Measurement and Modeling of Solar Radiation

October 30, 2017 | Author: Anonymous | Category: N/A
Share Embed

Short Description

POA irradiance is from three sources: DNI, DHI and solar An isotropic model assumes all diffuse ......


Measurement and Modeling of Solar Radiation

Dr. Manajit Sengupta April 2016

NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, operated by the Alliance for Sustainable Energy, LLC.

Why Do We Need Solar Radiation Data? • • • • • • • • • •

Agriculture Astronomy Atmospheric Science Climate Change Health Hydrology Materials Oceanography Photobiology Renewable Energy

Photosynthesis Solar Output Variation Numerical Weather Prediction Energy Balance UV effects on skin Evaporation Degradation Energy Balance Light and Life Sustainability


What Do We Want to Do with the Sun? • Harvest the energy to support our lives

Lower our cost of living (and maintain our standard of living?) o Reduce reliance on foreign energy supplies o Reduce harm to the environment o

• Transport the energy from one place to another o

Produce in low population density; use in high population density

• Store the energy for use at a later time o o

Use at night Optimize response to demand


Power and Energy Understanding Solar Resources Compare the Energy Density* of Solar Energy to Fossil Fuels In one average day at NREL, an area of 236 m2 receives 1.7MWh of solar energy 236 m2 ~13-15 parking spaces = 1 BOE/day

Credit: DOE

*1 Barrel of Oil = 1.7MWh

When considering conversion efficiencies (15-20%), this NREL PVcovered parking structure may supply the energy equivalent of a barrel of oil per day.


Why Solar Energy Resource Assessment Accurate resource data reduce risks for each project phase:


Policy Decisions Site / Technology Selection Increasing


Investor Commitment Project Approvals Spatial & Temporal

Due Diligence

Engineering Design System Integration Resolution


System Tests Operation & Maintenance Energy System Integration 5

The Best Practices Handbook

Chapter 3

Measuring Solar Radiation

Chapter 4

Modeling Solar Radiation



Most common solar parameters • • • • •

Global Horizontal Irradiance (GHI) Direct Normal Irradiance (DNI) Diffuse Horizontal Irradiance (DHI) Plane of Array (POA) Spectral

GHI = DNI * cos(Z) + DHI

• Z is the angle between the zenith and the sun • Z is 0° at the zenith and 90° at sunrise or sunset


Units of Solar Radiation • Watts per square meter (W/m2) • Watt hours per square meter (Wh/m2) Joules per square meter o BTU per square foot o Langleys o Calories per square centimeter o


Plane of Array Radiation  POA irradiance is from three sources: DNI, DHI and solar radiation reflected by land surface.  An isotropic model assumes all diffuse radiation is uniformly distributed over the sky.  Anisotropic models such as the Perez model accounts for the non-uniformity of diffuse light.



Spectral Distribution of Solar Energy

• • •

Source: Philip Ronan

Basic Solar Spectral Regions (nominally 280-3000 nm): Ultraviolet…..200 - 400 nm Visible……....400 - 700 nm Infrared……..700 - 3000 nm


What Influences the Amount of Solar Radiation? • • • • • • • • • •

Solar output Earth-Sun distance Clouds Water vapor Air pollution Smoke from forest fires Volcanic ash Location Time of day Season

11 year solar cycle 3.5% annual variation Dominant factor Selective absorber 40% less direct Natural or man-made Global effect for years Solar position


Change with Time and Location


Changes with Time


Changes with Time - Interannual MONTHLY MEAN DAILY TOTALS Solar Radiation Research Laboratory 1986-2000




Direct Trend

y = 16.103x - 26829 2

R = 0.0061





Global Trend y = 4.3303x - 4215.9 R2 = 0.0034

2000 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001


What Influences the Amount of Solar Radiation?

Earth’s Orbit: • Earth-Sun distance (minimal) • Relative axis tilt (significant) • Time of day (largest)


What Influences the Amount of Solar Radiation? Forest Fires can have a significant effect for days or weeks for hundreds of kilometers

• •

Direct down 22% Diffuse up by 100%


The Solar “Constant”

World Radiation Center, Davos, Switzerland 17

How Accurate Do the Data Need to Be? • What are the Cost/Benefit and risks? o o

Illuminated through Resource Assessment Higher value project (risk?) may demand higher accuracy

• What is the application? o o o o o

Residential (solar water preheat) Commercial (daylighting & building thermal performance) Industrial (concentrating collector solar power plant) Global integrated assessment model (IAM’s) Grid integration studies.

