Authors: (1) Maggie D. Bailey, Colorado School of Mines and National Renewable Energy Lab; (2) Douglas Nychka, Colorado School of Mines; (3) Manajit Sengupta, National Renewable Energy Lab; (4) Aron Habte, National Renewable Energy Lab; (5) Yu Xie, National Renewable Energy Lab; (6) Soutir Bandyopadhyay, Colorado School of Mines. Table of Links Abstract and Intro Data Bayesian Hierarchical Model (BHM) Solar Radiation Example Results Conclusion Appendix A: Simulation Study Appendix B: Regridding Coefficient Estimates References 6 Conclusion This study analyzes the uncertainty in regridding of spatial data from climate models, which is often the first step in multi-model climate analysis. Solar radiation data is regridded from its native grid, using kriging with an exponential covariance function and a loglinear transformation, to the same grid as the NSRDB. Second, we implement a BHM to estimate linear model weights while incorporating the uncertainty associated with the regridding step. Finally, we compare the two and provide an additional simulation study in Appendix A. The naive regridding model coefficient estimates were found to be within range of the posterior distributions of the model coefficients in most cases. Seasonally, the month of August produced a mismatch between the naive regridding coefficient and the posterior distribution for the WRF RCM forced by ERA-Interim. In particular, we saw that resulting coefficient estimates in this month for the WRF were higher in the naive method than BHM. This suggests that when regridding uncertainty is taken into account, there is a smaller increase in the WRF data for unit increases in the NSRDB, or that the regridding uncertainty may result in less bias from the WRF in this particular case. It was found that the posterior coverage for test data for the simulated fields were similar to the naive regridding estimates for the months of August and November. This suggests that when taking into account the regridding uncertainty of the simulated fields and the model parameters themselves, the true value of solar radiation in this case is still likely to be covered by the 95% credible interval. Therefore if the conditional mean of the regridded field were taken for ground truth, as it often is, downstream effects of regridding on modeling appear to be minimal in the case of solar radiation. However, the BHM had higher RMSE values than the naive regridding models in the months considered indicating that the addition of the regridding uncertainty increased prediction error for out of sample prediction. It is important to note that the naive regridding coefficient estimates give good predictions but are not appropriate to assess the model biases directly since model biases are dependent on regridding. Finally, this analysis serves as a framework for understanding regridding effects within the context of solar radiation. While this study did not find situations where the BHM regridding consistently outperformed the naive regridding method, we note that this analysis revolves around the chosen variable: GHI. It has been shown that the chosen regridding method has an impact on the extremes of distributions (McGinnis et al. (2010)), however extremes are not central to solar radiation. A future analysis applying the BHM regridding method to climate variables where extremes of the data are more widely studied, such as precipitation or temperature, may yield different results and provide an example where the method proposed in this paper might show higher uncertainty in downstream modeling. Additionally, this study takes into account a single type of regridding (kriging with an exponential covariance) and this analysis could be extended to other types of interpolation to understand downstream effects of those particular methods. This paper is under CC 4.0 license. available on arxiv

Navigating Regridding Uncertainty in Solar Radiation Climate Analysis

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

A Comprehensive Look at Solar Radiation Predictions through Bayesian Hierarchical Models

$PEPE, a Purple Lamborghini, and More: The Story Continues

102 Most Important Webism Quotes by Unknown Authors

100 Days of AI Day 3: Leveraging AI for Prompt Engineering and Inference

100 Days of AI, Day 8: Experimenting With Microsoft's Semantic Kernel Using GPT-4

10 Tips to Take Your ChatGPT Prompts to the Next Level

A Comprehensive Look at Solar Radiation Predictions through Bayesian Hierarchical Models

$PEPE, a Purple Lamborghini, and More: The Story Continues

102 Most Important Webism Quotes by Unknown Authors

100 Days of AI Day 3: Leveraging AI for Prompt Engineering and Inference

100 Days of AI, Day 8: Experimenting With Microsoft's Semantic Kernel Using GPT-4

10 Tips to Take Your ChatGPT Prompts to the Next Level

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps