The rapid expansion of intermittent grid-tied solar capacity is making the job of balancing electricity's real-time supply and demand increasingly challenging. To address the problem, recent work proposes mechanisms for actively controlling solar power output to the grid by enabling software to cap it as a fraction of its time-varying maximum output. Utilities can use these mechanisms to dynamically share the grid's solar capacity by controlling the solar output at each site. However, while enforcing an equal fraction of each solar site's time-varying maximum output results in "fair" short-term contributions of solar power, it does not result in "fair" long-term contributions of solar energy. This discrepancy arises from fundamental differences in enforcing "fair" access to the grid to contribute solar energy, compared to analogous fair-sharing in networks and processors. In this paper, we present a centralized and distributed algorithm to enable control of distributed solar capacity that enforces fair grid energy access. We implement our algorithm and evaluate it on synthetic data and real data across 18 solar sites. We show that traditional rate allocation, which enforces equal rates, results in solar sites contributing up to 18.9% less energy than an algorithm that enforces fair grid energy access over a single month.
more »
« less
A moment in the sun: solar nowcasting from multispectral satellite data using self-supervised learning
Solar energy is now the cheapest form of electricity in history. Unfortunately, significantly increasing the electric grid's fraction of solar energy remains challenging due to its variability, which makes balancing electricity's supply and demand more difficult. While thermal generators' ramp rate---the maximum rate at which they can change their energy generation---is finite, solar energy's ramp rate is essentially infinite. Thus, accurate near-term solar forecasting, or nowcasting, is important to provide advance warnings to adjust thermal generator output in response to variations in solar generation to ensure a balanced supply and demand. To address the problem, this paper develops a general model for solar nowcasting from abundant and readily available multispectral satellite data using self-supervised learning.
Specifically, we develop deep auto-regressive models using convolutional neural networks (CNN) and long short-term memory networks (LSTM) that are globally trained across multiple locations to predict raw future observations of the spatio-temporal spectral data collected by the recently launched GOES-R series of satellites. Our model estimates a location's near-term future solar irradiance based on satellite observations, which we feed to a regression model trained on smaller site-specific solar data to provide near-term solar photovoltaic (PV) forecasts that account for site-specific characteristics. We evaluate our approach for different coverage areas and forecast horizons across 25 solar sites and show that it yields errors close to that of a model using ground-truth observations.
more »
« less
- Award ID(s):
- 2020888
- NSF-PAR ID:
- 10340729
- Date Published:
- Journal Name:
- Proceedings of the Thirteenth ACM International Conference on Future Energy Systems (e-Energy)
- Page Range / eLocation ID:
- 251 to 262
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Though the partitioning of shortwave radiation ( K ↓ ) at the surface into its diffuse ( K ↓,d ) and direct beam ( K ↓,b ) components is relevant for, among other things, the terrestrial energy and carbon budgets, there is a dearth of large-scale comparisons of this partitioning across reanalysis and satellite-derived products. Here we evaluate K ↓ , K ↓,d , and K ↓,b , as well as the diffuse fraction ( k d ) of solar radiation in four current-generation reanalysis (NOAA-CIRES-DOE, NCEP/NCAR, MERRA-2, ERA5) datasets and one satellite-derived product (CERES) using ≈1400 site years of observations. Although the systematic positive biases in K ↓ is consistent with previous studies, the biases in gridded K ↓,d and K ↓,b vary in direction and magnitude, both annually and across seasons. The inter-model variability in cloud cover strongly explains the biases in both K ↓,d and K ↓,b . Over Europe and China, the long-term (10-year plus) trends in K ↓,d in the gridded products are noticeably differ from corresponding observations and the grid-averaged 35-year trends show an order of magnitude variability. In the MERRA-2 reanalysis, which includes both clouds and assimilated aerosols, the reduction in both clouds and aerosols reinforce each other to establish brightening trends over Europe, while the effect of increasing aerosols overwhelm the effect of decreasing cloud cover over China. The inter-model variability in k d seen here (0.27 to 0.50 from CERES to MERRA-2) suggests substantial differences in shortwave parameterization schemes and their inputs in climate models and can contribute to inter-model variability in coupled simulations. Based on these results, we call for systematic evaluations of K ↓,d and K ↓,b in CMIP6 models.more » « less
-
Abstract Because geostationary satellite (Geo) imagery provides a high temporal resolution window into tropical cyclone (TC) behavior, we investigate the viability of its application to short-term probabilistic forecasts of TC convective structure to subsequently predict TC intensity. Here, we present a prototype model that is trained solely on two inputs: Geo infrared imagery leading up to the synoptic time of interest and intensity estimates up to 6 h prior to that time. To estimate future TC structure, we compute cloud-top temperature radial profiles from infrared imagery and then simulate the evolution of an ensemble of those profiles over the subsequent 12 h by applying a deep autoregressive generative model (PixelSNAIL). To forecast TC intensities at hours 6 and 12, we input operational intensity estimates up to the current time (0 h) and simulated future radial profiles up to +12 h into a “nowcasting” convolutional neural network. We limit our inputs to demonstrate the viability of our approach and to enable quantification of value added by the observed and simulated future radial profiles beyond operational intensity estimates alone. Our prototype model achieves a marginally higher error than the National Hurricane