In this exclusive interview, Shane Marie Visaga, a key researcher at the Manila Observatory and Ateneo de Manila University, shares insights into their groundbreaking study on improving solar radiation forecasts in the Philippines. This research leverages advanced numerical weather prediction models, like WRF-Solar, combined with statistical techniques such as the Kalman Filter, to enhance forecast accuracy. With a focus on real-time applications for industries like solar energy and agriculture, Visaga discusses the unique challenges of forecasting in a tropical climate, the role of seasonal variations, and the transformative potential of accurate solar radiation forecasts for renewable energy planning. This study not only addresses critical climate-related challenges but also underscores the importance of collaboration and innovation in tackling the climate emergency in the Philippines.

Could you provide an overview of the research and explain why improving solar radiation forecasts is vital for the Philippines, particularly in the context of industries like solar energy and agriculture?

Our solar energy forecast study is part of the Manila Observatory’s efforts to mitigate the climate emergency in the Philippines. We have evaluated and enhanced the solar radiation forecasts generated by the freely available Numerical Weather Prediction tool known as the Weather Research and Forecasting-Solar (WRF-Solar) model, with a particular focus on real-time applications for the solar industry. By applying the post-processing technique known as the Kalman Filter, we can reduce the bias error in solar radiation forecasts by up to 94%. This technique utilizes the performance data from the previous three days to refine the forecast for the subsequent day.

The Kalman Filter (KF) played a significant role in your study. Could you elaborate on how KF improves the accuracy of WRF-Solar forecasts and why it was chosen over other algorithms?

One of the key advantages of the Kalman Filter (KF) is its adaptability. This adaptability is the reason why KF is widely used in weather forecasting, as well as in tracking and navigation using radars. The technique effectively reduces biases and errors in the WRF-Solar model by recursively comparing the model’s predictions with observed solar radiation and making adjustments based on the specified training days. While we should ideally explore and compare the performance of other algorithms, the KF has proven to be highly effective. Our post-processing efforts with the KF are the result of a fruitful collaboration with Dr. Alvarenga from UMR Espace-Dev, University of French Guiana, who generously shared their Python programs aimed at similar forecast improvements in their region.

What are the specific challenges in forecasting solar radiation in a tropical country like the Philippines, and how does the KF algorithm address these challenges?

The primary challenge in solar forecasting in tropical regions like the Philippines is accurately capturing the timing and location of cloud formation. Improving the representation of clouds in models is an ongoing area of research. For near industry-grade solar energy applications, we can rely on faster statistical techniques such as the Kalman Filter (KF) to leverage recent observations. The KF algorithm helps to statistically correct and improve the model’s biases and errors to enhance the accuracy of solar radiation forecasts.

The study found that the optimal training days for the Kalman Filter varied by season. Could you explain why these variations exist and how they impact forecast accuracy during the wet and dry seasons?

The seasons in our region, wet and dry, as it implies, are based on rainfall amounts, which brings us back to clouds! The WRF-Solar model we are using performs really well on clear-sky conditions and struggles with cloudy conditions. Since our dry season has more clear-sky conditions and our wet season has more cloudy conditions, this seasonal difference in cloud cover affects our statistical technique that relies on how well the initial WRF-Solar compares with the observed solar radiation.

How do you envision the improved forecasts benefiting industries like solar power and agriculture in the Philippines? Could you provide specific examples of potential impacts?

Improving the solar radiation forecasts is an initial step towards solar power applications. In the industry, we should take another step and do solar power plant modeling which takes into account the power plant modules, and power wire losses, etc.

Your team mentioned plans to apply the methodology to different topographies in the Philippines. What are the unique challenges and opportunities in scaling this research across the country’s diverse landscapes?

We are highly relying on quality observation datasets to correct for the model biases and errors. In this study, we had an opportunity to compare the models with a ground-based pyranometer deployed in the Manila Observatory by Anthony Bucholtz and Betsy Reid from the US CIRPAS Airborne Research Facility and Naval Research Laboratory. As the ECW project (https://panahon.observatory.ph/ecw/) continue to deploy more weather stations, we gain more opportunities for model evaluation and improvement for different regions of the country. This is also true as we continue to collaborate with other institutions like PAGASA with a vast network of weather stations, and existing and upcoming solar power plants.

One of the key benefits of your approach is its computational efficiency. How do you see this advantage being leveraged for broader adoption, particularly in resource-constrained environments?

In our paper, we emphasize that the KF WRF-Solar approach complements the more computationally intensive Numerical Weather Prediction (NWP) ensemble forecasts. Ensemble forecasting involves running multiple configurations of a model or using different types of NWP models. We use both approaches because NWP models offer valuable insights into land-air interactions and provide opportunities for further improvements. In resource-constrained environments, complementing a single NWP forecast with statistically improved solar forecasts using KF can be highly beneficial.

Looking ahead, how do you think this research could shape the future of renewable energy planning and weather forecasting in the Philippines? What broader implications might it have for tackling climate-related challenges?

This study is part of the Manila Observatory’s efforts to mitigate the climate emergency in the Philippines. As more researchers join the conversations and efforts on renewable energy, we hope to expedite the transition to a cleaner, greener economy in our country. We are also excited to explore further improvements to the WRF model through Kalman Filter. We have more students working on using WRF-Solar to understand which configurations work best for the Philippines. By refining our predictive models and understanding of atmospheric processes, we can better equip communities to face climate-related challenges.

IMAGE CREDIT: Ramon FVelasquez

Sign up for the Daily Dose Newsletter and get the morning’s best science news from around the web delivered straight to your inbox? It’s easy like Sunday morning.