Research Of Amdar Data Quality And The Cause Of Amdar Data Errors

AMDAR is the acronym for Aircraft Meteorological Data Relay. In many papers, AMDAR is used to refer to the meteorological observations taken from commercial aircraft. AMDAR programme is worldwide programme in the 1970s. In China, the volumes of AMDAR observations have increased and have been applied to numerical forecasts gradually. The quality of AMDAR observations is attracting a lot of attention because it contributes to the effect of operational forecast. In order to investigate the quality of AMDAR observation and the factors which affect the estimating observation errors for wind and temperature. On the basis of previous work, a more in-depth study of the quality of AMDAR observation, the characteristics of error and the factors which affect the estimating observation errors were studied. In this paper, through the comparison of temperature and wind measurements from AMDAR-equipped aircraft and radiosonde, we study the quality of AMDAR observations. In the same time, we also study the quality of AMDAR observation obtained from different aircraft by data classification. The comparison was made of temperature and wind measurements from AMDAR-equipped aircraft and radiosonde separated by less than 30km in distance, 15 min in time and 30 m in vertical distance: First, we studied the quality of temperature and wind observations from AMDAR-equipped aircraft in total through calculating difference of temperature and wind observations from AMDAR-equipped aircraft and data from radiosonde. Following, we studied the relative systematic errors of a single aircraft, and analyzed the difference of temperature and wind speed observations from one aircraft and another one. Finally, we analyzed the reliability of the statistical results and its scope of application. The results show that: The difference between AMDAR temperature observation and temperature observations from radiosonde is not large, -0.02K for the average value of above temperature difference, for temperature difference rms of 0.98K, which was obtained by AMDAR temperature minus temperature from radiosonde, therefore the temperature observation from radiosonde is slightly higher than the AMDAR temperature observation. The difference between AMDAR wind speed and wind speed from radiosonde is also not large. The average of wind speed deviation is only -0.22m/s, the wind speed rms error is 2.41m/s. The relative systemic errors of temperature and wind speed from AMDAR-equipped aircraft are very small, but the relative systemic errors of temperature and wind speed observation from some aircraft which equipped AMDAR are relatively larger, so the relative systemic error of temperature and wind speed observation from all the aircraft which equipped AMDAR aren't considered to be zero or the same. The Statistical relative systematic errors of temperature and wind speed are reliable to a certain extent, but its application in the strict revised work for AMDAR data is not suitable.A simple method is developed by NOAA in 1999 to estimate individual aircraft rms observation random error. In this method, the rms difference between pairs of observations from different aircraft is calculated. The rms difference includes aircrafts observation errors and the errors contributed by mesoscale variability. For the small separation, the mesoscale variability is taken to be zero and no related error between reports from two different aircraft exits. Following above method, the rms observation error is rms difference divided by square root 2. In this paper, we considered that rms observation random error for temperature and wind speed from different aircraft observing system error is not the same, and we use statistics relative system error to advance the method. This can be more reasonable to calculate the random error for temperature and wind speed. These datasets from different aircraft are separated by less than 10km in horizontal distance and 30m in vertical distance and 10min in time. Only data made on descents and ascents reported were used. The rms difference includes aircrafts observation random errors and the errors contributed by mesoscale variability. For the small separation, the mesoscale variability is taken to be zero and no related error between reports from two different aircraft exits. We calculated random errors for temperature and wind speed observations in each layer to study the difference of random errors for temperature and wind speed observations in different layer. In the same time, we studied how wind speed difference and wind direction difference between aircraft pairs change with the average wind speed and temperature difference versus average temperature. The results showed that an observation error of horizontal wind is 1.39 m/s and for temperature is 0.59 o C.In the bottom of trosphere(pressure more then 900hpa), temperature observation error rms and wind speed observation error rms are larger, with increasing of altitude, temperature observation error rms and wind speed observation error rms gradually decrease. The temperature difference between aircraft pairs is relevant to average temperature. The scope of temperature differences is greater in higher average temperature. The wind speed difference and wind direction difference are relevant to average wind speed especially for wind direction difference. When wind speed is more, the scope of wind direction differences and wind speed difference are small. We expect our results are helpful for the data assimilation.