Performance Evaluation of IMERG GPM Products During Tropical Storm Imelda

atmosphere

Article Performance Evaluation of IMERG GPM Products during Tropical Storm Imelda

Salman Sakib , Dawit Ghebreyesus and Hatim O. Sharif *

Department of Civil & Environmental Engineering, The University of Texas at San Antonio, One UTSA Circle, San Antonio, TX 78249, USA; [email protected] (S.S.); [email protected] (D.G.) * Correspondence: [email protected]; Tel.: +1-210-458-6478

Abstract: Tropical Storm Imelda struck the southeast coastal regions of Texas from 17–19 September, 2019, and delivered precipitation above 500 mm over about 6000 km2. The performance of the three IMERG (Early-, Late-, and Final-run) GPM satellite-based precipitation products was evaluated against Stage-IV radar precipitation estimates. Basic and probabilistic statistical metrics, such as CC, RSME, RBIAS, POD, FAR, CSI, and PSS were employed to assess the performance of the IMERG products. The products captured the event adequately, with a fairly high POD value of 0.9. The best product (Early-run) showed an average correlation coefﬁcient of 0.60. The algorithm used to produce the Final-run improved the quality of the data by removing systematic errors that occurred in the near-real-time products. Less than 5 mm RMSE error was experienced in over three-quarters (ranging from 73% to 76%) of the area by all three IMERG products in estimating the Tropical Storm Imelda. The Early-run product showed a much better RBIAS relatively to the Final-run product. The overall performance was poor, as areas with an acceptable range of RBIAS (i.e., between −10% and 10%) in all the three IMERG products were only 16% to 17% of the total area. Overall, the Early-run product was found to be better than Late- and Final-run. Citation: Sakib, S.; Ghebreyesus, D.; Sharif, H.O. Performance Evaluation Keywords: satellite data; GPM; IMERG; radar; Imelda; rainfall estimates; precipitation; remote of IMERG GPM Products during sensing; performance evaluation Tropical Storm Imelda. Atmosphere 2021, 12, 687. https://doi.org/ 10.3390/atmos12060687

1. Introduction Academic Editors: Panagiotis Nastos and Elissavet G. Feloni The U.S. has faced around 258 weather and climate disasters, each of which reached an overall damage exceeding a billion dollars, since 1980. The National Centers for En- Received: 5 May 2021 vironmental Information (NCEI), a subsidiary of the National Oceanic and Atmospheric Accepted: 24 May 2021 Association (NOAA), records weather and climate events with significant economic and Published: 27 May 2021 societal impacts in the United States [1]. The total cost of all of these disasters exceeded an estimated damage cost of 1.75 trillion dollars. During 2019 alone, the U.S. experienced Publisher’s Note: MDPI stays neutral 14 separate billion-dollar disasters, and 2019 was marked as the fifth consecutive year with regard to jurisdictional claims in (2015–2019) in recorded history with 10 or more individual billion-dollar disasters [1]. The published maps and institutional affil- natural disasters in 2019 included: three major inland floods, eight severe storms, two iations. tropical cyclones or depressions (Dorian and Imelda), and one major wildfire event; totaling an overall damage cost exceeding 14 billion dollars [1]. The state of Texas currently leads the nation in the total number of billion-dollar disasters since 1980, with a total estimated damage cost of around 250 billion dollars [2]. In recent history, Texas has experienced Copyright: © 2021 by the authors. some of the nation’s deadliest and most expensive natural disasters, such as the Memorial Licensee MDPI, Basel, Switzerland. Day Flood (2015), Tax Day Flood (2016), Hurricane Harvey (2017), The Great June Flood This article is an open access article (2018), and Tropical Storm Imelda (2019) [3]. Major drivers of these natural disasters were distributed under the terms and extreme precipitation and heavy downpour, which have left scientists and researchers conditions of the Creative Commons looking for solutions to better model and predict such extreme scenarios. About three Attribution (CC BY) license (https:// billion or nearly half of the world’s population lives in coastal areas, and in the United creativecommons.org/licenses/by/ States the southern coastal states like Louisiana, Florida, and Texas face recurring extreme 4.0/).

