Detrending Research Articles

Reactive surveillance of deaths involving a fall among the elderly in France, 2015 to March 2023

Abstract In 2022, France launched an elderly anti-fall plan with a goal to reduce fatal falls by 20% within 5 years. In a reactive monitoring approach ensuring comparability over the period, our study aims to develop an algorithm to identify deaths involving a fall using free-text medical causes in death certificates. This will enable proactive surveillance and evaluate plan effectiveness upon completion. We devised an algorithm based on textual search of fall-related expressions in death certificates. Performance metrics were assessed by contrasting algorithm-identified deaths with those flagged by ICD-10 fall-related codes. We compared also temporal dynamics between the algorithms from 2015 to 2021. Analysis was performed by age group, gender and place of death. Additionally, our algorithm enabled estimation of fall-related death trends from 2015 to 2023. The free-text algorithm exhibited robust performance, with a sensitivity and PPV of 95% when compared to deaths coded with an ICD-10 fall-related underlying or associated cause. Temporal trends between both were highly correlated (correlation coefficient=0.99). Instead of the usual 10000 deaths for fall, we revealed about 13000 deaths linked to falls considering both primary and associated causes. Among these, approximately three-quarters were coded as the underlying cause of death in ICD-10 coding. Regarding trends, the results are preliminary and suggested an increase in mortality rates related to falls detected by the algorithm between 2015 and 2022. Our approach distinguishes falls from other competing causes that often relegate them to secondary status in the chain of morbid events leading to death. With free-text medical causes accessible within 3 to 4 months, monitoring deaths involving falls can be done at a detailed geographic level and in a more reactive way. Looking ahead, we must also consider incorporating the impact of epidemics and/or heat waves to accurately evaluate the effectiveness of the anti-fall plan. Key messages • Algorithm using free-text of death certificates identify deaths involving falls. • Our innovative approach isolates falls from other competing causes, elevating their significance in death processes.

The association between county-level premature cardiovascular mortality related to cardio-kidney-metabolic disease and the social determinants of health in the US.

Cardio-kidney-metabolic (CKM) syndrome is defined by the American Heart Association as the intersection between metabolic, renal and cardiovascular disease. Understanding the contemporary estimates of CKM related mortality and recent trends in the US is essential for developing targeted public interventions. We collected state-level and county-level CKM-associated age-adjusted premature cardiovascular mortality (aaCVM) (2010–2019) rates from the CDC Wide-ranging Online Data for Epidemiologic Research (WONDER). We linked the county-level aaCVM with a multi-component social deprivation metric: the Social Deprivation Index (SDI: range 0–100) and grouped them as follows: I: 0–25, II: 26–50, III: 51–75, and IV: 76–100. We conducted pair-wise comparison of aaCVM between SDI groups with the multiplicity adjusted Wilcoxon test; we compared aaCVM in men versus women, metropolitan versus nonmetropolitan counties, and non-hispanic white versus non-hispanic black residents. In 3101 analyzed counties in the US, the median CKM associated aaCVM was 61 [interquartile range (IQR): 45, 82]/100 000. Mississippi (99/100 000) and Minnesota (33/100 000) had the highest and lowest values respectively. CKM associated aaMR increased across SDI groups [I – 45 (IQR: 36, 55)/100 000, II- 61 (IQR: 49, 77)/100 000, III- 77 (IQR: 61, 94)/100 000, IV- 89 (IQR: 70, 110)/100 000; all pair-wise p-values < 0.001]. Men had higher rates [85 (64, 91)/100 000] than women [41 (28, 58)/100 000](p-value < 0.001), metropolitan counties [54 (40, 72)/100 000] had lower rates than non-metropolitan counties [66 (49, 90)/100 000](p-value < 0.001), and non-Hispanic Black [110 (86, 137)/100 000] had higher aaMR than non-Hispanic White residents [59 (44, 78)/100 000](p-value < 0.001). In the US, CKM mortality remains high and disproportionately occurs in more socially deprived counties and non-metropolitan counties. Our inability to reduce CKM mortality rates over the study period highlights the need for targeted policy interventions to curb the ongoing high burden.

