Date: Thu, 10 Oct 2002 09:40:37 +0200 From: Escudier Philippe To: Berthias Jean-Paul , "John C. Ries" , Pascal Willis Cc: Escudier Philippe , Gilles Tavernier , Martine Feissel , "Jayles, Christian" , "Sengenes, Pierre" Subject: RE: satellite clock in the SAA region This paper gives a clear explanation of the problem according to my understanding, I let my collegues in the project team add other remarks. I would suggest the following comments/remarks : - it could be precised somewhere that thanks to the filtering which is applied, quality of precise orbit determination is not impacted by this phenomenom, it is why it shows up in the residuals. - this effect exists for all satellites but is presently larger for Jason. This is due to several potential causes : * orbit; T/P and Jason flies on a most severe radiation environment than SPOT or ENVISAT * receiver sensitivity to radiation; from the various tests performed at CNES and by other groups this sensitivity varies from one clock to the other. It also varies with time. There is usually some "saturation" of the quartz after a certain level of irradiation which significantly decreases this sensitivity. This may be part of the reason why present T/P data is impacted at a much lower level than Jason. We got some evidence that during the first months of T/P this effect was larger than today, carefull reprocessing to assess this point would be very interesting to support the precise understanding of the phenomenom. * different satellite shielding as mentionned in the paper, but from what we know today this should not be the larger cause because shielding is not very efficient for the high energy protons implied in this phenomenom. - concerning question 3, this quasi linear trend was the most striking effect which has made us looking for other causes for a while. Potential explanations of the variations with time of the effect are : * change in the radiation dose received by the clock, this is probably the most significant cause for short to long term variations. Physical causes are change of SAA characteristics (altitude, size, ...) according to local time and solar activity. Variations at 60 days period which appears in many parameters (clock analysis, residuals, localization,...) can be related to the change of local time of the SAA crossing. Considering the linear trend, we are in the decreasing part of the solar cycle, this is associated with a lowering of the altitude of the SAA (which maximum is above Jason orbit). Then this is probably the most credible explanation to the apparent linear increase of the effect. We have begun to work with a french group working on SAA dynamic modelling (ONERA/DESP), they confirmed the physical basis of this explanation and they declared a great interest in getting our results for validation/calibration of there dynamic modelling. I guess other groups may also be interested in these data to get a better understanding of the SAA dynamic. * change in the clock sensitivity, this is a relatively slow process and as mentionned above it should lead to a global reduction of the sensitivity in the future. - concerning question 4, it is true that the proper modelling of the effect is actually not a simple quadratic in the SAA. The way which has been investigated by Pierre is to use a static model of the radiation environment (AP8) which gives the radiation dose received all along the orbit, in the SAA this model has a "bell" shape (I hope this direct translation of french wording can be understood in english). Then, the clock frequency modelling is based on a the sum of a linear trend and the integrated radiation dose received by the satellite multiplied by a sensitivity coefficient. Actual data can be used to tune the 2 coefficients of this model, linear trend and sensitivity coefficient (the later being actually the product of the actual sensitivity and a multiplicative factor between actual and modelled radiations, this becomming a "new" DORIS product that could be used by radiation modelling groups to analyze the dynamic evolution of the SAA). First results are very encour! aging, I hope Pierre will be able to present the preliminary output during New orleans meeting. The first good news is that the sensitivity coefficient that can be computed this way is in the range of the known clock sensitivity values (from ground test). Once this preliminary analysis will be made, such a model would be available to anybody who would be interested in modelling that effect. The radiation model used has no dynamic so that it exactly repeats from one cyle to the other. We will provide the file giving the profile along a cycle, file that anyone may use for tuning of the parameters (the multiplicative factors may be tuned every x (TBD) hours/days, one may also think about different tuning parameters such as global rotation of the SAA or homothetic geographical increase, ...). When a consensus will be reached on the benefit of such a modelling and the best strategy for parameters tuning this could be implemented in the CNES pre processing software so that any group would benefit from that. An other question will be to decide wether or not such a processing will be beneficial to other satellites (even if the effect is smaller one may think that it has a significant part or the present error budget). Best regards Philippe