GNGTS 2013 - Atti del 32° Convegno Nazionale

and covering the time interval 1997 – 2011. As we can see, our regional model presents a higher spatial var- iability than the global CHAOS-4 model due to the regional character- istic of our model. This characteristic is presented for all the geomagnetic field components and for the total field intensity. In order to see the spatial predic- tion of the model, we plot different crustal field maps in Fig. 2. The crus- tal field model was also calculated at 450 km of altitude and was compared with the prediction given by the CHAOS-4 model. Our results show a clear agreement with the global mod- el, but present more details in terms of the spatial wavelength. Upward continuation transforms anomalies measured at one surface into those that would have been measured at a higher altitude sur- face. In the simplest case, i.e., lev- el-to-level, upward continuation is merely a convolution of the original data, performed in either the space or Fourier domain. We made a lev- el-to-level continuation of the Earth’s surface CM4 data to 450 km. We first transformed the data coordinates from spherical to Cartesian, then used a maximum entropy extrapola- tor to realize a square map and final- ly extended the grid by 10% with a periodic extrapolator. The results are shown in Fig. 3. Fig. 2 – Maps of the X, Y, Z and F (from top to bottom) crustal field at 450 km as modeled by a set of real satellite data (right) and the crustal field calculated with CHAOS-4 model (left). Fig. 3 – Level-to-level upward continuation of total intensity F. The CM4 data at ground level were continued to 450 km (a) with a level-to-level convolution filter. Only little differences (c) among CM4 and continued data occur in a wide central area (within 1 nT), while major differences appear at the edges, i.e., where the edge-effect and the errors for not having taken into account the Earth’s curvature are stronger. However, by changing border extrapolator (not shown here) we observed that the edge-effect changes led the continued field to be improved in some areas and to get worse in others. This should demonstrate that most of the error is due to the edge-effect. 168 GNGTS 2013 S essione 3.2