Establishing Climate Change on Rainfall Trend, Variation and Change Point Pattern in Umuahia, Nigeria

This study investigates the prevailing conditions of changing climatic trends and change point dates in Umuahia, Nigeria, using daily annual maximum series (AMS) rainfall data spanning 31 years. The Indian Meteorological Department (IMD) was employed to downscale the time series data into durations ranging from 5 minutes to 24 hours. Mann-Kendall (MK) trend and Sen's Slope Estimator (SSE) tests revealed a statistically significant increasing trend across all durations. The Sen's Slope magnitude ranged from 0.1373 mm/year for 5-minute duration to 0.9045 mm/year for 24-hour duration, corresponding to a variation rate of 9.05 mm/decade for the 24-hour duration. Trend change-point analysis, utilizing Distribution-free CUSUM and Sequential Mann-Kendall tests, identified 2002 as the initial rainfall change-point date. These results provide robust evidence of positive changing climatic conditions for rainfall in Umuahia, indicating that non-stationary Intensity-Duration-Frequency (IDF) modelling may be more appropriate for flood risk management and urban planning in this area.

Common Hippopotamus, Hippopotamus Amphibian (Linnaeus, 1758) in the Kouoptamo Subdivision-Cameroon

The Kouoptamo subdivision is one of the localities of the Noun Division that adjoins the Noun River. It served as a study area because of the presence of hippopotamus. This study carried out from January to July 2017, aimed to contribute to the sustainable management of wildlife by improving the observation of hippopotamuses (Hippopotamus amphibius) along the Noun River in the Kouoptamo subdivision. The data were obtained by administering a semi-structured survey form to fishermen, farmers, and local authorities. The results show that the average number of hippopotamuses observed along the River Noun in Kouoptamo is 10. On average, Mefuh records a very abundant potential with 13 individuals. This is followed by Buea and Mangouhnoun (7), Upstream of the dam (5), Makoumahou (4), and Downstream of the dam (3). According to habitat types, megaherbivores are highest in herbaceous savannah (6), followed by forest gallery (5), and finally shrub savannah (3). The months of observation are January and April with the greatest potential (20) and (15), followed by February and June (9) respectively, then March, May, and July (7) respectively. These pachyderms are most frequently observed in the afternoon between 2 p.m. and 6 p.m. (7), followed by the morning between 5 a.m. and 10 a.m. (6), then at midday between 11 a.m. and 1 p.m. (5) and finally at night between 7 p.m. and 4 a.m. (4). Very few studies have been carried out on hippopotamuses in Cameroon, so aspects of their biology and ecology remain unknown; precise estimates of population size and evolutionary trends are still incomplete. However, according to the CBD, knowledge of these resources is essential for better management.

A fast hybrid approach for continuous land cover change monitoring and semantic segmentation using satellite time series

Land cover change detection and classification, including both inter-class changes (land cover conversion, LCC) and intra-class changes (land cover modification, LCM), is critical for understanding the Earth’s dynamic processes and promoting sustainability. However, previous studies have predominantly focused on LCC, with less emphasis on LCM. Land cover classification remains challenging, and its mapping results are often affected by salt and pepper noise. Here, we propose a hybrid approach for continuous change detection and classification of LCC and LCM using Jinchang City, China, as a case study. Firstly, we combined the Continuous Change Detection and Classification (CCDC) and the Bayesian Estimator of Abrupt change, Seasonal change, and Trend (BEAST) algorithms to identify LCC and LCM using all available Landsat time series (TS) data from 2000 to 2020. Then, the harmonic regression coefficients and RMSE values derived from CCDC (hereafter called CCDC features) were fed into the DCNN model for LCC classification. Our findings indicate: (1) For LCC and LCM accuracy assessment, the CCDC and BEAST ensemble achieved a spatial F1 score of 82.7% and an average temporal F1 score of 79.7%. (2) In LCC classification, the DCNN model with CCDC features, particularly DeepLabV3+, outperformed the pixel-based XGBoost and other multi-year land cover products, with frequency-weighted intersection over union (FWIoU), overall accuracy, and Kappa scores of 88.7%, 94%, and 0.87, respectively. (3) Seasonal LCM showed a more concentrated distribution than trend LCM. (4) In Jinchang City, LCM larger than LCC in area, and grassland and cultivated land are the most distributed. Our approach can be contributed to wall-to-wall land surface monitoring and enhance land management capabilities.

Population dynamics of HIV drug resistance during treatment scale-up in Uganda: a population-based longitudinal study.

