Modeling river water temperature with limiting forcing data: Air2stream v1.0.0, machine learning and multiple regression

Geoscientific model development, Journal Year: 2023, Volume and Issue: 16(14), P. 4083 - 4112

Published: July 20, 2023

Abstract. The prediction of river water temperature is key importance in the field environmental science. Water datasets for low-order rivers are often short supply, leaving modelers with challenge extracting as much information possible from existing datasets. Therefore, identifying a suitable modeling solution large scarcity forcing great importance. In this study, five models, forced meteorological obtained fifth-generation atmospheric reanalysis, ERA5-Land, used to predict 83 (with 98 % missing data): three machine learning algorithms (random forest, artificial neural network and support vector regression), hybrid Air2stream model all available parameterizations multiple regression. hyperparameters were optimized tree-structured Parzen estimator, an oversampling–undersampling technique was generate synthetic training general terms, results study demonstrate vital hyperparameter optimization suggest that, practical perspective, when number predictor variables observed values limited, application models considered crucial. Basically, tested proved be best at least one station. root mean square error (RMSE) Nash–Sutcliffe efficiency (NSE) ensemble 2.75±1.00 0.56±0.48 ∘C, respectively. that performed overall random forest (annual – RMSE: 3.18±1.06 ∘C; NSE: 0.52±0.23). With technique, RMSE reduced 0.00 21.89 (μ=8.57 %; σ=8.21 %) NSE increased 1.1 217.0 (μ=40 σ=63 %). These proposed has potential significantly improve methods, well providing scope its larger other types dependent variables. also revealed existence logarithmic correlation among between predicted watershed time concentration. increases by average 0.1 ∘C 1 h increase concentration (watershed area: μ=106 km2; σ=153).

Language: Английский

---

Comparison of Machine Learning Models in Simulating Glacier Mass Balance: Insights from Maritime and Continental Glaciers in High Mountain Asia

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(6), P. 956 - 956

Published: March 8, 2024

Accurately simulating glacier mass balance (GMB) data is crucial for assessing the impacts of climate change on dynamics. Since physical models often face challenges in comprehensively accounting factors influencing glacial melt and uncertainties inputs, machine learning (ML) offers a viable alternative due to its robust flexibility nonlinear fitting capability. However, effectiveness ML modeling GMB across diverse types within High Mountain Asia has not yet been thoroughly explored. This study addresses this research gap by evaluating used simulation annual glacier-wide data, with specific focus comparing maritime glaciers Niyang River basin continental Manas basin. For purpose, meteorological predictive derived from monthly ERA5-Land datasets, topographical obtained Randolph Glacier Inventory, along target rooted geodetic observations, were employed drive four selective models: random forest model, gradient boosting decision tree (GBDT) deep neural network ordinary least-square linear regression model. The results highlighted that generally exhibit superior performance compared ones. Moreover, among models, GBDT model was found consistently coefficient determination (R2) values 0.72 0.67 root mean squared error (RMSE) 0.21 m w.e. 0.30 river basins, respectively. Furthermore, reveals climatic differentially influence simulations glaciers, providing key insights into dynamics response change. In summary, ML, particularly demonstrates significant potential simulation. application can enhance accuracy modeling, promising approach assess

Language: Английский

---

Comparison of selected surface level ERA5 variables against in‐situ observations in the continental Arctic

Quarterly Journal of the Royal Meteorological Society, Journal Year: 2024, Volume and Issue: 150(761), P. 2123 - 2146

Published: April 1, 2024

Abstract In this study, data from 17 ground‐based, continental Arctic observatories are used to evaluate the performance of European Centre for Medium‐Range Weather Forecasts Reanalysis version 5 (ERA5) reanalysis model. Three aspects evaluated: (i) overall reproducibility variables at all stations seasons one‐hour time resolution; (ii) seasonal performance; and (iii) between different temporal resolutions (one hour one month). Performance is evaluated based on slope, R 2 value, root‐mean‐squared error (RMSE). We focus surface meteorological including 2‐m air temperature (temperature), relative humidity (RH), pressure, wind speed, zonal meridional speed components, short‐wave downward (SWD) radiation flux. The comparison revealed best results pressure (0.98 ± 0.02, mean standard deviation [ σ ]), followed by (0.94 0.02), SWD flux (0.87 0.03) while (0.49 0.12), RH (0.42 0.20), (0.163 0.15) (0.129 0.17) displayed poor results. also found that certain (surface meridional, speed) showed no dependency others (temperature, RH, flux) performed better during seasons. Improved were observed when decreasing resolution month temperature, However, (RH pressure) comparatively worse monthly resolution. Overall, ERA5 performs well in Arctic, but caution needs be taken with which has implications use global climate models. Our useful scientific community as it assesses confidence placed each produced ERA5.