• What is the period of interest?

Instantaneous, daily, seasonal, annual? Longer averaging intervals can remove random errors o Recent data more accurate than historical records (climate trends and technology advancements) o


How do we measure/model solar radiation? •

Ground Measurements:

Pyranometers, Pyrheliometers, etc.

– Advantages: accurate, high temporal resolution. – Disadvantages: local coverage, regular maintenance and calibration.

Satellite derived estimates:

Physical (e.g. NREL PSM), empirical (e.g. SUNY), etc.

– Advantages: global coverage, reasonably long time series, – Disadvantages: spatial and temporal resolution, complicated retrieval process, accuracy depends on information content of satellite channels.

Numerical Weather Prediction based estimates


– Advantages: global coverage, long time series (reanalysis data), increasing computing capability results in increasing resolution. – Disadvantages: level of accuracy especially in cloud formation and dissipation (initialization and model physics issues).

NOTE: Methods that combine all 3 will ultimately provide the best solutions. 19

Measuring and Using Solar Data

Topics • • • • •

Measuring and Using Solar Data Calibration Uncertainty and Data Quality Station Maintenance Measurement Station Siting and Deployment


Measurement Options First establish project accuracy requirements.

• Start from the end—what will the analysts need for the best analysis? • This allows you to base instrument selection and the levels of effort for operating and maintaining the system on an overall cost-performance determination.

Anticipate your future requirements

Always consider financial resources available for all operations, including maintenance.

• “First-class” instrumentation should not be used if the project resources cannot support the maintenance required to ensure high measurement quality. • Without regular maintenance, the best instruments may be worse than lesser instruments. This can result in poor data at high cost.


Measurement Options Three fundamental measurements • Global Horizontal Irradiance (GHI) • Direct Normal Irradiance (DNI) • Diffuse Horizontal Irradiance (DHI)

Measuring all 3 components provides measurement redundancy


Measurement Options Additional Measurements

• Global tilt – Fixed or tracking plane of array (POA) • Spectral, Infrared, Ultraviolet • Albedo (ratio of upwelling and downwelling globals) • Meteorological (temperature, wind, humidity, pressure, precipitation) • Sky imager


Measurement Technologies Thermopile • Based on the thermoelectric effect • Broad spectrum blackbody absorber • Voltage output • Susceptible to soiling (because of protective optics) • Higher cost

Silicon Photovoltaic • Based on the photoelectric effect • Narrow spectrum sensitivity • Current output (converts to voltage with a shunt resistor) • Less susceptible to soiling • Lower cost


Instrumentation Pyranometers Pyrheliometer (Eppley NIP) (requires separate tracker)


Offset compensation

Rotating Shadowband Radiometer

Measures GHI, DNH, Calculates DNI


Low cost silicon

Eppley 8-48

Campbell-Stokes Sunshine Recorder

Low cost; Often used for diffuse 26

How Do I Choose an instrument?


How Do I Choose an instrument? What are your requirements? • Measurement Uncertainty (How much accuracy does your application need?) • Measurement speed (What is my integration time; instantaneous?) • Spectral Response (Is broadband necessary, or will silicon suffice?) • Ease of Maintenance (Can I keep to the necessary maintenance schedule?) • Cost (What is reasonable?) • Calibration Interval (Will the instrument hold calibration; can I afford to have it out of service?) Habte, A.; Wilcox, S.; Stoffel, T. (2014). Evaluation of Radiometers Deployed at the National Renewable Energy Laboratory's Solar Radiation Research Laboratory. 187 pp.; NREL Report No. TP-5D0060896. 28


Calibrations “Calibration is a comparison of an output signal with a measurement standard traceable to a recognized standard to identify and eliminate deviations in accuracy.” • One instrument has known characteristics of uncertainty (the standard instrument) • The other has an unknown or lesser uncertainty (the unit under test). • The comparison yields calibration results, when applied to the unit under test, produces a corrected reading that agrees with the standard (with a stated uncertainty). • The standard will have known and traceable characteristics (uncertainty) relative to a consensus reference.