Atmosphere 2021, 12, 687. https://doi.org/10.3390/atmos12060687 https://www.mdpi.com/journal/atmosphere Atmosphere 2021, 12, 687 2 of 18

hydrological events [4–6]. In the future, Texas will likely face similar extreme natural disasters, and the need for accurate precipitation estimates is key to better understand and model these hydrological scenarios. The National Hurricane Center started monitoring an area of disturbed weather for possible tropical development near the west coast of Florida on 14 September 2019 [7]. The area of disturbed weather had moved westward by 16 September, reaching the coast of Southeast Texas and Southeast Louisiana. Before daybreak on 17 September, the disturbed weather area underwent significant organization and was being observed by the GOES satellite and the Houston, TX and Lake Charles, and LA WSR-88D radars. By midday, the weather area strengthened, and soon after, it was classified as Tropical Storm Imelda by the National Hurricane Center (NHC) tropical cyclone update. Around 1:30 PM on 17 September 2019, Tropical Storm Imelda made landfall near Freeport, Texas, and was weakened to a tropical depression by the evening. It moved inland shortly after and hovered between Houston and Lufkin in Texas throughout 19 September. During the stalling of Imelda over Southeast Texas, precipitation events caused significant flooding in the Houston and Galveston areas. Along the I-10 corridor from Winnie eastward to Fannett, Beaumont, Vidor, and Orange Texas faced devastating flooding in many areas. Tropical Storm Imelda was the fifth wettest tropical cyclone in the history of the continental United States and the fourth wettest in the state of Texas [7]. It inflicted damage worth over 5 billion dollars [7]. To analyze the impacts of these severe natural disasters, scientists have been trying to evaluate different precipitation estimate tools in terms of their performance and accuracy. Precipitation modeling is an essential part of weather prediction, hydrological analysis, climate modeling, and agricultural applications [8–10]. Accurate precipitation predictions are crucial for the analysis of impacts caused by natural disasters such as hurricanes, tropical storms, and tornadoes [11]. Obtaining reliable and accurate precipitation data has been a challenge for scientists and users, even though precipitation data is vital for many environmental and hydrological analyses [12,13]. To verify and enhance predictions, precipitation measurements are generally obtained through rain gauges, weather radars, and satellite sensors [14,15]. All of these data sources provide measurements with specific spatiotemporal resolutions and are distinguished by their instrumental and environmental uncertainty. The difference in the spatiotemporal resolutions of the data introduces a novel challenge for the comparative evaluation of the measurements in hydrological modeling and analysis studies [16]. Historically, rain gauges have been the main source of precipitation data with vari- able temporal frequencies at ground level. Rain gauge measurements are considered to be ground-truth observations at the point level in comparison to radar and satellite sensors [17]. Conversely, the sparse distribution of rain gauges makes it harder to interpret the actual precipitation for coverage over larger areas [18,19]. Spatial interpolation of the gauge data in mountainous regions is more challenging, considering the geography of the landscape [20]. The use of ground-based radar allows users to have a higher spatiotemporal resolution of the precipitation data and increases the understanding of storm structures and precipitation estimates. However, the acquisition of radar data is sometimes limited in mountainous regions, as the terrain complexity and radar beam obstruction effects distort the measurements [21–28]. Moreover, the construction and maintenance of radar systems in difficult terrain is expensive and unfeasible for almost every country [29]. Combining the strengths and weaknesses of both measurement systems, methods such as Bayesian combination or conditional merging, ordinary kriging correction, and mean bias correction have been developed for better precipitation estimates [30–33]. The Global Precipitation Measurement (GPM) mission, a successor of the Tropical Rainfall Measuring Mission (TRMM), has been providing precipitation estimates since 2014 [29]. Precipitation estimates using the GPM satellite constellation have major advan- tages over the other previous satellite precipitation products in terms of global spatial coverage and temporal and spatial resolution [29]. After the introduction of the Inte- Atmosphere 2021, 12, 687 3 of 18

