Professional Grade Validation October 2019

Introduction

One of the most often asked questions about Vaisala’s satellite-derived irradiance datasets is simply: “How accurate is it?” Unfortunately, determining accuracy is one of those things that seems so simple in theory but is difficult in practice. The theory says just compare your predictions to high-quality observations at many locations to get a sense for accuracy. 

The difficulty is that obtaining high-quality observations for a public-facing validation study is easier said than done. Many public stations of high quality were already used to tune the algorithms behind satellite-derived estimates, so therefore they are not truly independent when being used for validation purposes. Public observations of lesser quality are available, but quality control of those data often reveal significant problems and removing those problematic data reduces the length of the comparison period. Obviously, many high-quality observation datasets have been collected by those developing utility scale solar projects — but those data are typically not available for use in public validation studies. 

The difficulty of obtaining high-quality and independent data for public-facing validation studies has frustrated both the providers and users of solar resource datasets. Those that are familiar with the difficulties of validation ask a follow-on question to the “How accurate?” question. They ask: “What data did you use to determine the accuracy of your data, and are those data truly independent from the observations you used to develop your datasets?”

...