Language: Английский

---

Exploring Recent (1991–2020) Trends of Essential Climate Variables in Greece

Atmosphere, Journal Year: 2024, Volume and Issue: 15(9), P. 1104 - 1104

Published: Sept. 11, 2024

Europe and the Mediterranean are considered climate change hot spots. This is reason why this paper focuses on analysis of trends essential variables in a country, Greece. The analyzed period 1991–2020, dataset used ERA5-Land (produced by European Center for Medium-Range Weather Forecasts), which has global coverage an improved resolution ~9 × 9 km compared to other datasets. Significant climatic changes across Greece have been put evidence during period. More specifically, country averaged 30-year trend temperature +1.5 °C, locally exceeding +2 increasing positively correlated with distance areas from coasts. Accordingly, number frost days decreased throughout country. In terms rainfall, major part experienced annual rainfall amounts, while 86% Greek area positive heavy (>20 mm). Finally, multiple signal consecutive dry was found (statistically non-significant Greece).

Language: Английский

---

Large-sample hydrology – a few camels or a whole caravan?

Hydrology and earth system sciences, Journal Year: 2024, Volume and Issue: 28(17), P. 4219 - 4237

Published: Sept. 12, 2024

Abstract. Large-sample datasets containing hydrometeorological time series and catchment attributes for hundreds of catchments in a country, many them known as “CAMELS” (Catchment Attributes MEteorology Studies), have revolutionized hydrological modelling enabled comparative analyses. The Caravan dataset is compilation several (CAMELS other) large-sample with uniform attribute names data structures. This simplifies hydrology across regions, continents, or the globe. However, use instead original CAMELS other may affect model results conclusions derived thereof. For dataset, meteorological forcing are based on ERA5-Land reanalysis data. Here, we describe differences between precipitation, temperature, potential evapotranspiration (Epot) 1252 CAMELS-US, CAMELS-BR, CAMELS-GB these dataset. Epot unrealistically high catchments, but there are, unsurprisingly, also considerable precipitation We show that from impairs calibration vast majority catchments; i.e. drop performance when using compared to datasets. mainly due Therefore, suggest extending included wherever possible so users can choose which they want at least indicating clearly come quality loss recommended. Moreover, not (and attributes, such aridity index) recommend should be replaced (or on) alternative estimates.

Language: Английский

---

Evaluation of ERA5-Land reanalysis datasets for extreme temperatures in the Qilian Mountains of China

Frontiers in Ecology and Evolution, Journal Year: 2023, Volume and Issue: 11

Published: Feb. 24, 2023

An increase in extreme temperature events could have a significant impact on terrestrial ecosystems. Reanalysis data are an important set for estimation mountainous areas with few meteorological stations. The ability of ERA5-Land reanalysis to capture the index published by Expert Team Climate Change Detection and Indices (ETCCDI) was evaluated using observational from 17 stations Qilian Mountains (QLM) during 1979–2017. results show that can well daily maximum temperature, two warm extremes (TXx TX90p) one cold (FD0) QLM. ERA5-Land’s is best summer worst spring winter. In addition, trends all indices except range (DTR). main bias due difference elevation between ground observation station grid point. simulation accuracy increases decrease difference. provide reference study local data.

Language: Английский

---

Spatiotemporal responses of net primary productivity of alpine ecosystems to flash drought: The Qilian Mountains

Journal of Hydrology, Journal Year: 2023, Volume and Issue: 624, P. 129865 - 129865

Published: June 27, 2023

Language: Английский

---

Using quantile mapping and random forest for bias‐correction of high‐resolution reanalysis precipitation data and CMIP6 climate projections over Iran

International Journal of Climatology, Journal Year: 2024, Volume and Issue: 44(12), P. 4495 - 4514