Clinical studies have reported rising pre-treatment HIV drug resistance during antiretroviral treatment (ART) scale-up in Africa, but representative data are limited. We estimated population-level drug resistance trends during ART expansion in Uganda. We analyzed data from the population-based open Rakai Community Cohort Study conducted at agrarian, trading, and fishing communities in southern Uganda between 2012 and 2019. Consenting participants aged 15-49 were HIV tested and completed questionnaires. Persons living with HIV (PLHIV) provided samples for viral load quantification and virus deep-sequencing. Sequence data were used to predict resistance. Population prevalence of class-specific resistance and resistance-conferring substitutions were estimated using robust log-Poisson regression. Data from 93,622 participant-visits, including 4,702 deep-sequencing measurements, showed that the prevalence of NNRTI resistance among pre-treatment viremic PLHIV doubled between 2012 and 2017 (PR:1.98, 95%CI:1.34-2.91), rising to 9.61% (7.27-12.7%). The overall population prevalence of pre-treatment viremic NNRTI and NRTI resistance among all participants decreased during the same period, reaching 0.25% (0.18% - 0.33%) and 0.05% (0.02% - 0.10%), respectively ( p- values for trend = 0.00015, 0.002), coincident with increasing treatment coverage and viral suppression. By the final survey, population prevalence of resistance contributed by treatment-experienced PLHIV exceeded that from pre-treatment PLHIV, with NNRTI resistance at 0.54% (0.44%-0.66%) and NRTI resistance at 0.42% (0.33% - 0.53%). Overall, NNRTI and NRTI resistance was predominantly attributable to rtK103N and rtM184V. While 10.52% (7.97%-13.87%) and 9.95% (6.41%-15.43%) of viremic pre-treatment and treatment-experienced PLHIV harbored the inT97A mutation, no major dolutegravir resistance mutations were observed. Despite rising NNRTI resistance among pre-treatment PLHIV, overall population prevalence of pre-treatment resistance decreased due to treatment uptake. Most NNRTI and NRTI resistance is now contributed by treatment-experienced PLHIV. The high prevalence of mutations conferring resistance to components of current first-line ART regimens among PLHIV with viremia is potentially concerning. National Institutes of Health and the Gates Foundation. Evidence before the study: We searched PubMed for studies matching the keywords "hiv" "resistance" "longitudinal" "cohort" "population" published since 2004 (the beginning of antiretroviral therapy (ART) availability in sub-Saharan Africa) and identified 50 studies. We excluded 34 studies not based in sub-Saharan Africa, five studies primarily concerned with infection with other pathogens (e.g. HBV, M. tuberculosis ), two studies concerned with insulin resistance, one sequencing-methods paper, and one paper concerned with host susceptibility to HIV infection. The remaining seven studies were not population-based meaning that the study population was not all persons but e.g. people living with HIV enrolled in care at a given clinic. Population-based cohort are essential for monitoring HIV drug resistance in both treated and untreated individuals, including those people who may go undetected in clinical settings, capturing evolutionary dynamics of resistance in real-world conditions. Added value of this study: We estimated the prevalence of drug resistance over five consecutive survey rounds of a population-based open-cohort study in southern Uganda between 2012 and 2019 during a period of intense treatment scale-up. We show that among the entire population regardless of HIV status, 0.8% and 0.5% of individuals harbor viremic resistance to non-nucleoside reverse transcriptase inhibitors (NNRTIs) and nucleoside-reverse transcriptase inhibitors (NRTIs), respectively, of which the majority is dual-class NNRTI/NRTI resistance. Despite a two-fold increase in the prevalence of NNRTI resistance among pre-treatment viremic PLHIV, the overall prevalence of pre-treatment viremic resistance in the entire population decreased by more than 50% due to increased treatment initiation and population viral load suppression. The majority of resistance in recent survey rounds was contributed by treatment-experienced PLHIV. Among treatment-experienced viremic PLHIV, we observe a substantial burden of mutations that confer resistance to the NNRTI and NRTI components of dolutegravir and cabotegravir based regimens e.g. rtM184V (34%) rtY181C (15%), rtG190A (12%), rtK65R (12%), and rtK101E (9.5%). The integrase strand transfer inhibitor (INSTI) resistance mutation inT97A was observed in about a tenth of viremic PLHIV.These results provide the first longitudinal population-based estimates of temporal trends in the prevalence of drug resistance during ART program expansion in a high-burden setting. Further, they provide critical insight into the landscape of prevalent drug resistance substitutions circulating in this population.Implications of all the available evidence: Scale-up of HIV treatment has increased the prevalence of drug resistance mutations among viremic people living with HIV in sub-Saharan Africa. The relatively high prevalence of NNRTI resistance has prompted a recent shift to first-line regimens including dolutegravir (an INSTI) in combination with NRTIs. The high prevalence of mutations conferring resistance to components of current first-line regimens in our population warrants continued monitoring of treatment failures and the prevalence of drug resistance in high burden settings.

