In weather forecasting and meteorological research, satelite images are used more and more, for example to measure the water density on a patch of sky. Early 2006, at the Utrecht Stochastics Seminar, a Dutch meteorologist from KNMI discussed the possibility of validating satellite observations with ground observations. In my thesis I present such a method of validation using statistical testing and the principle of kernel estimation.
The aim of my thesis was finding a statistical test that can validate satellite measurements of the so-called Liquid Water Path, LWP, which is the density of water in a certain area in the sky. This should be possible using observations from groundstations. The data of both observations differ greatly in, for example, resolution and presentation, and a signifficant part of my research involved making both types of data comparable. An important tool used is the method of kernel estimation.
In my speech I will explain the principles of kernel estimation and explain its application. I will then discuss the testing hypothesis and its qualities, aided by some examples and simulations.