Calibrations The Solar Reference is linked to the International System of units (SI) through the World Radiometric Reference. The standard for solar measurements (DNI) is the World Radiometric Reference, derived from a consensus set of instruments in Davos, Switzerland at the World Radiation Center (the World Standard Group). • The World Standard Group has a long and wellcharacterized history, and is comprised of instruments from several manufacturers and technologies. Currently the WSG has six instruments. • The WRR is derived statistically from the World Standard Group based on individual instrument performance. • The estimated uncertainty of the WRR is about ±0.3% with respect to SI. • The standard for global and diffuse is calculated from the DNI

World Radiometric Reference in Davos, Switzerland


Calibrations • An International Pyrheliometer Comparison (IPC) is held in Davos every five years to transfer the WRR to participating countries. • NREL Pyrheliometer Comparison (NPC) is held in Golden, CO every year to transfer calibration to participants


Calibrations Traceability “property of a measurement result whereby the result can be related to a reference through a documented unbroken chain of calibrations, each contributing to the measurement uncertainty." International Standard Organization VIM, 3rd ed., definition 2.41, 2012


Calibrations: Solar Measurement Traceabilty


Calibrations: Reference Instruments Cavity Radiometers Are Usually Chosen for Reference Solar

Cavity Radiometers Being Compared for Traceability to the World Radiometric Reference

•Reference instruments with potentially lowest uncertainty. •Self calibrating, though must still be traceable to WRR. •Typically fair-weather instruments without a window (absolute), though can be operated in all conditions with a weather-proof window. •Expensive and delicate; require expert operation. •Require a separate controller and data acquisition 35

Calibrations • Pyrheliometers are compared directly with the DNI from the cavity to determine a calibration coefficient • A simple comparison between two instruments


Calibrations Component Sum Calibrations

We create global reference from DNI, zenith angle , and diffuse.


DNI Global Reference = DNI * cos(Z) + Diffuse

The global reference has greater uncertainty than the DNI.

Schematic of Component Sum Calibration 37

Calibrations Advantages of Component Sum Calibration

• The simple calculated global reference allows easy simultaneous calibration of large numbers of instruments. • Reference and test instruments are sampled simultaneously. • No “moving parts” (no need to shade the instrument) – Less labor intensive – All-day calibrations are more practical

• Instruments under test are in relative equilibrium (no thermal shock from constant shade/ unshade). 38

Calibrations NREL maintains ISO* 17025 certification for its calibration processes. • ISO is the world's largest developer and publisher of International Standards NREL is expected to – • Maintain high quality traceable results • Use peer reviewed quality/calibration procedure • Provide Nationally/Internationally accepted calibration • Undergo periodic audits • Use controlled process for continuous improvement and early detection of problems/solutions • Provide consistent reporting of calibration results and associated uncertainties *International Organization for Standardization


Calibrations How Often Should You Calibrate? • Interval can be based on manufacturers recommendations • Can be based on your own experience • NREL recommends annual calibrations for radiometers – Some radiometers age at more than 3% per year – Radiometers can malfunction without total failure (malfunction may not be obvious) – A more conservative approach minimizes the chance of error (especially for a campaign lasting only a year or two).


Uncertainty and Data Quality Assessment

Goal • Develop a consensus standard to estimate radiometric data uncertainty using the Guide to the expression of uncertainty in measurement (GUM) method.  At

present the tendency is to look at instrument datasheets and take the instrument calibration uncertainty as the measurement uncertainty.


Calibration and Measurement Uncertainty • NREL radiometer calibrations are done outdoors. Calibration certificate reports the calibration results under specific environmental conditions that are different from conditions in the field. • In field deployments, add uncertainties due to o o o o o o o o o

Environmental effects (temperature, wind, atmos. constituents) Calibration Spectral Response Zenith Angle Maintenance----Soiling (dust, rain, birds) Data logger uncertainty Temperature dependence Non-linearity Aging 43

Calibration and Measurement Uncertainty ASTM Standard (under development): Standard Guide for Evaluating Uncertainty in Calibration and Field Measurements of Broadband Irradiance with Pyranometers and Pyrheliometers Excel® spreadsheet- Radiometric Data Uncertainty Estimate Using GUM method The spreadsheet provides a comprehensive estimation of measurement uncertainty associated with measurands using GUM method.


Data Quality Assessment Data quality is a function of your knowledge of the measurement environment and infrastructure. 1. Data quality is fixed (unchangeable) at the time a measurement is taken. 2. No amount of “quality control” after the fact can improve the fundamental data quality. 3. Data sets without good documentation are of unknown quality.


