Search

Systematic biases and coarse resolutions are major limitations of current precipitation datasets. Many deep learning (DL) based studies have been conducted for precipitation bias correction and downscaling. However, it is still challenging for the current approaches to handle complex features of hourly precipitation, resulting in incapability of reproducing small scale features, such as extreme events. This study developed a customized DL model by incorporating customized loss functions, multitask learning, and physically relevant covariates to bias correct and downscale hourly precipitation data. We designed six scenarios to systematically evaluate the added values of weighted loss functions, multitask learning, and atmospheric covariates compared to the regular DL and statistical approaches. The model was trained and tested using the Modern-Era Retrospective analysis for Research and Applications version 2 (MERRA2) reanalysis and the Stage IV radar observations over northern coastal region of Gulf of Mexico. We found that all the scenarios with weighted loss functions performed notably better than the other scenarios with conventional loss functions and a quantile mappingbased approach at hourly, daily, and monthly time scales as well as extremes. Multitask learning showed improved performance on capturing hourly precipitation climatology, aggregated precipitation at daily and monthly scales, and detailed features of extreme events, while the improvement is not as large as from weighted loss functions. Accounting for atmospheric covariates further improved the model performance for capturing extreme events. We show that the customized DL model can better downscale and bias correct precipitation datasets and provide improved precipitation estimates at fine spatial and temporal resolutions where regular DL and statistical methods experiencing challenges. [ABSTRACT FROM AUTHOR]

Land surface soil moisture (SM) plays a critical role in hydrological processes and terrestrial ecosystems in areas affected by desertification. Passive microwave remote sensing products such as the Soil Moisture Active Passive (SMAP) have been shown to monitor surface soil water well. However, the coarse spatial resolution and lack of full coverage of these products greatly limit their application in areas undergoing desertification. In order to overcome these limitations, a combination of multiple machine learning methods, including multiple linear regression (MLR), support vector regression (SVR), artificial neural networks (ANN), random forest (RF) and extreme gradient boosting (XGB), have been applied to downscale the 36 km SMAP SM products and produce higher spatial- resolution SM data based on related surface variables, such as vegetation index and surface temperature. Areas affected by desertification in Northern China, which are very sensitive to SM, were selected as the study area, and the downscaled SM with a resolution of 1 km on a daily scale from 2015 to 2020 was produced. The results show a good performance compared with in situ observed SM data, with an average unbiased root mean square error value of 0.049 m3/m3. In addition, their time series are also consistent with precipitation and perform better than some common gridded SM products. The data can be used to assess soil drought and provide a reference for reversing desertification in the study area. This dataset is freely available at https://doi.org/10.6084/M9.FIGSHARE.16 430478.V5 (Rao et al., 2021). [ABSTRACT FROM AUTHOR]

Amines and amides are important atmospheric organic-nitrogen compounds but high time resolution, highly sensitive, and simultaneous ambient measurements of these species are rather sparse. Here, we present the development of a high-resolution time-of-flight chemical ionization mass spectrometer (HR-ToF-CIMS) method, utilizing protonated ethanol as reagent ions to simultaneously detect atmospheric gaseous amines (C1 to C6) and amides (C1 to C6). This method possesses sensitivities of 5.6-19.4 Hz pptv-1 for amines and 3.8-38.0 Hz pptv-1 for amides under total reagent ion signals of -0.32 MHz. Meanwhile, the detection limits were 0.10-0.50 pptv for amines and 0.29-1.95 pptv for amides at 3s of the background signal for a 1 min integration time. Controlled characterization in the laboratory indicates that relative humidity has significant influences on the detection of amines and amides, whereas the presence of organics has no obvious effects. Ambient measurements of amines and amides utilizing this method were conducted from 25 July to 25 August 2015 in urban Shanghai, China. While the concentrations of amines ranged from a few parts per trillion by volume to hundreds of parts per trillion by volume, concentrations of amides varied from tens of parts per trillion by volume to a few parts per billion by volume. Among the C1- to C6-amines, the C2-amines were the dominant species with concentrations up to 130 pptv. For amides, the C3-amides (up to 8.7 ppb) were the most abundant species. The diurnal and backward trajectory analysis profiles of amides suggest that in addition to the secondary formation of amides in the atmosphere, industrial emissions could be important sources of amides in urban Shanghai. During the campaign, photo-oxidation of amines and amides might be a main loss pathway for them in daytime, and wet deposition was also an important sink. [ABSTRACT FROM AUTHOR]