This study presents an innovative, automated deep learning-based technique for near real-time satellite monitoring of river ice conditions in northern watersheds of the United States and Canada. The method leverages high-resolution imagery from the VIIRS bands onboard the NOAA-20 and NPP satellites and employs the U-Net deep learning algorithm for the semantic segmentation of images under varying cloud and land surface conditions. The system autonomously generates detailed maps delineating classes such as water, land, vegetation, snow, river ice, cloud, and cloud shadow. The verification of system outputs was performed quantitatively by comparing with existing ice extent maps in the northeastern US and New Brunswick, Canada, yielding a Probability of Detection of 0.77 and a False Alarm rate of 0.12, suggesting commendable accuracy. Qualitative assessments were also conducted, corroborating the reliability of the system and underscoring its utility in monitoring hydraulic and hydrological processes across northern watersheds. The system’s proficiency in accurately capturing the phenology of river ice, particularly during onset and breakup times, testifies to its potential as a valuable tool in the realm of river ice monitoring.

This study presents a comprehensive analysis of an extreme dust event recorded in the Arabian Peninsula and the United Arab Emirates (UAE) between 31 March and 3 April 2015. Simulations of the dust event with the Weather Research and Forecasting model coupled with the Chemistry module (WRF-Chem) were analyzed and verified using MSG-SEVIRI imagery and aerosol optical depth (AOD) from the recent 1-km Multi-Angle Implementation of Atmospheric Correction (MAIAC) algorithm for MODIS Terra/Aqua. Data from the National Centers for Atmospheric Prediction/National Center for Atmospheric Research (NCEP/NCAR) and the upper-air radiosonde observations were used to understand the synoptic of the event. In addition, the impact of the event on atmospheric and air quality conditions is investigated. The Air Quality Index (AQI) was calculated prior, during, and after the event to assess the degradation of air quality conditions. Simulated temperature, relative humidity, wind speed, and surface radiation were compared to observations at six monitoring stations in the UAE giving R 2 values of 0.84, 0.63, 0.60, and 0.84, respectively. From 1 to 2 April 2015, both observations and simulations showed an average drop in temperature from 33 to 26 °C and radiance reduction from about 950 to 520 Wm −2 . The AOD modeled by WRF-Chem showed a good correlation with Aerosol Robotic Network (AERONET) measurements in the UAE with R 2 of 0.83. The AQI over the UAE reached hazardous levels during the peak of the dust event before rapidly decreasing to moderate–good air quality levels. This work is the first attempt to demonstrate the potential of using WRF-Chem to estimate AQI over the UAE along with two satellite products (MODIS-MAIAC and MSG-SEVIRI) for dust detection and tracking.

This study investigates an extreme weather event that impacted the United Arab Emirates (UAE) in March 2016, using the Weather Research and Forecasting (WRF) model version 3.7.1 coupled with its hydrological modeling extension package (WRF-Hydro). Six-hourly forecasted forcing records at 0.5∘ spatial resolution, obtained from the National Center for Environmental Prediction (NCEP) Global Forecast System (GFS), are used to drive the three nested downscaling domains of both standalone WRF and coupled WRF–WRF-Hydro configurations for the recent flood-triggering storm. Ground and satellite observations over the UAE are employed to validate the model results. The model performance was assessed using precipitation from the Global Precipitation Measurement (GPM) mission (30 min, 0.1∘ product), soil moisture from the Advanced Microwave Scanning Radiometer 2 (AMSR2; daily, 0.1∘ product) and the NOAA Soil Moisture Operational Products System (SMOPS; 6-hourly, 0.25∘ product), and cloud fraction retrievals from the Moderate Resolution Imaging Spectroradiometer Atmosphere product (MODATM; daily, 5 km product). The Pearson correlation coefficient (PCC), relative bias (rBIAS), and root-mean-square error (RMSE) are used as performance measures. Results show reductions of 24 % and 13 % in RMSE and rBIAS measures, respectively, in precipitation forecasts from the coupled WRF–WRF-Hydro model configuration, when compared to standalone WRF. The coupled system also shows improvements in global radiation forecasts, with reductions of 45 % and 12 % for RMSE and rBIAS, respectively. Moreover, WRF-Hydro was able to simulate the spatial distribution of soil moisture reasonably well across the study domain when compared to AMSR2-derived soil moisture estimates, despite a noticeable dry and wet bias in areas where soil moisture is high and low. Temporal and spatial variabilities of simulated soil moisture compare well to estimates from the NOAA SMOPS product, which indicates the model's capability to simulate surface drainage. Finally, the coupled model showed a shallower planetary boundary layer (PBL) compared to the standalone WRF simulation, which is attributed to the effect of soil moisture feedback. The demonstrated improvement, at the local scale, implies that WRF-Hydro coupling may enhance hydrological and meteorological forecasts in hyper-arid environments.