Published: Aug. 15, 2024

Abstract Climate change is expected to cause important changes in precipitation patterns Iran until the end of 21st century. This study aims at evaluating projections climate over by using five model outputs (including ACCESS‐ESM1‐5, BCC‐CSM2‐MR, CanESM5, CMCC‐ESM2 and MRI‐ESM2‐0) Coupled Model Intercomparison Project phase 6 (CMIP6), performing bias‐correction a novel combination quantile mapping (QM) random forest (RF) between years 2015 2100 under three shared socioeconomics pathways (SSP2‐4.5, SSP3‐7.0 SSP5‐8.5). First, was performed on ERA5‐Land reanalysis data as reference period (1990–2020) QM method, then corrected considered measured data. Based historical simulations (1990–2014), future (2015–2100) were also bias‐corrected utilizing method. Next, accuracy method validated comparing with for overlapping 2020. comparison revealed persistent biases; hence, QM‐RF applied rectify result, highest RMSE both SSP2‐4.5 amounting 331.74 201.84 mm·year −1 , respectively. Particularly, exclusive use displayed substantial errors projecting annual based SSP5‐8.5, notably case ACCESS‐ESM1‐5 (RMSE = 431.39 ), while reduced after (197.75 ). Obviously, significant enhancement results observed upon implementing 139.30 ) 151.43 showcasing approximately reduction values 192.43 50.41 Although each output evaluated individually, multi‐model ensemble (MME) created project pattern Iran. By considering that lower correcting outputs, we used technique create MME. SSP2‐4.5, MME highlight imminent reductions (>10%) across large regions Iran, conversely increases ranging from 10% 20% southern areas SSP3‐7.0. Moreover, projected dramatic declines especially impacting central, eastern, northwest Notably, most pronounced possibly decline are arid (central plateau) eastern SSP5‐8.5.

Language: Английский

---

Modelling point mass balance for the glaciers of the Central European Alps using machine learning techniques

˜The œcryosphere, Journal Year: 2023, Volume and Issue: 17(7), P. 2811 - 2828

Published: July 13, 2023

Abstract. Glacier mass balance is typically estimated using a range of in situ measurements, remote sensing and physical temperature index modelling techniques. With improved data collection access to large datasets, data-driven techniques have recently gained prominence natural processes. The most common used today are linear regression models and, some extent, non-linear machine learning such as artificial neural networks. However, the entire host capabilities has not been applied glacier modelling. This study monthly meteorological from ERA5-Land drive four models: random forest (ensemble tree type), gradient-boosted regressor support vector (kernel networks (neural type). We also use ordinary least squares baseline model against which compare performance models. Further, we assess requirement for each hyperparameter tuning. Finally, importance variable estimation permutation importance. All outperform model. network depicted low bias, suggesting possibility enhanced results event biased input data. ensemble tree-based models, regressor, outperformed all other terms evaluation metrics interpretability variables. best coefficient determination value 0.713 root mean squared error 1.071 m w.e. feature values associated with suggested high variables ablation. line predominantly negative observations. conclude that promising estimating can incorporate information more significant opposed simplified set

Language: Английский

---

Evaluating and Correcting Temperature and Precipitation Grid Products in the Arid Region of Altay, China

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(2), P. 283 - 283

Published: Jan. 10, 2024

Temperature and precipitation are crucial indicators for investigating climate changes, necessitating precise measurements rigorous scientific inquiry. While the Fifth Generation of European Centre Medium-Range Weather Forecasts Atmospheric Reanalysis (ERA5), ERA5 Land Surface (ERA5-Land), China Meteorological Forcing Dataset (CMFD) temperature products widely used worldwide, their suitability Altay region arid semi-arid areas has received limited attention. Here, we as study area, utilizing meteorological station data implementing residual revision method coefficient to rectify inaccuracies in monthly records from ERA5-Land, ERA5, CMFD. We evaluate accuracy these datasets before after correction using bias, Taylor diagrams, root-mean-square error (RMSE) metrics. Additionally, employ Tropical Rainfall Measuring Mission satellite (TRMM) a benchmark assess performance CMFD correction. The results revealed significant differences capture capabilities region. Overall, exhibit substantial errors not directly suitable research. However, applied methods. After this revision, showed significantly improved capabilities, especially ERA5-Land. In terms temperature, post-revision-CMFD (CMFDPR) demonstrated better capabilities. All three weaker mountainous regions compared plains. Notably, post-revision-ERA5 (ERA5PR) seemed unsuitable capturing Concerning rain, CMFDPR, post-revision-ERA5-Land (ERA5-LandPR) ERA5PR outperformed TRMM precipitation. CMFDPR ERA5-LandPR both outperform ERA5PR. summary, effectively compensated sparse distribution stations region, providing reliable support studying change areas.

Language: Английский

---