Data Quality Assessment The Quality Assurance Cycle Data Acquisition

Quality Assessment

Analysis/ Feedback


Data Quality Assessment Quality Assessment is not Quality Control • Quality Assessment Requires Judgment and Analysis. This happens after the measurements.

• Quality Control is a Supervisory Process. This happens before and during the measurements.


Data Quality Assessment Expected Values in Data Quality Assessment 1. When performing our data quality assessment, we want to examine a measurement in the context of an expected value. 2. We want to reduce the range of expected values as much as possible.


Data Quality Assessment

SERI QC: An Empirical Approach Operates in K-space: Fraction of Possible Irradiance Variable



Global / (ETRN * cos (Z))


Direct / ETRN


Diffuse / (ETRN * cos(Z))

Kt = Kn + Kd ETRN = Extraterrestrial Radiation Normal to the Sun (DNI above the atmosphere) K space subscripts: t = total (global), n = normal (DNI), d = diffuse 49

Data Quality Assessment Top of Atmosphere Extraterrestrial (ETR) ETR Measurement

Kt= Ground/ETR Kt is usually between 0.0 and 1.0 But not always…

Kt can occasionally exceed ETR for short periods, usually due to cloud reflections.

Ground Measurement


Reducing the Area of Expected Values Two Dimension K-space Template All data in outer space would plot here


But never here 0.8



Atmospheric effects (attenuation) plot data here.



0.0 0.0








Reducing the Area of Expected Values Maximum Kt and Kn 1.0

Kn Max 0.8





Kt Max

0.0 0.0








Reducing the Area of Expected Values Example Data (High Quality)


Reducing the Area of Expected Values Example Data (Poor Quality)


Reducing the Area of Expected Values Parameterizing the Envelope with The Gompertz Function

Kn is a function of Kt such that: Kn = αß



Where α, ß, Ɣ, and δ are variable parameters


Reducing the Area of Expected Values Examples Using QCFIT Austin, Texas

U of T Network Sep 1998-2002 5-minute Resolution

Eugene, Oregon

U of O Network Jun 1999-2000 60-minute Resolution


Reducing the Area of Expected Values Three Component Data With three component data, the expected relationship between the three measurements is unambiguous.

KЄ = Kt - Kn - Kd Ideally KЄ = 0 • Any non-zero value of KЄ indicates some disagreement among instruments. • KЄ = 0 could occur with errors if errors cancel. • Errors cancelling uniformly throughout a day are not likely


Reducing the Area of Expected Values The two component analysis is useful for checking three component data.


Daily Quality Checks

•A full month of data and quality flags at a glance •Shows data for each of the three solar components •Errors become instantly evident. •You can correlate flags with irradiance values •You can view the three components in context with each other.

Days of the Month

SERI QC Cylinder Plots

Hours of the Day 59

Daily Quality Checks Tracker Failure June 6 and 7 Low Kn

High Kd

High SERI QC flags.


Feedback • Feedback to station operators is a critical component When operators know that data are being examined, they are more likely to do a good job. o Regular feedback keeps data quality in the forefront. o

• Feedback is not just problem reports o o

Feedback should include positive reports of a job well-done With a well-run network, problem reports (negative feedback) should be a small exception. Data Acquisition

Quality Assessment

Analysis/ Feedback


Station Maintenance

Data Quality: Routine Maintenance The Rationale for Maintenance • A manufacturer’s instrument uncertainty is usually stated with the assumption of well-maintained environmental conditions (e.g. kept clean, aligned, calibrated) • Uncertainty will increase as the environment degrades measurement conditions (e.g. soiling, temperature extremes, equipment failures). • Compromised measurements can occur within minutes and may not be easily detectable (may “look” normal, but aren’t). These marginal effects can persist for days, months, or years if not detected.

Frost crystals on LI-COR pyranometer 63

Data Quality: Routine Maintenance How often is maintenance required to maintain the best quality data?

