The velocity, temperature and humidity in the chamber of the Continuous Flow Streamwise Thermal-Gradient CCN Counter manufactured by DMT were successfully simulated using the computational fluid dynamics software ANSYS FLUENT. With velocity of 0.0164, 0.0185, 0.0205, 0.0226, 0.0246 m/s and temperature settings of ΔT 2, 8 and 17 K, the influence of velocity and temperature on the flow were tested. Results show that, velocity and temperature setting of CCNC influenced both velocity field and temperature field. When ∆T=8 K, v=0.0205 m/s, the supersaturaiton on the centerline was about 0.27%. Supersaturation in the CCNC chamber was simulated successfully.

With micro-pulse lidar (MPL) data, the max mixing height (MMH) in the summer of Naqu is retrieved by gradient method. With radiosonde data (twice daily at 08:00 and 20:00) and max potential temperature of Naqu in the corresponding date, MMH is retrieved by parcel method. The intercomparison between MMHs from the different data shows that the result from 08:00 radiosonde data consists well with MPL results, with correlation coefficient 0.85, root-mean-square error (RMSE) 0.30 km, mean absolute error (MAE) 0.25 km, and the t-test (0.95) passed. But the result from 20:00 radiosonde data has larger deviation from MPL results, with the correlation coefficient 0.84, RMSE 0.67 km, MAE 0.54 km and failed in the t-test. Deviation analysis shows that some factors such as the residual mixing layer in 20:00 soundings, some local weather process before the 20:00 launch time and the temporal-spatial variation of the mixing height induced by the thermal convective bubbles or/and the entrainment process may cause the inconsistency in the results. All above lead to the non-availability of the 20:00 radiosonde data in retrieving the mixing height. The character of the diurnal variation of the potential temperature profile will also affect the accuracy of retrieval result, which would produce a higher height from 08:00 data, and some statistical corrections should be used to improve the result.

A WSD method is presented, using domain keywords and word vector model built from unlabelled data. The effectiveness of the proposed approach is proved, compared with other WSD methods including Lesk on evaluation corpus in environmental domain. Through employing knowledge from different fields, proposed method can be adapted into the WSD task of other domains.

In search services, users can get information more conveniently by reading the succinct answers to their questions. This paper introduces a feature-based method for the query-focused summarization to extract the answer summary of a user query. A convolutional neural network (CNN) is used to learn the semantic representation of a sentence, by which the similarity between a candidate answer sentence and a user query is evaluated. The neural network is trained under the framework of max-margin learning. Experiments in Baidu Knows verify that the proposed method can generate the concise answer of a user query.