Time trends in adolescent depressive symptoms from 2010 to 2019 in Norway: real increase or artifacts of measurements?

Whether the recent rise in adolescent self-reported depressive symptoms is influenced by changing reporting behavior is much debated. Most studies use observed sum scores to document trends but fail to assess whether their measures are invariant across time, a prerequisite for meaningful inferences about change. We examined whether measurement noninvariance, indicative of changing perceptions and reporting of symptoms, may influence the assessment of time trends in adolescent depressive symptoms. Data stem from the nationwide repeated cross-sectional Ungdata-surveys (2010-2019) of 560 712 responses from adolescents aged 13 to 19 years. Depressive symptoms were measured with the Kandel and Davies' six-item Depressive Mood Inventory. Using structural equation modeling, we examined measurement invariance across time, gender and age, and estimated the consequences of noninvariance on cross-cohort time trends. Across most conditions, the instrument was found measurement invariant across time. The few noninvariant parameters detected had negligible impact on trend estimates. From 2014, latent mean depressive symptom scores increased among girls. For boys, a U shaped pattern was detected, whereby an initial decrease in symptoms was followed by an increase from 2016. Larger issues of noninvariance were found across age in girls and between genders. From a measurement perspective, the notion that changed reporting of symptoms has been an important driver of secular trends in depressive symptoms was not supported. Thus, other causes of these trends should be considered. However, noninvariance across age (in girls) and gender highlights that depressive symptoms are not necessarily perceived equivalently from early to late adolescence and across gender.

Changes in Vegetation Greenness and Responses to Land Use Changes in the Yongding River Basin (in North China) from 2002 to 2022

Vegetation is an important component of an ecosystem, fulfilling various ecological functions in areas such as soil and water conservation, climate regulation, and water source maintenance. This study focuses on the Yongding River Basin as a research area. This study used vegetation indices with long time series as a data source in combination with Landsat land use data. This study applied linear trend estimation to analyze the interannual variation trend in vegetation greenness from 2002 to 2022 in the Yongding River Basin and quantitatively analyzed the impact of land use changes on vegetation greenness. The results show that, from 2002 to 2022, the vegetation greenness in the Yongding River Basin has shown an overall increasing trend. The average growth season and the maximum annual normalized difference vegetation index (NDVI) growth rates were 0.006/10a and 0.008/10a, respectively, and the area of increased vegetation greenness accounted for 90% of the total area. During the main growth season (April to October) in the Yongding River Basin, the NDVI generally showed a spatial pattern of being higher in mountainous areas and lower in water areas, with the largest coefficient of variation in vegetation in the river water areas, and the most stable vegetation in forest land. In terms of the changes in vegetation greenness, the contribution rate of arable land was between 36.73% and 38.63%, followed by grassland and forest land, with contribution rates of 26.86% to 27.11% and 23.94% to 26.43%, respectively. The total contribution rate of water areas, construction land, and unused land was around 10.18%. This study can provide a theoretical basis for environmental protection and rational land use in the Yongding River Basin.

Fast and simultaneous prediction of inner quality parameters on intact mangos by near infrared spectroscopy: Impact of spectra pre-processing on prediction accuracy