The forecasting of the downwelling surface solar irradiances is required by the operators of solar powered plants to predict the power output of the plant. The purpose of this study is to use aerosol optical depth and columnar water vapor forecasts by the chemistry transport model CHIMERE, coupled with the Weather Research and Forecast (WRF) model, as inputs in the REST2 model to predict the day-ahead clear-sky global horizontal, direct normal and diffuse horizontal irradiances at hourly steps. A dataset of quality-assured and cloud-screened surface solar irradiance measurements collected from 44 sites scattered throughout Saudi Arabia was used to validate the predicted surface solar irradiances. The validation for the day-ahead forecasting of the clear-sky global horizontal irradiance for the data of all 44 stations combined exhibits a bias (relative to the mean reference value) of 0%, a relative root mean square error of 4% and a correlation coefficient of 0.993. Respectively, the clear-sky direct normal irradiance exhibits values of 0%, 15% and 0.769, while the clear-sky diffuse horizontal irradiance values are 0%, 35% and 0.542. The errors in the inputs are also investigated. The results are satisfactory and pave the way to the inclusion of the attenuation due to clouds with the final aim of predicting the surface solar irradiance under all-sky conditions.

Atmospheric models are known to underestimate land surface temperature and, by association, 2 m air temperature over dry arid regions during the day due to the treatment of the thermal roughness length also known as roughness length of heat. The thermal roughness length can be controlled by the Zilitinkevich parameter, known as Czil, which is a tunable parameter within the models. Three different scenarios with the WRF model are run to test the impact of the Czil parameter on the simulations using two land surface models: the Noah and Noah-MP models. In this study, a modified version of the Noah-MP model is tested, in which the Czil parameter, and, therefore, the thermal roughness length varies depending on the land cover and vegetation height. The model domain is over the United Arab Emirates (UAE) where the major land cover type is desert. The following configurations are tested: the Noah model with Czil = 0.1, Noah model with Czil = 0.5 and the Noah-MP model with Czil = 0.5 over desert. Results of 2 m air temperature are verified against three stations in the UAE. Mean gross error of the diurnal 2 m temperature was reduced by up to 1.48 and 1.54 °C in the 24 and 48 h forecasts, respectively. This reduced the cold bias in the model. This improvement in air temperature showed to improve the diurnal cycle of relative humidity at the three monitoring stations as well as the duration of the sea breeze in some cases.

Dust storms are among the most severe environmental problems in arid and semi-arid regions of the world. The predictability of seven dust events, viz. D1: April 2–4, 2014; D2: February 23–24, 2015; D3: April 1–3, 2015; D4: March 26–28, 2016; D5: August 3–5, 2016; D6: March 13–14, 2017 and D7:March 19–21, 2017, are investigated over the Arabian Peninsula using a regionally adapted chemistry transport model CHIMERE coupled with the Weather Research and Forecast (WRF) model. The hourly forecast products of particulate matter concentrations (PM10) and aerosol optical depths (AOD) are compared against both satellite-based (MSG/SEVRI RGB dust, MODIS Deep Blue Aerosol Optical Depth: DB-AOD, Ozone Monitoring Instrument observed UV Aerosol Absorption Index: OMI-AI) and ground-based (AERONET AOD) remote sensing products. The spatial pattern and the time series of the simulations show good agreement with the observations in terms of the dust intensity as well as the spatiotemporal distribution. The causative mechanisms of these dust events are identified by the concurrent analyses of the meteorological data. From these seven storms, five are associated with synoptic scale meteorological processes, such as prefrontal storms (D1 and D7), postfrontal storms of short (D2), and long (D3) duration types, and a summer shamal storm (D6). However, the storms D4 and D6 are partly associated with mesoscale convective type dust episodes known as haboobs. The socio-economic impacts of the dust events have been assessed by estimating the horizontal visibility, air quality index (AQI), and the dust deposition flux (DDF) from the forecasted dust concentrations. During the extreme dust events, the horizontal visibility drops to near-zero values co-occurred withhazardous levels of AQI and extremely high dust deposition flux (250μgcm−2day−1).

