Efficiency test of earthquake prediction around Thessaloniki from electrotelluric precursors

Since the completion of the network in January 1983, the electric field of the earth has been continuously monitored at four sites near Thessaloniki, the capital of northern Greece. From the present study and from previous investigations by similar measurements in Greece, it is evident that transient changes of the electrotelluric field occur prior to earthquakes. The analysis of these electric forerunners leads in many cases to a successful prediction of the epicentral area, the magnitude and the time of the impending event. Predictions prior to regional earthquakes are issued and documented with telegrams.From November 1983 until the end of May 1984 twelve earthquakes (M_L > 3.5) occurred in the vicinity of Thessaloniki. Ten of these were predicted and warnings given by telegram, whereas two smaller seismic events were missed. Two additional predictions were unsuccessful. Independent of their magnitudes, predicted events took place within a time window of 6 hrs to 6 days after the observations of the electrotelluric anomalies. The accuracy of the predicted epicenters in eight cases is better than 100 km, which corresponds roughly to the mean distance between the electric stations. Magnitude estimates deviate by less than 0.5 magnitude units from the seismically observed ones.Considering the two largest earthquakes, it is shown that the probability of making each of these predictions by chance is of the order of 10⁻².     

Comparison of standard and digital observation at the Geomagnetic Adolf Schmidt Observatory Niemegk

For more than 4 yr the geomagnetic field has continuously been recorded by means of 3 proton magnetometers (F, X. Y) at Niemegk Observatory (1 value/min on punched tape). With this observation material, for all 7 geomagnetic components hourly means are calculated. The preliminary hourly means forF, Y, andX are published in the monthly report. As a rule, their comparison with the observatory data is satisfactory. Only with great geomagnetic activity (K ₁7) what was the case especially in 1982 relatively often, deviations in the hourly means up to 30 nT occurred. Obviously, during great magnetic disturbances the scanning rate of 1 min is not sufficient. In case of quick data variations, however, small technical defects in the electronic circuit can occur. Besides, these great differences could also be explained by errors which could be conditioned by the standard observational and processing methodology.     

Measurements of solar occulation: the error in a naive retrieval if the constituent's concentration changes

The stratospheric concentrations of many minor constituents change rapidly at sunrise or sunset. If this happens, there is an inherent error when retrieving the vertical profiles of the constituents from measurements of their absorption of sunlight. For retrievals of NO at sunset the error can be estimated from in-situ measurements alone, without appeal to a model of stratospheric photochemistry. Below 20 km this error can approach 100% so that the retrieved NO is zero. But at 40 km, and at 25 km when the absorption is strong and Lorentzian, it can be less than 20%. Precise calculations of the error, even if small, require model calculations of the sunset and sunrise changes. With a model, we have calculated the error for NO, NO₂, OH and ClO.