Near infrared spectroscopy or known as NIRS has been widely employed in many fields including agriculture, especially for sorting and grading of agricultural products. Spectra pre-processing is one of the main factors affecting model accuracy and prediction capabilities of NIRS. The objective of the present study was to study the impact of different spectra corrections namely mean centering (MC), mean normalization (MN), de-trending (DT), multiplicative scatter correction (MSC), standard normal variate (SNV) and orthogonal signal correction (OSC), to the prediction accuracy of quality parameters: titratable acidity (TA) and soluble solids content (SSC) in intact mango. A total of 91 mango samples (cv. Kent) were used as dataset for calibration and external prediction which was separated by means of systematic sampling based on a property (SSBP) approach. Diffuse reflectance spectra (log1/R) were acquired and recorded in wavelength range of 1000 – 2500 nm by Antaris Fourier transform NIR instrument. Judging from calibration and prediction performance, MSC found to be the best spectra pre-processing method prior to prediction model development with R2 prediction are 0.72 for TA and 0.76 for SSC. Although MSC increase the prediction performances based on R2, RMSE, RPD and RER metrics compared to the baseline, the achieved RPD, 1.9 for TA and 1.8 for SSC of this findings are still poor and need improvements to achieve even higher levels of accuracy and reliability necessitates for real-time applications.

Evaluating the effects of nest management on a recovering raptor using integrated population modeling

AbstractEvaluating population responses to management is a crucial component of successful conservation programs. Models predicting population growth under different management scenarios can provide key insights into the efficacy of specific management actions both in reversing population decline and in maintaining recovered populations. Bald eagle (Haliaeetus leucocephalus) conservation in the United States has seen many successes over the last 50 years, yet the extent to which the bald eagle population has recovered in Arizona, an important population within the Southwest region, remains an area of debate. Estimates of the species' population trend and an evaluation of ongoing nest‐level management practices are needed to inform management decisions. We developed a Bayesian integrated population model (IPM) and population viability analysis (PVA) using a 36‐year dataset to assess Arizona bald eagle population dynamics and their underlying demographic rates under current and possible future management practices. We estimated that the population grew from 77 females in 1993 to 180 females in 2022, an average yearly increase of 3%. Breeding sites that had trained personnel (i.e., nestwatchers) stationed at active nests to mitigate human disturbance had a 28% higher reproductive output than nests without this protection. Uncertainty around population trends was high, but scenarios that continued the nestwatcher program were less likely to predict abundance declines than scenarios without nestwatchers. Here, the IPM‐PVA framework provides a useful tool both for estimating the effectiveness of past management actions and for exploring the management needs of a delisted population, highlighting that continued management action may be necessary to maintain population viability even after meeting certain recovery criteria.

Fingerprints of El Niño Southern Oscillation on global and regional oceanic chlorophyll-a timeseries (1997–2022)

With the advancement of present-day ocean color satellites, the spatiotemporal variations of oceanic Chlorophyll-a (Chl-a) concentration, a proxy for phytoplankton population, can be monitored at regional and global scales. Estimating long-term changes in Chl-a concentration, however, is mainly constrained by the limited availability of ocean color data and the significant influence of climate oscillations like the El Niño Southern Oscillation (ENSO). In this study, we investigate the influence of ENSO on regional and global Chl-a timeseries using two ocean color datasets spanning from September 1997 to December 2022. Our analysis found that global Chl-a concentration exhibits a significant increasing trend over the study period. Most of the increasing trends are detected in the Southern Hemisphere (SH) and high-latitudinal regions, including the Southern Ocean. In light-limited regions like the Southern Ocean, enhanced Chl-a concentration shows a strong positive correlation with warming sea surface temperature (SST). Conversely, most oceans in the Northern Hemisphere (NH) exhibit decreasing trends, which are highly correlated with warming SST and ENSO phases. Additionally, we identify distinctive fingerprints of four extreme ENSO events: two extreme El Niño events (1997–1998 and 2015–2016) and two triple La Niña events (1998–2001 and 2020–2022) in these timeseries. Furthermore, empirical orthogonal function (EOF) analysis reveals that the dominant mode of interannual Chl-a variability is associated with an ENSO-like pattern, accounting for about 13 % of the total variability. Intriguingly, this principal mode is highly correlated with the phases of ENSO (R = -0.87) and Pacific Decadal Oscillation (PDO) (R = −0.90) on an interannual scale. This study provides insights into assessing the impacts of ENSO on long-term Chl-a trend estimation.