grated Multi-Satellite Retrievals for the Global Precipitation Mission (IMERG) products in 2014, the use of satellite-based precipitation estimates in hydrologic applications has seen great interest from researchers. Researchers have validated the IMERG products using ground-based observations (radar and rain gauges) and found an overall improvement in precipitation estimates over previous similar products, such as TRMM, PERSIANN, CMORPH, GSMAP, and MSWEP [34,35]. Stage-IV radar data have been widely used as a validation dataset to evaluate the performance of the IMERG products, as researchers aim to improve the algorithm of the IMERG products [36–39]. Data acquisition from the satellite-based measurements can be almost real-time, with only a 4-hr latency between capture and acquisition and 3.5 months latency for the availability of post-real-time research data. IMERG GPM precipitation product estimates overcome the limitations of both gauge and radar systems by warranting real-time data and global spatial coverage [29]. These products are less affected by localized weather or terrain conditions during weather events such as storms, tornadoes, and flooding, and by other technical issues, than measurements obtained with ground-based radar instruments. Although these products provide precipitation estimates with near-global spatial coverage, especially valuable for mountainous regions, the spatial resolution is still rather coarse [40–43]. Hence, many researchers have tried to evaluate and validate the performance of IMERG GPM precipitation estimates in different hydrological scenarios and applications [29]. Detailed analysis of the IMERG GPM precipitation estimates and quantification of the errors associated with these products will contribute towards both hydrology and climatology applications if IMERG is used as reference data for poorly instrumented or remote regions [44]. Although IMERG products provide almost global spatial coverage and high temporal resolution, earlier studies have shown unsatisfactory performance of their estimates in Northern China and the Indian subcontinent based on seasonal variations [45–47]. The inconsistency of the IMERG products’ performance over various regions can be attributed to various sources of errors, which were investigated in multiple error modeling studies [48–52]. Owing to their limitations in terms of spatial resolution and performance inconsistencies, the need remains for the evaluation of IMERG GPM products at regional scales and under major precipitation and flood events, such as hurricanes and tropical storms [53]. Accurate precipitation estimates of major natural disasters are immensely important for many hydrological applications. Better prediction and modeling of the hydrological events will provide key insights for environmental design and will offer opportuni- ties to influence policy and planning for resilient infrastructure networks during crisis events [54–59]. Therefore, researchers need to assess the performance of the IMERG GPM products during severe natural disasters and evaluate their potential for filling crucial precipitation measurement gaps [40,47,48,60,61]. The objective of this study was to evaluate the performance of the IMERG GPM products at a local scale in southeast Texas, which is vulnerable to hurricane and storm events. The study also focused on assessing the performance accuracy of GPM satellite-based precipitation products during the Imelda Tropical Storm that made its landfall on 17 September 2019. Section2 provides a detailed description of the study area, the datasets used, and the methodology, while Section3 elaborates the results and relevant discussion on the findings. Section4 represents the study conclusions.

2. Materials and Methods 2.1. Study Area The current study area focuses on the southeast region of the state of Texas, the largest state in the contiguous United States, with three climatic regions. The regions differ from each other based on hydrometeorological conditions, topography and land cover. The western region is typically arid and dry, the central and eastern regions are under humid climatic conditions, and the southeastern region is predominantly wet, with subtropical weather conditions [15,62,63]. The southeast region is one of the most vulnerable regions Atmosphere 2021, 12, 687Atmosphere 2021, 12, x FOR PEER REVIEW 4 of 18 4 of 18

in the United States to extreme precipitation events such as hurricanes, tropical storms, and flash-floods.regions Tropical in the Storm United Imelda States struck to extreme the southeast precipitation coastal events regions such of as Texas hurricanes, and tropical caused massivestorms, flooding and in flash the-floods. Houston Tropical and Galveston Storm Imelda areas stru extendingck the southeast along the coastal I-10 regions of corridor from WinnieTexas and eastward caused to massive Fannett, flooding Beaumont, in the Vidor, Houston and and Orange Galveston Texas. areas It was extending the along the I-10 corridor from Winnie eastward to Fannett, Beaumont, Vidor, and Orange Texas. second most expensive natural disaster in the contiguous United States in 2019 costing a It was the second most expensive natural disaster in the contiguous United States in 2019 total of 3.5 billion. Our current study area ranges from 93.4◦ to 106.7◦ West in longitude and costing a total of 3.5 billion. Our current study area ranges from 93.4° to 106.7° West in 25.7◦ to 36.6◦ North in latitude. Within this study area, data from radar and GPM IMERG longitude and 25.7° to 36.6° North in latitude. Within this study area, data from radar and products were used for precipitation estimate analysis. Figure1 shows the path tracking of GPM IMERG products were used for precipitation estimate analysis. Figure 1 shows the the storm as it made landfall in Freeport, TX on 17 September 2019, and continued inward path tracking of the storm as it made landfall in Freeport, TX on 17 September 2019, and in the latter days. continued inward in the latter days.

