| description abstract | An objective analysis scheme for meteorological variables on constant potential temperature surfaces is presented. The analysis uses a form of multivariate statistical interpolation and is designed th retain mesoscale detail in disparate observations including rawinsonde, surface, aircraft, satellite, and wind profiler data while combining them with a forecast background (first guess) field. The wind and mass field analyses are interdependent. The horizontal correlation of forecast error on isentropic surfaces is modeled with an analytical function from statistics collected for this study; the vertical correlation of forecast error is modeled as a function of potential temperature separation. These correlations determine the weights applied to observed-minus-forecast increments in the analysis. The analysis is two-dimensional except with respect to single-level data where it is three-dimensional. Comparisons of isentropic and isobaric analysts are shown, and examples of the effects of single-level (aircraft and surface) observations on isentropic analyses are presented. Although variable in space and time, these datasets are often of higher density than the rawinsonde network, and they support increased resolution of mesoscale features in the analysis. More importantly, the examples reveal that three-dimensional analysis increment structures, especially in the vicinity of fronts, appear to be more physically reasonable in an isentropic analysis than in an isobaric analysis. | |
