Effects Of Observation Error On Precipitation Measurement And Trend Estimates Over North And Northeast China

By using a dataset of daily precipitation, wind speed and weather phenomena of158stations during1960to2009, and the data of experimental observations of precipitation from11stations, this paper analyzes the effects of observational errors of precipitation on total precipitation amount and the long-term precipitation trends over North China. Also analyzed is the effects of wind-induced error on different types of winter snowfall amounts and the long-term trends over Northeast China. The results are as follows:(1) In the past50years, effect of observation error on the average annual precipitation is large in North China. The measured average annual rainfall is559.4mm, and the error-corrected average annual rainfall increases to607.6mm, with the average absolute error48.3mm and relative error8.1%; Before and after correction the rainfall intensities are6.8mm/d and7.4mm/d respectively.(2) Effects of observation error on different categories of precipitation is significantly different. With the increase of rainfall intensity, the relative error generally decreases. The effect of observation error on light rain is relatively larger, reaching13.7%. The relative effect of observation error is6%for moderate rainfall, and5%for the greater rainfall; Similarly, the effect of observation error on precipitation measurement are different for each of seasons, with the absolute error being larger in summer, and smaller in winter. The relative error, however, is larger in winter, with the region averaged relative error reaching24.5%, and significantly higher than those of other seasons.(3) Significant effect of observation error on the linear trends of annual precipitation amount and intensity in North China during the1960-2009period is found. Due to the changes in precipitation observation error over time, the measured precipitation trend is-10.9mm/10a, and the error-corrected precipitation trend is reduced to-14.2mm/10a, with a difference of-3.3mm/10a, accounting for23.2%of the real precipitation trend. The long-term trend of the measured precipitation intensity is0.18mm/d, and the trend of the error-corrected precipitation intensity becomes0.16mm/d, indicating that the upward trend is slowed.(4) Due to the effect of observation error, the trends of precipitation amounts and precipitation intensity at different levels and in all seasons are over estimated, in spite of the fact that the magnitude of the biases are obviously different. In view of the different categories of precipitation amounts, the relative biases of the trend estimates becomes smaller with the increase of the precipitation intensity, with those of light rain, moderate rain and heavy rain are32.7%,25.0%and16.6%respectively. With regard to the different categories of precipitation intensity, the most obvious effects of observation error on the trend is for moderate precipitation, with the trend is0.02(mm/d)/10a for original data, and it is reduced to-0.03(mm/d)/10a for error-corrected data.(5) The largest effect of the observation error on the precipitation trend is in spring. The trend of the seasonal precipitation is0.25mm/10a for the original data, and it is-0.5mm/10a for the error-corrected data, a change from a weak rise to a weak decline, indicating that measured precipitation trend is significantly overestimated. For summer and autumn, the trend differences between the original and corrected precipitation are relative small, with the relative biases are about15%. The relative bias in winter is86.5%. Therefore, the observation error effect on seasonal precipitation amounts trends is generally larger in spring and winter, and relatively small in summer and autumn.(6) Wind-induced error has a great influence on the winter snowfall measurement in Northeast China. The original observational records of the winter snowfall are generally significantly underestimated. The region averaged measurement of winter snowfall is15.1mm, and the wind-speed error corrected winter snowfall is22.5mm, with an average relative error of34.3%. (7) Due to the weakening of surface wind speed, the wind-induced error decreases over the past50years, and this has a significant effect on the estimates of winter snowfall trend in Northeast China. Region averaged trend of actually observed winter snowfall is0.4mm/10a, and that of the error-corrected winter snowfall decreases to0.1mm/10a.