• Daily to every other day for domed and windowed instruments (which tend to soil faster) • Weekly to every other week for diffuser instruments (LI-COR, RSR), which tend to soil slower. • More often after unusual weather events (high winds, rain, snow, frost, etc.) • Some manufacturers recommend weekly cleaning for windowed instruments. Some don’t maintain diffuser instruments at all. • Our recommendations are conservative. Although instruments may remain clean between daily cleanings, the more frequent protocol is designed to minimize corruption of the data. You want to be able to defend data quality. • With experience, some locations may suffice with less frequent cleanings (but some additional uncertainty should be added). Conversely, some may require more frequent cleaning. 64

Data Quality: Routine Maintenance Cleaning Issues GHI Increases 1.5%

DNI Increases 5%

Typical cleaning event after ten days (Windowed instruments)

A knowledgeable critic will assume that these conditions exist in your data set unless you can show otherwise with a well documented maintenance protocol. Insect webs on instrument (no cleaning schedule)

Both DNI and GHI Increase by 3.5%

Atypical cleaning event after five days (RSR)

• Avoid doubt • Avoid hard questions • Avoid having to make a lot of excuses or explanations. Design a robust maintenance protocol and keep good documentation.

Difficult to clean insect droppings (a small, but significant effect on measurement)


Data Quality: Routine Maintenance • As a rule, it is better to not seek volunteers. • Plan for maintenance costs in future budget. • Follow-up training for maintenance people should be provided as necessary. • Training should include reality that mistakes and data problems occur. • The trainer should provide contact information (telephone and email) and encourage good communications.


Maintenance Documentation At a Minimum, Documentation Should Include • • •

• •

Date and time of station visit Name of person doing maintenance A checklist of EACH instrument and associated support equipment (e.g. not only anemometer, but tower and guy wires) Keys to any codes (e.g. clouds) A place for free-hand comments (unusual circumstances not encountered in checklist). Keep a complete record of observations and conditions for each station visit. This shows not only problems discovered, but it provides good evidence the station was confirmed in excellent condition most of the time . 67

Measurement Station Siting and Deployment

Station Design Considerations Collecting useful (accurate) solar resource data requires careful design and implementation of a measurement station. Errors in the design stage of a measurement campaign can lead to serious analytical problems in the future.


Targeted Measurement Area


T1 15 km

T2 15 km

Background source: Google Earth

• Ideally, the measurement area should be concurrent with or close to the area of interest, but MUST be representative of the area of interest. • Some spatial offset (distance) can be tolerated, but it depends on the local climate and terrain. Lower variability in terrain and climate generally means lower solar variability at a distance. • More complex terrain (mountains, valleys, bodies of water) can mean more variability in the solar resource.

In this example, transferring the measurement to Target-1 may be adequate, but maybe not for Target-2. 70

Spatial Variability Variability is Variable Maps showing spatial variability across the United States based on variance within a 30x30 km or 50x50 km matrix. Variability ranges from a fraction of a percent to greater than 5%. Variability generally increases with distance. Monthly variability is greater.


Complicating Factors The need for spatially concurrent data must be weighed against other factors

• Availability of power, communications, and access for routine maintenance. Each of these limitations can be usually be overcome, but each can have considerable cost. • Local sources of dust—for example a nearby dirt road with heavy traffic—can introduce error. Moving the station a few hundred meters away could improve resulting data significantly.


Complicating Factors • In metropolitan areas, consider sources of radio frequency interference. Sensors produce a signal of only a few millivolts; strong interference could cause unwanted noise. • High locations with no obstructions may also host high powered transmitters. Consider harmful effects on humans working at the station. 73

Nearby Obstructions • Trees, buildings, power poles can cast shadows on the instruments, interfering with measurements that are intended to represent the resource at the target area. Even the best measurements from the best instruments will be degraded by nearby shading, and may impose a large uncertainty in the data. • Nearby objects may also reflect sunlight onto the sensor, resulting in an overestimation of the solar resource. • Nearby mountains may shade the instruments, but may be a legitimate obstacle if the same shading occurs at the target area. 74

Nearby Obstructions

•Tools exist to calculate the sun’s location along the horizon throughout the year. •This will help determine if an object will be a problem with DNI measurements.

University of Oregon Sun Path Chart Program


Station Accessibility and Security Measurement stations hold very valuable equipment. • Such equipment is not the typical target of thieves, but they may mistakenly believe the equipment has broad resale value • Vandals are a more likely problem; vandals care less about what they’re destroying than the act of destruction itself and causing problems for the owner.


Power Requirements- Ground and Shielding Reliable data acquisition requires reliable power for equipment. Some applications cannot tolerate data gaps. These may require backup battery power or Uninterruptable Power Supply (UPS).

