MODELED VARIATIONS OF PRECIPITATION OVER THE GREENLAND ICE-SHEET

A parameterization of the synoptic activity at 500 hPa and a simple or ographic scheme are used to model the spatial and temporal variations of precipitation over the Greenland ice Sheet for 1963-88 from analyze d geopotential height fields produced by the National Meteorological C enter (NMC). Model coefficients are fitted to observed accumulation da ta, primarily from the summit area of the ice sheet. All major spatial characteristics of the observed accumulation distribution are reprodu ced apart from the orographic accumulation maximum over the northweste rn coastal slopes. The modeled time-averaged total precipitation amoun t over Greenland is within the range of values determined by other inv estigators from surface-based observations. A realistic degree of inte rannual variability in precipitation is also simulated. A downward tre nd in simulated ice sheet precipitation over the 26 years is found. Th is is supported by a number of lines of evidence. It matches the accum ulation trends during this period from ice cores drilled in south-cent ral Greenland. The lower tropospheric specific humidities at two south coastal radiosonde stations also decrease over this interval. A syste matic shift away from Greenland and a decrease in activity of the domi nant storm track are found for relatively low precipitation periods as compared to relatively high precipitation periods. This negative prec ipitation trend would mean that the Greenland Ice Sheet, depending on its 1963 mass balance state, has over the 1963-88 period either decrea sed its negative, or increased its positive, contribution to recently observed global sea level rise. Superimposed on the declining simulate d precipitation rate for the entire ice sheet is a pronounced 3-5-yr p eriodicity. This is prominent in the observed and modeled precipitatio n time series from Summit, Greenland. This cycle shows some aspects in common with the Southern Oscillation. Some deficiencies in the NMC an alyses were highlighted by this work. A large jump in simulated precip itation amounts at Summit around 1962, which is not verified by accumu lation data, is inferred to be due to an artificial increase in cyclon ic activity at 500 hPa associated with the NMC change from manual to n umerical analyses. The activity of the storm track along the west coas t of Greenland appears to be anomalously low in the NMC analyses, perh aps due to mesoscale cyclogenesis that is not resolved by the NMC anal ysis scheme.