Figure 1. Path track of the Tropical Storm Imelda before and after its landfall in Freeport, TX at 17:45–18:00 UTC 17 Sep- Figure 1. Path track of the Tropical Storm Imelda before and after its landfall in Freeport, TX tember 2019 (green line shows the path of the Tropical Strom Imelda). Precipitation estimates are based on radar data. at 17:45–18:00 UTC 17 September 2019 (green line shows the path of the Tropical Strom Imelda). Precipitation estimates2.2. Precipitation are based on Data radar data. 2.2. Precipitation DataThe IMERG GPM data was downloaded from the Precipitation Measurement Mis- The IMERGsions GPM website data was that downloaded uses the algorithm from the version Precipitation V06B Measurement (http://pmm.nasa.gov/data Mis- -ac- sions website thatcess/downloads/gpm/ uses the algorithm accessed version on V06B 22 M (http://pmm.nasa.gov/data-access/arch 2020). The IMERG is the unified U.S. algorithm that provides the multi-satellite precipitation product for the U.S. GPM team. The downloads/gpm/ accessed on 22 March 2020). The IMERG is the uniﬁed U.S. algorithm IMERG GPM is available in three products, namely Early, Late, and Final. All of the that provides the multi-satellite precipitation product for the U.S. GPM team. The IMERG IMERG products have a temporal resolution of one hour and a spatial resolution of 0.1° × GPM is available in three products, namely Early, Late, and Final. All of the IMERG 0.1°. The Early-run product is produced with the first run of the IMERG algorithm, which products have a temporal resolution of one hour and a spatial resolution of 0.1◦ × 0.1◦. has a latency of about 4 h from the observation time. Another product is produced when The Early-run product is produced with the ﬁrst run of the IMERG algorithm, which has the algorithm is run for the second time, with more information from the satellites at a a latency of about 4 h from the observation time. Another product is produced when the algorithm is run for the second time, with more information from the satellites at a latency

Atmosphere 2021, 12, 687 5 of 18

of about 14 h. This product is called Late-run. The Final-run product is developed when the monthly gauge analysis data is obtained, and it has a latency period of about 3.5 months from the observation time. The baseline for the post-real-time Final-run half-hour estimates is calibrated so that they sum to the monthly satellite-gauge Final-run combination. This product is developed mainly for research purposes. The Final-run product is adjusted with a gauge-network which converts it to a more accurate product relative to the Early- and Late-run products. Since the radar products have an hourly temporal resolution, the IMERG products used in this study were also converted to an hourly format at the end of each hour. The National Weather Service/National Centers for Environmental Prediction (NWS/ NCEP) Stage-IV Quantitative Precipitation Estimates (QPEs) gridded radar-rainfall (http: //data.eol.ucar.edu/ accessed on 22 March 2020) serves as the source of radar precipitation measurements. The operational Stage-IV precipitation radar data production started its journey in 2001 at NWS River Forecast Centers (RFCs) and has been continued since then [64]. It contains bias-adjusted, quality-controlled hourly rain-gauge precipitation records with a 4 × 4-km spatial resolution. The measurements are in a cumulative hourly format at the end of each hour.

2.3. Methodology Tropical Storm Imelda predominantly affected the southeastern counties of Texas. To better track the devastation of the event, this study focuses on the precipitation downpour around the Houston and Galveston areas. IMERG products at Early-, Late-, and Final-run stages of the event were compared to the NCEP Stage-IV gridded radar product at an hourly temporal resolution for eight consecutive days, starting from 14 September and ending on 21 September, during Tropical Storm Imelda. Three days before and after landfall (17 September 2019) were included in the analysis to reduce the impact of the presence of the extreme storm event on the probabilistic statistical indices. Euclidean distance was used to identify the nearest radar grid to each satellite grid. For each grid of the satellite product, a nearest radar grid was found using the advantage of the satellite product’s spatial spacing being three times higher than that of the radar grids. Subsequently, statistical metrics were calculated to compare the precipitation estimates of the IMERG grids and their nearest radar grids at an hourly temporal resolution. There are 40,434 NWS/NCEP radar grids versus 6577 IMERG satellite grids in Texas. However, only 2570 grids of Stage-IV radar data and 418 IMERG grids were used that cover the analysis area spanning across 17 counties located in southeast Texas.