Improved estimates of child malnutrition trends in Bangladesh using remote-sensed data

This study investigates the trends in chronic malnutrition (stunting) among young children across Bangladesh’s 64 districts and 544 sub-districts from 2000 to 2018. We utilized remote-sensed data–nighttime light intensity to indicate urbanization, and environmental factors like precipitation and vegetation levels–to examine patterns of stunting. Our primary data source was the Bangladesh Demographic and Health Survey, conducted six times within the study period. Using Bayesian multilevel time-series models, we integrated cross-sectional, temporal, and spatial data to estimate stunting rates for years not covered by the direct survey information. This approach, enhanced by remote-sensed data, allowed for greater prediction accuracy by incorporating information from neighboring areas. Our findings show a significant reduction in national stunting rates, from nearly 50% in 2000 to about 30% in 2018. Despite this overall progress, some districts have consistently high levels of stunting, while others show fluctuating levels. Our model gives more precise sub-district estimates than previous methods, which were limited by data gaps. The study highlights Bangladesh’s advancements in reducing child stunting, highlighting the value of integrating remote-sensed data for more precise and credible analysis.

Calibrating estimates of ionospheric long-term change

Abstract. Long-term reduction (∼20 km) in the height of the ionospheric F2 layer, hmF2, is predicted to result from increased levels of tropospheric greenhouse gases. Sufficiently long sequences of ionospheric data exist in order for us to investigate this long-term change, recorded by a global network of ionosondes. However, direct measurements of ionospheric-layer height with these instruments is not possible. As a result, most estimates of hmF2 rely on empirical formulae based on parameters routinely scaled from ionograms. Estimates of trends in hmF2 using these formulae show no global consensus. We present an analysis in which data from the Japanese ionosonde station at Kokubunji were used to estimate monthly median values of hmF2 using an empirical formula. These were then compared with direct measurements of the F2 layer height determined from incoherent-scatter measurements made at the Shigaraki MU Observatory, Japan. Our results reveal that the formula introduces diurnal, seasonal, and long-term biases in the estimates of hmF2 of ≈±10% (±25 km at an altitude of 250 km). These are of similar magnitude to layer height changes anticipated as a result of climate change. The biases in the formula can be explained by changes in thermospheric composition that simultaneously reduce the peak density of the F2 layer and modulate the underlying F1 layer ionization. The presence of an F1 layer is not accounted for in the empirical formula. We demonstrate that, for Kokobunji, the ratios of F2 / E and F2 / F1 critical frequencies are strongly controlled by changes in geomagnetic activity represented by the am index. Changes in thermospheric composition in response to geomagnetic activity have previously been shown to be highly localized. We conclude that localized changes in thermospheric composition modulate the F2 / E and F2 / F1 peak ratios, leading to differences in hmF2 trends. We further conclude that the influence of thermospheric composition on the underlying ionosphere needs to be accounted for in these empirical formulae if they are to be applied to studies of long-term ionospheric change.

Тенденции распределения доходов между трудом и капиталом в странах с низкой долей труда в ВВП

The work is devoted to obtaining quantitative estimates of trends in the distribution of income between labor and capital in countries with a low share of labor in GDP. UN data was used for a set of European countries, post-Soviet countries, Israel, Canada, the USA and Turkey. The lowest labor share levels were observed in Ireland, Kyrgyzstan, Romania and Turkey. To assess trends in the share of labor in GDP on the rate of economic growth, linear econometric models of changes in the share of labor compensation in GDP by year in the period from 2012 to 2021 were built. Hungary, Ireland, Kazakhstan, Kyrgyzstan, Uzbekistan and Ukraine have seen a decline in the share of labor in GDP. In Uzbekistan, this trend is weakly expressed. Germany, Greece, Iceland, Luxembourg, the Czech Republic, Switzerland and Estonia have seen an increase in the labor share of GDP. In Germany and Switzerland this trend is weakly expressed. An increase in the share of labor in GDP is observed in countries such as Azerbaijan, Belarus, Bulgaria, Georgia, Israel, Cyprus, Latvia, Lithuania, Malta, Norway, Poland, the Russian Federation, Romania, Serbia, Slovakia, Turkmenistan and Turkey. Moreover, this trend is weakly expressed in Belarus and Turkey. There are no significant trends in the redistribution of income between labor and capital in countries such as Albania, Malta, North Macedonia, Tajikistan and Montenegro. Trends in the redistribution of income between labor and capital can be determined by institutional conditions in the country’s economy.