Center’s official forecasts despite excluding environmental factors, such as vertical wind shear and sea surface temperature. We also demonstrate that it is possible to reasonably predict short-term evolution of TC convective structure via radial profiles from Geo infrared imagery, resulting in interpretable structural forecasts that may be valuable for TC operational guidance. Significance Statement This work presents a new method of short-term probabilistic forecasting for tropical cyclone (TC) convective structure and intensity using infrared geostationary satellite observations. Our prototype model’s performance indicates that there is some value in observed and simulated future cloud-top temperature radial profiles for short-term intensity forecasting. The nonlinear nature of machine learning tools can pose an interpretation challenge, but structural forecasts produced by our model can be directly evaluated and, thus, may offer helpful guidance to forecasters regarding short-term TC evolution. Since forecasters are time limited in producing each advisory package despite a growing wealth of satellite observations, a tool that captures recent TC convective evolution and potential future changes may support their assessment of TC behavior in crafting their forecasts.more » « less
-
more » « less
Abstract There is a need for nowcasting tools to provide timely and accurate updates on the location and rate of spread (ROS) of large wildfires, especially those impacting communities in the wildland urban interface. In this study, we demonstrate how fixed‐site weather radars can be used to fill this gap. Specifically, we develop and test a radar‐based fire‐perimeter tracking tool that leverages the tendency for local maxima in the radar reflectivity to be collocated with active fire perimeters. Reflectivity maxima are located using search radials from points inside a fire polygon, and perimeters are updated at intervals of ∼10 min. The algorithm is tested using publicly available Next Generation Weather Radar radar data for two large and destructive wildfires, the Camp and Bear Fires, both occurring in northern California, USA. The radar‐based fire perimeters are compared with available, albeit limited, satellite and airborne infrared observations, showing good agreement with conventional fire‐tracking tools. The radar data also provide insights into fire ROS, revealing the importance of long‐range spotting in generating ROS that exceeds conventional estimates. One limitation of this study is that high‐resolution fire perimeter validation data are sparsely available, precluding detailed error quantification for the radar estimates drawn from samples spanning a range of environmental conditions and radar configurations. Nevertheless, the radar tracking approach provides the basis for improved situational awareness during high‐impact fires.
-
null (Ed.)The launch of the National Oceanic and Atmospheric Administration (NOAA)/ National Aeronautics and Space Administration (NASA) Suomi National Polar-orbiting Partnership (S-NPP) and its follow-on NOAA Joint Polar Satellite Systems (JPSS) satellites marks the beginning of a new era of operational satellite observations of the Earth and atmosphere for environmental applications with high spatial resolution and sampling rate. The S-NPP and JPSS are equipped with five instruments, each with advanced design in Earth sampling, including the Advanced Technology Microwave Sounder (ATMS), the Cross-track Infrared Sounder (CrIS), the Ozone Mapping and Profiler Suite (OMPS), the Visible Infrared Imaging Radiometer Suite (VIIRS), and the Clouds and the Earth’s Radiant Energy System (CERES). Among them, the ATMS is the new generation of microwave sounder measuring temperature profiles from the surface to the upper stratosphere and moisture profiles from the surface to the upper troposphere, while CrIS is the first of a series of advanced operational hyperspectral sounders providing more accurate atmospheric and moisture sounding observations with higher vertical resolution for weather and climate applications. The OMPS instrument measures solar backscattered ultraviolet to provide information on the concentrations of ozone in the Earth’s atmosphere, and VIIRS provides global observations of a variety of essential environmental variables over the land, atmosphere, cryosphere, and ocean with visible and infrared imagery. The CERES instrument measures the solar energy reflected by the Earth, the longwave radiative emission from the Earth, and the role of cloud processes in the Earth’s energy balance. Presently, observations from several instruments on S-NPP and JPSS-1 (re-named NOAA-20 after launch) provide near real-time monitoring of the environmental changes and improve weather forecasting by assimilation into numerical weather prediction models. Envisioning the need for consistencies in satellite retrievals, improving climate reanalyses, development of climate data records, and improving numerical weather forecasting, the NOAA/Center for Satellite Applications and Research (STAR) has been reprocessing the S-NPP observations for ATMS, CrIS, OMPS, and VIIRS through their life cycle. This article provides a summary of the instrument observing principles, data characteristics, reprocessing approaches, calibration algorithms, and validation results of the reprocessed sensor data records. The reprocessing generated consistent Level-1 sensor data records using unified and consistent calibration algorithms for each instrument that removed artificial jumps in data owing to operational changes, instrument anomalies, contaminations by anomaly views of the environment or spacecraft, and other causes. The reprocessed sensor data records were compared with and validated against other observations for a consistency check whenever such data were available. The reprocessed data will be archived in the NOAA data center with the same format as the operational data and technical support for data requests. Such a reprocessing is expected to improve the efficiency of the use of the S-NPP and JPSS satellite data and the accuracy of the observed essential environmental variables through either consistent satellite retrievals or use of the reprocessed data in numerical data assimilations.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1145/3538637.3538854