Remote applications distant from the grid will require a local power supply (PV, wind, or petroleum generator). PV systems with batteries should be sized to sustain power for several days of clouds and shorter winter days.Test backup systems regularly

Station equipment needs to be protected from lightning strikes and shielded from radio frequency interference. 77

Data Communications Data should be collected as soon as possible after acquisition. • Data quality assessment methods are more effective with more rapid data turnaround. • The potential for data loss increases when a data set sits in memory on the logger.

One minute updates are technically feasible with cell phone modems.

• Real-time updates may be posted on the internet at oneminute intervals. • Rapid updates may not be practical in areas with poor cell phone connectivity. 78

Station Validation • Confirm proper station performance with reference instruments. • Verify correct installation (a complicated process; many things can go wrong) • Document all equipment—”If it isn’t documented, it didn’t happen…” • Train maintenance personnel • Establish a baseline condition for future evaluations.

Commissioning Report – with Equipment Lists, Photographs, Validation plots


Station Setup Key Points • Choose a location representative of your target area • Assure good solar access; avoid shading • Design for easy access and a safe operations • Provide for adequate site security • Plan for reliable power • Ground for equipment protection and low noise • Set up reliable and fast data communications • Validate installation for baseline performance


Resources NREL Solar Radiation Research Lab • Best Practices Handbook for the Collection and Use of Solar Resource Data •


Introduction to Satellite Resource Assessment

Topics for discussion • Introduction to Satellite Resource Assessment • Developing Datasets for India • Validation using Ground Station Data • Dissemination and Uses (discussed in later session)


What does Satellite Based Resource Assessment Provide? • Long-term datasets covering large areas at high temporal and spatial resolution. • Satellite instrument calibrated daily provides uniform measurements. • Only source of long-term data for various studies (currently 2000-2014 for India). • Source of data for performance models such as SAM, PVWatts etc.


What Impacts Satellite Measurements (looking from the top)? Meteosat Channel 1: 0.45-1 μm - Visible channel impacted by sunlight reflected from cloud or ground.

Clear Sky

Cloudy Sky

Surface reflectivity (albedo)

Cloud reflectivity (albedo)

First Order:

Second Order:


First Order:

Second Order:

Upper atmospheric water vapor

Surface Reflectivity: Dynamically calculated using darkest pixel from 30 day moving average of satellite scenes. Snow cover determined using observations. Climatological land database used for extreme events like 30 days of continuous cloudiness (rarely happens). Meteosat Channel 3: 10.5-12.5 μm – Infrared channel only influenced by temperature (height) of clouds and surface in clear sky situations.

Clear Sky

First Order:

Surface temperature Second Order:

Water vapor content

Cloudy Sky

First Order:

Cloud top temperature Second Order:



What Impacts GHI & DNI – i.e. Sunlight Reaching the Surface?

Clear Sky

Cloudy Sky

Aerosols, Mixed gases (Rayleigh)

Cloud optical thickness

First Order:

Second Order:

Water vapor, ozone Lower order: None

First Order:

Second Order:

Water Vapor Lower order: Aerosols, Surface Reflectivity


Meteosat 1st Generation: Meteosat 7 for Resource Assessment 3 channels: Image Band VIS WV IR spectral range 0.45-1.0 um 5.7-7.1 um 10.5-12.5 um resolution at nadir 2.5 km 5.0 km 5.0 km

Indian Ocean Data Coverage (IODC) Meteosat-5,6,7 provides the Indian Ocean Data Coverage (IODC) service. The current near real-time data are rectified to 57deg E.


How is surface radiation modeled from satellites?

•Empirical Approach:

–Build model relating satellite measurements and ground observations. –Use those models to obtain solar radiation at the surface from satellite measurements.

