Mutual evaluation of global gravity models (EGM2008 and GOCE) and terrestrial data in Amazon Basin, Brazil

The gravity observations of the satellite GOCE have a global homogeneous coverage and precision. This data set constitutes an independent new tool to control the quality of terrestrial gravity data. Terrestrial data reach higher resolution and precision, but can be affected by errors due to factors such as different vertical geodetic datums, wrong position in latitude and longitude, geodynamic effects and gravimeter drift, which tends to accumulate over long distances. Terrestrial data recover gravity signals at shorter wavelengths compared to the GOCE satellite, but the average gravity anomaly values can be compared to the GOCE derived values which are bandlimited to lower frequencies. We consider the area of the Amazon Craton, and in particular the Solim\u303oes , Amazon and Parnaiba Basins, and part of the Tocantins and S\u303ao Francisco Provinces in Brazil, to estimate the systematic errors in terrestrial gravity data. We calculate the average terrestrial gravity anomaly by spatial averages applying Gaussian, inverse distance and simple averages, which allows to compare the long- and medium-wavelength part of the terrestrial gravity anomalies with the gravity field derived from GOCE. We also consider the combined satellite-terrestrial model EGM2008 up to degree and order 250 (i.e. maximum expansion from satellite GOCE). The results show that the systematic errors range from about 12 28.1 to 25.2 mGal with a standard deviation value of 6.4 mGal. The mean value over the study area is about zero, obtaining 0.27mGal difference between the Gaussian average of the terrestrial gravity data and the gravity data from the GOCE satellite-only model and is smaller than the commission error associated to the geopotential model. Also, we verified that 64.8percent of the study area does not present systematic errors, as their difference is within the commission error of 5.1mGal of the GOCE model in the harmonic expansion up to degree 250. The comparison of the terrestrial data with the model EGM2008 gives slightly smaller differences, which can be attributed to the fact that the EGM2008 contains terrestrial data. The results vary only slightly according to the type of averaging used, with improved values for the Gaussian average. The analysis also shows where the terrestrial data are scarce and require an improvement in data coverage in order to correctly represent the gravity field. The method we propose can be directly used to control other gravity databases and constitutes a tool for the quality assessment of terrestrial gravity observations