TY - GEN
AU - Alan Geer
AU - Peter Bauer
AB - Cycle 36r3 will include a complete revision of observation errors, quality control, thinning and resolutionmatching for the all-sky assimilation of microwave imagers. The new approach makes use of the symmetric nature of cloud and rain-affected first guess (FG) departures to predict the observation and background error for quality control purposes. All-sky FG departures are Gaussian when normalised by this model, allowing for the first time an effective quality control for cloud and rain-affected observations. The model is also used to provide observation errors that increase with the mean amount of cloud in model and observation. In the previous approach, observation error was inflated as a function of distance from grid point, but this has been abandoned in the new system. In practice, the increase of FG departure errors with distance is not important over the 20 to 50 km distances involved. The spatial scale of the observations has also been examined. Instead of taking the nearest single all-sky observation to a grid point, we calculate an average or ‘superob' of all observations falling into a grid box, prior to assimilation. Finally, the new approach screens out observations where the model shows ‘cold sector' and ‘heavy snowfall' biases. These biases are too difficult to deal with using a predictor-based bias correction scheme. The new approach gives a substantial increase in the weight of all-sky observations in the analysis, with improved analysis and FG departure fits to radiosonde and dropsonde humidities, microwave sounder humidity channels and infrared sounder lower-tropospheric temperature channels. Short-range own-analysis forecast errors in humidity and vector wind are larger over the tropical oceans, but this is caused by bigger increments, rather than real errors. Vector wind forecast errors against radiosonde do not show the same effect.
BT - ECMWF Technical Memoranda
DA - 04/2010
DO - 10.21957/mi79jebka
LA - eng
M1 - 620
N1 - Also published as EUMETSAT/ECMWF RR20
N2 - Cycle 36r3 will include a complete revision of observation errors, quality control, thinning and resolutionmatching for the all-sky assimilation of microwave imagers. The new approach makes use of the symmetric nature of cloud and rain-affected first guess (FG) departures to predict the observation and background error for quality control purposes. All-sky FG departures are Gaussian when normalised by this model, allowing for the first time an effective quality control for cloud and rain-affected observations. The model is also used to provide observation errors that increase with the mean amount of cloud in model and observation. In the previous approach, observation error was inflated as a function of distance from grid point, but this has been abandoned in the new system. In practice, the increase of FG departure errors with distance is not important over the 20 to 50 km distances involved. The spatial scale of the observations has also been examined. Instead of taking the nearest single all-sky observation to a grid point, we calculate an average or ‘superob' of all observations falling into a grid box, prior to assimilation. Finally, the new approach screens out observations where the model shows ‘cold sector' and ‘heavy snowfall' biases. These biases are too difficult to deal with using a predictor-based bias correction scheme. The new approach gives a substantial increase in the weight of all-sky observations in the analysis, with improved analysis and FG departure fits to radiosonde and dropsonde humidities, microwave sounder humidity channels and infrared sounder lower-tropospheric temperature channels. Short-range own-analysis forecast errors in humidity and vector wind are larger over the tropical oceans, but this is caused by bigger increments, rather than real errors. Vector wind forecast errors against radiosonde do not show the same effect.
PB - ECMWF
PY - 2010
EP - 41
T2 - ECMWF Technical Memoranda
TI - Enhanced use of all-sky microwave observations sensitive to water vapour, cloud and precipitation
UR - https://www.ecmwf.int/node/9518
ER -