• Semi-Empirical Approach:

–Retrieve “cloud index” using counts from visible satellite measurements –Use clear sky radiative transfer models and scale by cloud index

•Physical Approach:

–Retrieve cloud and aerosol information from satellites –Use the information in a radiative transfer model 88

Developing the datasets for India

The Satellite Based Solar Radiation Product • Developed using the Semi-Empirical Approach • Uses the visible channel of Meteosat (5, 6 and 7) satellites at 10 km resolution • Hourly data • Currently covers 2000-2014 • Validation in collaboration with NIWE (planned); other partners are always welcome


Steps in the Semi-Empirical Approach • Scale visible satellite counts by solar zenith angle • Develop a dynamic of values for the satellite counts that represent the clear to cloudy range • Identify a clear sky radiative transfer model (preferably a fast model) to calculate clear sky GHI and DNI • Develop inputs for the clear sky model from ancillary source (primarily aerosol optical depth and precipitable water vapor) • Calculate the cloud index from the satellite counts • Scale the clear sky GHI by the cloud index • Calculate DNI from GHI using an empirical conversion model.


Creating Dynamic Range


Calculating the Clear Sky GHI and DNI SOLIS Model: Used to compute GHIclear and DNIclear Simplified Clear Sky model developed by Ineichen (2006,2008) Incorporates effect of : Precipitable water Site elevation Solar zenith angle Aerosol attenuation Ineichen, P., (2006): Comparison of eight clear sky broadband models against 16 independent data banks, Solar Energy 80, 4, pp. 468-478 Ineichen P., (2008): A broadband simplified version of the Solis clear sky model. Solar Energy, 82, 8, pp. 758-762


Basic principle

F+G F+TOA Cloud Index

Richard Perez, et al. 94

= a – b F-TOA F+TOA Clearness Index

Calculating Cloud Index and GHI CI = (UB – CCC) / (UB – LB) UB = Counts Upper Bound LB = Counts Lower Bound CCC = Counts Cosine Corrected


DNI Estimation from GHI Historical background • Most radiometric stations in the world measured GHI only; some additionally measured DIF • Systematic measurement of DNI with pyrheliometers or RSRs is recent • Liu & Jordan (1960) first suggested a statistical relationship between DIF and GHI, allowing the estimation of DIF and DNI • Main predictor: Kt = I / I0 [“clearness index”; removes a large part of the dependence of GHI on Z] • I0: ET irradiance on the horizontal ⩧ Solar constant corrected for actual sun-earth distance • Three possible forms used in the literature: K = Id / I = f(Kt)

Kd = Id / I0 = g(Kt) Kn = In / I0n = h(Kt) • All are mathematically equivalent, because I = Id + In cosZ and I0 = I0n cosZ): Kd = K / Kt Kn = Kt – Kd

Significant developments in “Separation models” • Addition of other predictors: Z, PW, albedo, sunshine • Consideration of persistence in the GHI daily record to separate clear and cloudy periods • Aggregation of datasets from many different climates for a “universal” relationship 96

DNI Estimation from GHI Temporal effects • Relatively good correlation for long time steps, e.g., monthly values • Cloud effects become increasingly dominant when time step shortens • Erbs model (1982) used to derive DNI and DIF in various datasets, e.g., NASA/SSE • Perez model (1992) used in some commercial datasets (SUNY/CPR, GeoModel, 3Tier…) [DIRINT MODEL]


DNI Estimation from GHI Validation

• Many validation studies exist, e.g.,

• Strong scatter (noise) at hourly or sub-hourly time scales • Most empirical models do not consider effects due to high ground albedo, haze, cloud enhancement… • More physical models would be needed


Surface Albedo Effects First effect: backscattering • The incident radiation reflected by the earth’s surface is backscattered in part by the atmosphere (more strongly if cloudy), and increases the diffuse component (DIF)—and hence GHI • The magnitude of this process is normally small, but can become large if snowy ground under low overcast sky (whiteout phenomenon) • The surface albedo to consider is the average over an area ≈20 km around the location under scrutiny • Albedo of various surfaces differ significantly, both for spectral and broadband radiation

Second effect: local reflections • To be considered when calculating GTI or GNI on tilted surfaces • Albedo of many surfaces is nearly Lambertian (isotropic) • Important exceptions do exist (e.g., sand): sun-tilted surface or sun-ground-satellite geometry is important • At grazing angles, reflection on water or ice becomes specular


Sources of Atmospheric Data Water vapor • • • •

Measured in terms of “precipitable water” Most common unit: cm Other units: mm, kg/m2, g/cm3 [1 mm ⩧ 1 kg/m2; 1 cm ⩧ 1 g/cm3] Relatively high in tropical environment but there is seasonal as well as locational variability

Precipitable water: Annual average for 2009 (CFSR reanalysis) 100

Sources of Atmospheric Data Water vapor • Water vapor has spatial variations and temporal variations (daily, seasonal, interannual) • Many sources of data: o Ground observations: – GPS (meteorological services, IGS network, …) – Sunphotometers (AERONET network, …) o Radiosondes (meteorological services) o Satellite observations (MODIS, SSMI…) o Satellite-derived products (CM-SAF…) o Reanalysis (MERRA, CFSR…) • Spatial and temporal resolutions vary! • Usual problem of observations: missing data!


