No abstract

No abstract

Vegetation restoration has significant effects on soil properties and vegetation cover and thus affects soil detachment by overland flow. Few studies have been conducted to evaluate this effect in the Loess Plateau where a Great Green Project was implemented in the past decade. This study was carried out to quantify the effects of age of abandoned farmland under natural vegetation restoration on soil detachment by overland flow and soil resistance to erosion as reflected by soil erodibility and critical shear stress. The undisturbed soil samples were collected from five abandoned farmlands with natural restoration age varying from 3 to 37 years. The samples were subjected to flow scouring in a 4.0 m long by 0.35 m wide hydraulic flume under six different shear stresses ranging from 5.60 to 18.15 Pa. The results showed that the measured soil detachment capacities in currently cultivated farmland were 24.1 to 35.4 times greater than those of the abandoned farmlands. For the abandoned farmlands, soil detachment capacities fluctuated greatly due to the complex effects of root density and biological crust thickness, and could be simulated well by flow shear stress and biological crust thickness with a power function (NSE = 0.851). Soil erodibility of abandoned farmlands decreased gradually with restoration age and reached a steady stage when restoration age was greater than 28 years. The critical shear stress of the natural abandoned farmlands declined when restoration age was less than 18 years and then increased due to the episodic influences of vegetation recovery and biological crust development. More studies in the Loess Plateau are necessary to quantify the relationship between soil detachment capacity and biological crust thickness for better understanding the mechanism of soil detachment under natural vegetation restoration. Copyright © 2013 John Wiley & Sons, Ltd.

No abstract

All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher. Effects of Near Soil Surface Characteristics on Soil Detachment by Overland Flow in a Natural Succession Grassland Soil & Water Management & Conservation S oil detachment was defined by Govers et al. (1990) as the dislodgment of soil particles from the soil mass at a particular location on the soil surface by the erosive forces of rainfall or surface flow of water. Soil detachment by overland flow, such as rill erosion, is generally considered as the most important process of sediment production and contributes greatly to total soil loss (Poesen et al., 2003). It occurs when the effect of flowing water exceeds the certain threshold of soil resistance (Knapen et al., 2007), which implies that the erosive forces of overland flow and the resistance of the topsoil mass are two factors influencing soil detachment in rills. Overland flow is the external driving force for rill formation and development. Both laboratory and field experiments have found that hydraulic parameters of flowing water were closely related to the process of soil detachment by overland

No abstract

This paper proposes an efficient memory transformer Emformer for low latency streaming speech recognition. In Emformer, the longrange history context is distilled into an augmented memory bank to reduce self-attention's computation complexity. A cache mechanism saves the computation for the key and value in self-attention for the left context. Emformer applies a parallelized block processing in training to support low latency models. We carry out experiments on benchmark LibriSpeech data. Under average latency of 960 ms, Emformer gets WER 2.50% on test-clean and 5.62% on test-other. Comparing with a strong baseline augmented memory transformer (AM-TRF), Emformer gets 4.6 folds training speedup and 18% relative real-time factor (RTF) reduction in decoding with relative WER reduction 17% on test-clean and 9% on test-other. For a low latency scenario with an average latency of 80 ms, Emformer achieves WER 3.01% on test-clean and 7.09% on test-other. Comparing with the LSTM baseline with the same latency and model size, Emformer gets relative WER reduction 9% and 16% on test-clean and testother, respectively.

Transformer-based acoustic modeling has achieved great success for both hybrid and sequence-to-sequence speech recognition. However, it requires access to the full sequence, and the computational cost grows quadratically with respect to the input sequence length. These factors limit its adoption for streaming applications. In this work, we proposed a novel augmented memory self-attention, which attends on a short segment of the input sequence and a bank of memories. The memory bank stores the embedding information for all the processed segments. On the librispeech benchmark, our proposed method outperforms all the existing streamable transformer methods by a large margin and achieved over 15% relative error reduction, compared with the widely used LC-BLSTM baseline. Our findings are also confirmed on some large internal datasets.