In this study, we intercompare seven different PBL schemes in WRF in the United Arab Emirates (UAE) and we assess their impact on the performance of the simulations. The study covered five fog events reported in 2014 at Abu Dhabi International Airport. The analysis of Synoptic conditions indicated that during all examined events, the UAE was under a high geopotential pressure and light wind that does not exceed 7m/s at 850hPa (~1.5km). Seven PBL schemes, namely, Yonsei University (YSU), Mellor-Yamada-Janjic (MYJ), Moller-Yamada Nakanishi and Niino (MYNN) level 2.5, Quasi-Normal Scale Elimination (QNSE-EDMF), Asymmetric Convective Model (ACM2), Grenier-Bretherton-McCaa (GBM) and MYNN level 3 were tested. In situ observations used in the model's assessment included radiosonde data from the Abu Dhabi International Airport and surface measurements of relative humidity (RH), dew point temperature, wind speed, and temperature profiles. Overall, all the tested PBL schemes showed comparable skills with relatively higher performance with the QNSE scheme. The average RH Root Mean Square Error (RMSE) and BIAS for all PBLs were 15.75% and −9.07%, respectively, whereas the obtained RMSE and BIAS when QNSE was used were 14.65% and −6.3% respectively. Comparable skills were obtained for the rest of the variables. Local PBL schemes showed better performance than non-local schemes. Discrepancies between simulated and observed values were higher at the surface level compared to high altitude values. The sensitivity to lead time showed that best simulation performances were obtained when the lead time varies between 12 and 18h. In addition, the results of the simulations show that better performance is obtained when the starting condition is dry.

A three dimensional chemistry transport model, CHIMERE, was used to simulate the aerosol optical depths (AOD) over the Arabian Peninsula desert with an offline coupling of Weather Research and Forecasting (WRF) model. The simulations were undertaken with: (i) different horizontal and vertical configurations, (ii) new datasets derived for soil/surface properties, and (iii) EDGAR-HTAP anthropogenic emissions inventories. The model performance evaluations were assessed: (i) qualitatively using MODIS (Moderate-Resolution Imaging Spectroradiometer) deep blue (DB) AOD data for the two local dust events of August 6th and 23rd (2013), and (ii) quantitatively using AERONET (Aerosol Robotic Network) AOD observations, CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation) aerosol extinction profiles, and AOD simulations from various forecast models. The model results were observed to be highly sensitive to erodibility and aerodynamic surface roughness length. The use of new datasets on soil erodibility, derived from the MODIS reflectance, and aerodynamic surface roughness length (z0), derived from the ERA-Interim datasets, significantly improved the simulation results. Simulations with the global EDGAR-HTAP anthropogenic emission inventories brought the simulated AOD values closer to the observations. Performance testing of the adapted model for the Arabian Peninsula domain with improved datasets showed good agreement between AERONET AOD measurements and CHIMERE simulations, where the correlation coefficient (R) is 0.6. Higher values of the correlation coefficients and slopes were observed for the dusty periods compared to the non-dusty periods.

Reliable and prompt information on river ice condition and extent is needed to make accurate hydrological forecasts to predict ice jams breakups and issue timely flood warnings. This study presents a technique to detect and monitor river ice using observations from the MODIS instrument onboard the Terra satellite. The technique incorporates a threshold-based decision tree image classification algorithm to process MODIS data and to determine the extent of ice. To differentiate between ice-covered and ice-free pixels within the riverbed, the algorithm combines observations in the visible and near-infrared spectral bands. The developed technique presents the core of the MODIS-based river ice mapping system, which has been developed to support National Oceanic and Atmospheric Administration NWS's operations. The system has been tested over the Susquehanna River in northeastern USA, where ice jam events leading to spring floods are a frequent occurrence. The automated algorithm generates three products: daily ice maps, weekly composite ice maps and running cloud-free composite ice maps. The performance of the system was evaluated over nine winter seasons. The analysis of the derived products has revealed their good agreement with the aerial photography and with in situ observations-based ice charts. The probability of ice detection determined from the comparison of the product with the high-resolution Landsat imagery was equal to 91%. A consistent inverse relationship was found between the river discharge and the ice extent. The correlation between the discharge and the ice extent as determined from the weekly composite product reached 0.75. The developed CREST River Ice Observation System has been implemented at National Oceanic and Atmospheric Administration-Cooperative Remote Sensing Science and Technology Center as an operational Web tool allowing end users and forecasters to assess ice conditions on the river. Copyright (c) 2012 John Wiley & Sons, Ltd.

Tidal harmonic analysis simulations along with simulations spanning four specific historical time periods in 2003 and 2004 were conducted to test the performance of a northern Gulf of Mexico tidal model. A recently developed method for detecting inundated areas based on integrated remotely sensed data (i.e., Radarsat-1, aerial imagery, LiDAR, Landsat 7 ETM+) was applied to assess the performance of the tidal model. The analysis demonstrates the applicability of the method and its agreement with traditional performance assessment techniques such as harmonic resynthesis and water level time series analysis. Based on the flooded/non-flooded coastal areas estimated by the integrated remotely sensed data, the model is able to adequately reproduce the extent of inundation within four sample areas from the coast along the Florida panhandle, correctly identifying areas as wet or dry over 85% of the time. Comparisons of the tidal model inundation to synoptic (point-in-time) inundation areas generated from the remotely sensed data generally agree with the results of the traditional performance assessment techniques. Moreover, this approach is able to illustrate the spatial distribution of model inundation accuracy allowing for targeted refinement of model parameters.