2.4. Statistical Indices The IMERG products were evaluated using two types of statistical comparative metrics to assess their performance (Table1), using the radar product as reference. The basic statistical indices included Pearson’s correlation coefﬁcient (CC), relative bias (RBIAS), and root mean squared error (RMSE). These statistical parameters were calculated to examine the consistency of the data in contrast to the precipitation radar data. The probabilistic statistical indices: probability of (rainfall) detection (POD), false alarm ratio (FAR), critical success index (CSI), and Peirce skill score (PSS) were calculated to obtain information about the probability and accuracy of precipitation detection by the IMERG products compared to the reference radar products. Probabilistic statistical indices are widely used to assess the probabilistic quality of satellite products in this category [9,15]. The probability of rainfall detection by IMERG satellites during the precipitation reporting of the radar data is evaluated with POD. The FAR index provided information on the inconsistencies between the IMERG satellites falsely detecting rainfall while the radar products were without detection. CSI index aggregates all reports captured by both satellite and radar. The CSI index provides more harmony over the precipitation reporting of both radar and satellite products over the study area, while also providing more insight into the satellite data in various spatiotemporal resolutions [15]. The system’s overall accuracy Atmosphere 2021, 12, 687 6 of 18

in capturing rainfall events was indicated by PSS, which is the difference between the probability of detection and the probability of false detection. The Kling–Gupta model efficiency coefficient is another powerful statistical metric that measures the goodness-of-fit of a model. Gupta et al. [65] developed this method as an alternative to the mean squared error (MSE) and Nash–Sutcliffe efficiency (NSE) in the context of hydrological modeling. This coefficient has three main components: correlation, bias, and variability. In this study, the updated version, i.e., Kling et al. [66] was used to ensure the variability and bias components were not cross-related. This means the coefficient of variation was used as a measure of variability instead of standard deviation. Table1 lists all the parameters with their formulae, ranges, and perfect values.

Table 1. Statistical performance indicators used for evaluation of the satellite products.

Category Index Formula Range Perfect Value

1 N Correlation N ∑n=1(Satn−Sat)(Radn−Rad) 2 (−1) − (+1) 1 Coefficient (CC) (SDSat)(SDRad) N ∑n=1(Satn−Radn) Relative Bias (RBIAS) N −∞/ + ∞ 0 ∑n=1 Radn Root Mean Square q Basic Statistical 1 N 2 − + ∑n=1(Satn − Radn) 0 ∞ 0 Indices 1 Error (RMSE) N Kling–Gupta model q 2 2 2 efficiency coefficient 1 − [sr × (r − 1)] + sβ × (β − 1) + [sγ × (γ − 1)] (−∞) − (+1) 1 (KGE) Probability of CSR 0 − 1 1 detection (POD) CSR+CR MS False Alarm Ratio CS MR 0 − 1 0 Probabilistic (FAR) CSR+CS MR Statistical Critical Success Index 1 CSR 0 − 1 1 Indices (CSI) CSR+CS MR+CR MS Peirce Skill Score (PSS) CSR − CS MR (−1) − (+1) 1 CSR+CR MS MSR+CS MR 1 n: Sample size; Satn: IMERG rainfall estimate; Radn: Radar rainfall estimate; Sat: IMERG average rainfall estimate; Rad: Radar average rainfall estimate; SDSat: Standard deviation of IMERG rainfall estimates; SDRat: Standard deviation of radar rainfall estimates; CSR: Number of detected rainfall hours recorded by GPM satellite and radar; CRMS: Number of detected rainfall hours recorded by radar but absent in GPM satellite estimates; CSMR: Number of detected rainfall hours recorded by GPM satellite and absent in radar estimates; MSR: Number of rainfall hours missed by both IMERG and radar products (no estimate). 2 In this study, the negative correlation coefficients were also examined to better understand the discrepancies.

3. Results and Discussion 3.1. Exploratory Data Analysis and Visualization Figure2 depicts the spatial distribution of the cumulative precipitation captured by both the radar and GPM IMERG products during Tropical Storm Imelda. Cumulative rainfall estimates show variability in spatial patterns between the Stage-IV gridded product and the three IMERG satellite products for the same grids during the study period (Figure2) . Both the Early- and Late-run IMERG products showed similar spatial patterns in the coastal and inland areas in the southeast coastal region receiving a high amount of precipitation during the study period (7 days). The products showed the area of highly hit areas (areas receiving more than 250 mm) as ~15,000 km2 and ~14,000 km2 for Early- and Late-run respectively. The Final-run product showed a different pattern, with a more expanded distribution of the tropical storm, with a ~21,500 km2 area getting more than 250 mm during the seven-day event (Figure2C). The Stage-IV radar product (Figure2D) showed a different spatial distribution in the cumulative rainfall estimation, showing two hot spots along the southeast coastal region. However, when considering the extreme peak that is about the limit of the NWS rain gauges (~500 mm), the pattern reverses. The Early- and Late-run reported areas about 3700 km2 and 3500 km2 to have received precipitation more than 500 mm, respectively. While, the Final-run showed only a 900 km2 area with precipitation higher than 500 mm. The most extreme rainfall area was observed on the Stage-IV radar product, with almost 6000 km2 getting precipitation higher than 500 mm. Atmosphere 2021, 12, x FOR PEER REVIEW 7 of 18

