysteme: Technologie - Berechnung - Simulation | Quaschning, V. | 2009 | Hanser, München |
Diffuse fraction correlations. | Reindl D.T: et al. | 1990 | Solar Energy, Vol. 45, 1990; p. 1-7. |
Five satellite products deriving beam and global irradiance validation on data from 23 ground stations. | Ineichen P, | 2011 | University of Geneva, 40 pp. |
Notes: $^1)$ As published $^2)$ Full name of first author. $^3)$ Publication. $^4)$ Publisher: Company or Institution.