Quasi-likelihood analysis for Student-Lévy regression

We consider the quasi-likelihood analysis for a linear regression model driven by a Student-t Lévy process with constant scale and arbitrary degrees of freedom. The model is observed at high frequency over an extending period, under which we can quantify how the sampling frequency affects estimation accuracy. In that setting, joint estimation of trend, scale, and degrees of freedom is a non-trivial problem. The bottleneck is that the Student-t distribution is not closed under convolution, making it difficult to estimate all the parameters fully based on the high-frequency time scale. To efficiently deal with the intricate nature from both theoretical and computational points of view, we propose a two-step quasi-likelihood analysis: first, we make use of the Cauchy quasi-likelihood for estimating the regression-coefficient vector and the scale parameter; then, we construct the sequence of the unit-period cumulative residuals to estimate the remaining degrees of freedom. In particular, using full data in the first step causes a problem stemming from the small-time Cauchy approximation, showing the need for data thinning.

LLMDiff: Diffusion Model Using Frozen LLM Transformers for Precipitation Nowcasting.

Precipitation nowcasting, which involves the short-term, high-resolution prediction of rainfall, plays a crucial role in various real-world applications. In recent years, researchers have increasingly utilized deep learning-based methods in precipitation nowcasting. The exponential growth of spatiotemporal observation data has heightened interest in recent advancements such as denoising diffusion models, which offer appealing prospects due to their inherent probabilistic nature that aligns well with the complexities of weather forecasting. Successful application of diffusion models in rainfall prediction tasks requires relevant conditions and effective utilization to direct the forecasting process of the diffusion model. In this paper, we propose a probabilistic spatiotemporal model for precipitation nowcasting, named LLMDiff. The architecture of LLMDiff includes two networks: a conditional encoder-decoder network and a denoising network. The conditional network provides conditional information to guide the denoising network for high-quality predictions related to real-world earth systems. Additionally, we utilize a frozen transformer block from pre-trained large language models (LLMs) in the denoising network as a universal visual encoder layer, which enables the accurate estimation of motion trend by considering long-term temporal context information and capturing temporal dependencies within the frame sequence. Our experimental results demonstrate that LLMDiff outperforms state-of-the-art models on the SEVIR dataset.

Network traffic prediction based on PSO-LightGBM-TM

Network traffic prediction is critical in wireless network management by allowing a good estimate of the traffic trend, which is also an important approach for detecting traffic anomalies in order to enhance network security. Deep-learning-based method has been widely adopted to predict network traffic matrix (TM) though with the main drawbacks in high complexity and low efficiency. In this paper, we propose a traffic prediction model based on Particle Swarm Optimization (PSO) and LightGBM (PSO-LightGBM-TM), which optimizes the LightGBM parameters for each network flow by PSO so that LightGBM can adapt to each of the network traffic flow. Compared with existing commonly used deep learning models, our model has a more straightforward structure and yet outperforms existing deep learning models. Sufficient comparison tests on three real network traffic datasets, Abilene, GÉANT, and CERNET have been conducted, and the results show that our model provides more accurate results and higher prediction efficiency.

Reconciled estimation of Antarctic ice sheet mass balance and contribution to global sea level change from 1996 to 2021

AbstractThe Antarctic Ice Sheet (AIS) has been losing ice mass and contributing to global sea level rise (GSLR). Given its mass that is enough to cause ∼58 m of GSLR, accurate estimation of mass balance trend is critical for AIS mass loss monitoring and sea level rise forecasting. Here, we present an improved approach to reconciled solutions of mass balance in AIS and its regions from multiple contributing solutions using the input-out, altimetric, and gravimetric methods. In comparison to previous methods, such as IMBIE 2018, this approach utilizes an adaptive data aggregation window to handle the heterogeneity of the contributing solutions, including the number of solutions, temporal distributions, uncertainties, and estimation techniques. We improved the regression-based method by using a two-step procedure that establishes ensembled solutions within each method (input-output, altimetry, or gravimetry) and then estimates the method-independent reconciled solutions. For the first time, 16 contributing solutions from 8 Chinese institutions are used to estimate the reconciled mass balance of AIS and its regions from 1996 to 2021. Our results show that AIS has lost a total ice mass of ∼3213±253 Gt during the period, an equivalent of ∼8.9±0.7 mm of GSLR. There is a sustained mass loss acceleration since 2006, from 88.1±3.6 Gt yr−1 during 1996–2005 to 130.7±8.4 Gt yr−1 during 2006–2013 and further to 157.0±9.0 Gt yr−1 during 2014–2021. The mass loss signal in the West Antarctica and Antarctic Peninsula is dominant and clearly presented in the reconciled estimation and contributing solutions, regardless of estimation methods used and fluctuation of surface mass balance. Uncertainty and challenges remain in mass balance estimation in East Antarctica. This reconciled estimation approach can be extended and applied for improved mass balance estimation in the Greenland Ice Sheet and mountain glacier regions.