Sources of Atmospheric Data Aerosols: Ground observations • Main world network: AERONET, • Many stations, but most were short lived • Cimel sunphotometer: 7 channels for AOD(λ), 1 channel for PW • Automatic tracking and operation, posttreatment to remove cloud-contaminated data • More stations are desirable in India

All stations

>5 years


Sources of Atmospheric Data Aerosols: Satellite observations vs. Modeled data




GOCART model


MACC model

Long-term mean annual AOD at 550 nm 103

Aerosol Effects on Irradiance Effect of using monthly-average AOD data on the modeled daily irradiance • Ignoring the daily variability in AOD leads to incorrect estimates of daily and hourly DNI or GHI • Risk of stepped changes at monthly seams: small for GHI, significant for DNI


Aerosol Effects on Irradiance Effect of using monthly-average AOD data on the modeled daily irradiance • DNI is affected much more than GHI • Effects on GTI would be intermediate • This is detrimental to the correct performance simulation of CSP systems and their bankability • The DNI frequency distribution is skewed differently at sunny and cloudy sites, both on hourly and daily time scales • The mean monthly or annual DNI is not a sufficient statistic to describe the DNI resource


Statistical Properties of Aerosol Data Frequency distributions of AOD • AOD (at any wavelength) follows a log-normal distribution • Log(τaλ) follows a normal distribution • Flattening of the distribution when mean AOD increases


Sources of Uncertainty • The lower bound of dynamic range is a function of the ground reflectivity (the albedo) • Albedo is variable over time • Albedo is a function of both sun-earth and sun-satellite geometry.


Sources of Uncertainty • Higher occurrence of variability of clouds • Higher concentration and dynamics of aerosols and water vapor • Mountainous terrain, high elevation and deep valleys. • Coastal Zones and regions with mixed land and water patterns • Urban and Industrial Environments • Snow and high-albedo surfaces (salt-beds, white sand areas)


Dissemination and Uses

Meteorological data for India Accurate meteorological data for use in the India dataset

• Various Reanalysis Datasets compared with ground measurements (Integrated Surface Database (ISD)) to identify best data. • NASA Modern EraRetrospective Analysis (MERRA) dataset, NOAA’s North American Regional Reanalysis (NARR) dataset and NOAA’s Climate Forecast System Reanalysis (CFSR) compared. • MERRA found to be the most accurate.

(A) (B)



Comparison between ISD and MERRA, CFSR and NARR dataset for Denver International Airport. (A) Dew Point, (B) Preciptable Water, (C) Atmospheric Pressure and (D) Wind Speed comparison


The datasets on the web • Hourly and TMY datasets are freely available from NREL’s website • Website: • Downloadable pixel by pixel or regions • Plans to distribute the full dataset using Amazon cloud services. • No restrictions on data except providing acknowledgement to NREL/USDOE and MNRE 111

Typical Meteorological Year Based on Hall, Prairie, Anderson, and Boes (Sandia National Labs)

A Typical Meteorological Year is a single year of data that represents the climate in a longer term data set. Useful for fast simulations using minimum data • Start with a multi-year data set (1530 years) • Build cumulative distribution functions for each conglomerated month (e.g. 30 Januarys, 30 Februarys, etc) for the entire pool of data • Build CDfs for each individual month • Compare individual monthly CDFs with long-term monthly CDFs using Finkelstein-Schafer (FS) statistic • Select the month with the best match (with other factors) 112

Regional applicability of ground measurements



Summary • • • •

Datasets freely available on the web for all users. Both solar and ancillary information is available. Hourly and TMY data is available for all locations. Satellite based data can determine the use of a ground station for surrounding regions. • Measurement at certain locations maybe more applicable to other surrounding locations than others. • Additional GIS based tools can make the data more useful for various stakeholders.


Thank You! Questions? Contact: [email protected]


View more...


Copyright © 2017 PDFSECRET Inc.