Page - 145 - in Contributions to GRACE Gravity Field Recovery - Improvements in Dynamic Orbit Integration, Stochastic Modelling of the Antenna Offset Correction, and Co-Estimation of Satellite Orientations

which is similarly homogeneous, but larger in magnitude. It could be argued that in both figs. 9.9c and 9.9e some systematic differences (compare red markers) can be seen at≈50°N and at≈30°N, here especially in Mexico and off the west coast of North America. This signal on its own is however not strong enough to draw any definitive conclusion from it. For June 2010, the month of deteriorated quality due to larger opening angles, the differences are in general larger and more spatially distinct. Especially at southern mid- latitudes, two bands of increased differences can be seen in both figs. 9.10c and 9.10e, representing the differences due to the AOC covariance matrices and TLS respectively (compareredmarkers).The latitudesof thesebandscorrespondroughlytothe latitudes of the northern bands observed in figs. 9.9c and 9.9e. These northern bands are not as distinct for June 2010, but they are still present. Table 9.3 gives the average EWH RMS for these two monthly solutions, for all three processing strategies and for different geographical areas. The RMS is evaluated globally and separately over the continents and oceans. Using perfect background models and flawless processing, the RMS over the oceans would be expected to be close to zero, as no time-variable signal should remain in this area. Over the continents however, the signal of the hydrological cycle and the other processes described in chapter 3 should remain visible in the results. To further isolate the ocean signal from possible interference from continental signals, a separate remote area in the south Pacific was analysed. This is the area marked in red in figs. 9.9a and 9.10a. Table 9.3: Global and geographically restricted RMS of recovered equivalent water heights for April 2008, a month of good data; and June 2010, a month of deteriorated data. A 300km Gaussian filter was applied to the data. Period / Area Old model [cm] Incl. AOC [cm] With TLS [cm] April 2008 Global 3.81 3.81 3.72 Continents 4.60 4.60 4.55 Ocean 3.44 3.43 3.33 Pacific patch 3.43 3.41 3.19 June 2010 Global 3.95 3.97 3.76 Continents 4.79 4.82 4.66 Ocean 3.55 3.57 3.33 Pacific patch 3.22 3.27 2.97 9.3 Results 145