An approach is proposed to estimate soil moisture from Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR‐E) 6.9 GHz passive microwave observations. The approach was evaluated over two watersheds in the Mackenzie River Basin in northwestern Canada as a contribution to the Canadian Global Energy and Water Cycle Experiment (GEWEX) study and the Mackenzie GEWEX Study (MAGS). Based on the sensitivity of the emitted microwave signal to soil roughness and vegetation parameters, a two‐stage method was applied to calibrate a microwave radiative transfer model. Roughness parameters were determined using observations taken under dry conditions. Vegetation parameters were determined using observations taken under wet conditions. Obtained soil roughness and vegetation parameters were then integrated in the radiative transfer model to retrieve soil moisture. The performances of the proposed approach were evaluated against in situ observations, estimates from the NASA soil moisture product (AMSR‐E), model‐based soil moisture estimates from the NARR and gauge‐based precipitation observations. The lowest RMSE of 0.0254 g/cm3 was obtained between the retrieved soil moisture and in situ soil moisture. But, the RMSE between the NARR estimates and in situ soil moisture was 0.055 g/cm3 and between the NASA AMSR‐E product and in situ observation was 0.072 g/cm3. This implies that the proposed approach led to an improvement of 55% and 72% in the obtained RMSE over NARR and NASA AMSR‐E soil moisture, respectively. It is noteworthy that the proposed approach is expandable to larger watersheds and very appropriate for remote regions like the Mackenzie River Basin where information on roughness and vegetation are scarce.

A method is presented to obtain dynamic estimates of basin wetness (BWI) and fractional water surface (FWS) indices at the scale of the Mackenzie River Basin using SSM/l remotely sensed brightness temperature measurements. The approach accounts for the seasonal evolution of the vegetation state and the basin surface heterogeneity. Results demonstrate that the approach can filter out the vegetation effect and produce reasonable estimates of FWS at the basin scale. However, the low resolution of SSM/T and other passive microwave sensors precludes the use of this approach for monitoring soil wetness at a smaller scale. Also, the FWS cannot distinguish moisture effects from open water bodies from that of soil surface. A methodology based on the combined use of passive microwave and visible data, along with topographic information, has been developed to separate the open water from the soil surface component in estimating the BWI, and to downscale the index at the digital elevation models scale using a topographic index (TI) which is continuously adjusted to account for vegetation growth. When applied to the Peace-Athabasca-DeIta area, this method improved the correlation between soil wetness and precipitation measured at a meteorological station, compared to an approach based on a time invariant Tl.

This work proposes a method for detecting inundation between semi‐diurnal low and high water conditions in the northern Gulf of Mexico using high‐resolution satellite imagery. Radarsat 1, Landsat imagery and aerial photography from the Apalachicola region in Florida were used to demonstrate and validate the algorithm. A change detection approach was implemented through the analysis of red, green and blue (RGB) false colour composites image to emphasise differences in high and low tide inundation patterns. To alleviate the effect of inherent speckle in the SAR images, we also applied ancillary optical data. The flood‐prone area for the site was delineated a priori through the determination of lower and higher water contour lines with Landsat images combined with a high‐resolution digital elevation model. This masking technique improved the performance of the proposed algorithm with respect to detection techniques using the entire Radarsat scene. The resulting inundation maps agreed well with historical aerial photography as the probability of detection reached 83%. The combination of SAR data and optical images, when coupled with a high‐resolution digital elevation model, was shown to be useful for inundation mapping and have a great potential for evaluating wetting/drying algorithms of inland and coastal hydrodynamic models. Copyright © 2011 John Wiley & Sons, Ltd.

This work proposes a method for detecting variation in water front between low and high tide conditions in the Gulf of Mexico area using high resolution satellite imagery. Radarsat 1, Landsat images and aerial photography from the Apalachicola region in Florida were used to test the proposed algorithm. A change detection approach was implemented through the analysis of RGB false color composites. In order to alleviate the effect of the inherent speckle in the SAR images ancillary data were used. The flood prone area was a priori delineated through the determination of lower and higher water contour lines with Landsat images and digital elevation model. This technique improved the performance of the proposed algorithm with respect to the detection technique using the entire scene.