Data-Driven Radial Compressor Design Space Mapping

Abstract Estimates of turbomachinery performance trends inform system-level compromises during preliminary design. Existing empirical correlations for efficiency use limited experimental data, while analytical loss models require calibration to yield predictive results. From a set of 3708 radial compressor computations, this paper maps efficiency as a function of mean-line aerodynamics, and determines the governing loss mechanisms. An open-source turbomachinery design code creates annulus and blade geometry, then runs a Reynolds-averaged Navier–Stokes simulation for compressors sampled from the mean-line design space. Polynomial surface fits yield a continuous eight-dimensional representation of the design space for analysis, predicting efficiency with a root-mean-square error of 1.2% points. The results show a balance between surface dissipation in boundary layers and mixing loss due to casing separations sets optimum values for inlet Mach number, hub-to-tip ratio, de Haller number, and backsweep angle. Surface dissipation drives the effect of flow coefficient, with high surface areas at low values, and high velocities at high values. Compact compressor designs are achieved by increasing inlet Mach number, reducing hub-to-tip ratio, and minimizing the radial loading coefficient—all of which reduce efficiency approaching design space boundaries. An interactive web-based tool makes the results available to practising engineers, demonstrating large ensembles of automated designs and simulations as a higher-fidelity replacement for legacy empirical correlations in preliminary design.

Changes in the spatiotemporal distribution of the timing and duration of the soil freeze–thaw status from 1979 to 2018 over the Tibetan Plateau

AbstractChanges in the soil freeze–thaw status will inevitably affect the thermal conditions and properties of the Tibetan Plateau (TP), thereby affecting its upper atmosphere, and further afield in East Asia and even globally. In this study, using the soil temperature simulated by the Community Land Model Version 5.0 (CLM5.0), the timing and duration of the soil freeze–thaw status were divided into freeze start‐date, freeze end‐date and freeze duration. Then, using linear trend estimation, correlation analysis and other methods, the changes in the spatiotemporal distribution of the timing and duration of the soil freeze–thaw status from 1979 to 2018 over the TP were analysed, and the relationships between them and surface temperature, altitude and latitude were analysed. The results obtained were as follows: (1) The soil temperature simulated by CLM5.0 can reasonably reproduce the seasonal changes in multilayer soil temperature, and correlated well with observations. Simulated by CLM5.0 of the time and duration of the soil freeze–thaw status also correlated well with observations. (2) The spatial distribution of the soil freeze–thaw status is characterized by a trend of delayed freezing, advanced thawing and shortened freeze duration from northwest to southeast over the TP. From 1979 to 2018, the freeze start‐date postponed by 7.3 days and became delayed at a rate of 1.9 days per decade, while the freeze end‐date advanced by 6.4 days at a rate of 1.7 days per decade, and the freeze duration shortened by 13.7 days at a rate of 3.6 days per decade. The timing and duration of the soil freeze–thaw status vary across different regions of the TP. The freeze start‐date in all areas of the TP has been delayed in the past 39 years. Except for the subcold zone and arid regions of the TP, the freeze end‐date has occurred earlier and the freeze duration has shortened, with the most significant changes in the subcold zone and humid regions, while the freeze end‐date has advanced at a rate of 3.6 days per decade and the freeze duration has shortened at a rate of 6.3 days per decade. (3) The timing and duration of the soil freeze–thaw status are significantly correlated with surface air temperature, elevation and latitude, exceeding the 99% confidence level. The correlation between the timing and duration of the soil freeze–thaw status and surface temperature is strongest, followed by altitude, and correlation with latitude is weaker. The correlation between surface air temperature and the timing and duration of the soil freeze–thaw status in the western TP is stronger than that in the eastern TP. The rate of change in the soil freeze–thaw status increases with altitude to 3000 m above sea level, while this rate decreases with elevation above 3000 m. The rate of change in the soil freeze–thaw status is greatest at 29°N, while the rate of delay in the freeze start‐date is minimal at 33°N and the rates of advancement in the freeze end‐date and shortening of the freeze duration are minimal at 35°N.