Early- and Late-run reported areas about 3700 km2 and 3500 km2 to have received precipitation more than 500 mm, respectively. While, the Final-run showed only a 900 km2 area with precipitation higher than 500 mm. The most extreme rainfall area was observed on the Atmosphere 2021, 12, 687 Stage-IV radar product, with almost 6000 km2 getting precipitation higher than7 of 18 500 mm.

Figure 2. The spatialFigure distribution 2. The spatial of the distribution cumulative of precipitation the cumulative estimation precipitation (≥10 mm) estimation captured (≥ by10 IMERG mm) captured (A) Early-, (B) Late-, (C) Final-run products, and (D) Stage-IV products during Tropical Storm Imelda from 14 to 21 September 2019. by IMERG (A) Early-, (B) Late-, (C) Final-run products, and (D) Stage-IV products during Tropical Storm Imelda from 14 to 21 September 2019. When it comes to the maximum recorded grid totals by the products, all three IMERG Whenproducts it comes toreported the maximum close enough recorded totals, grid totalsranging by from the products, 740 mm allto three775 mm. IMERG However, the products reportedmaximum close precipitation enough totals, from ranging the radar from product 740 mm was to about 775 mm.1.5 times However, the IMERG the product maximumat precipitation the peak grid from reported the radar value product of 1150 was mm. about The 1.5 maps times shown the IMERG in Figure product 2 show that the at the peakIMERG grid reported satellite value product of 1150 generally mm. Theoverestimated maps shown the inprecipitation Figure2 show in the that places the which re- IMERG satelliteceived product lower precipitation, generally overestimated but underestimated the precipitation the storm in places the places where which there was very received lowerhigh precipitation, precipitation. but The underestimated satellite also didthe storm not detect in places precipitation where there in was some very coastal areas high precipitation.where radar The products satellite recorded also did values. not detect All precipitationproducts successfully in some recoded coastal areasthe center of the where radarstorm products near recordedthe southeast values. side All of products the Houston successfully area. Overall, recoded the the radar center product of the estimates storm nearwere the southeast more concentrated side of the over Houston the storm area. center Overall, and thesatellite radar products product covered estimates a larger spa- were more concentratedtial extent of the over southeast the storm Texas center region and satellite (Houston, products Galveston, covered and a larger Beaumont spatial areas). extent of the southeast Texas region (Houston, Galveston, and Beaumont areas). 3.2. Temporal Evolution of the Tropical Storm 3.2. Temporal Evolution of the Tropical Storm Figure 3 illustrates the temporal evolution (17–20 September 2019) of the storm after Figureits3 illustrates landfall and the the temporal spatial patterns evolution captured (17–20 September by IMERG 2019) Final- ofrun the product storm after(Figure 3A) and its landfallStage and- theIV (Figure spatial patterns3B). On the captured day of bythe IMERG landfall Final-run (17 September product), the (Figure Final-3runA) and Stage- and Stage-IVIV showed (Figure3 aB). similar On the pattern. day of On the the landfall next day (17 ( September),18 September the), the Final-run two products and reported Stage-IV showeddifferent a spatial similar distributions pattern. On of the the next storm. day The (18 satellite September), product the missed two products the hotspot in the reported differentsouthern spatial part of distributions the region, as of shown the storm. in Figure The 3. satellite On the productday of the missed heavy the precipitation hotspot inevent the southern (19 Septe partmber of), the the region, satellite as-based shown product in Figure failed3. On to thecapture day ofthe the intensity heavy of the trop- precipitationical event storm, (19 with September), a relatively the satellite-basedsmaller area (4200 product km2 failed) getting to capture more than the intensity250 mm. However, 2 of the tropicalthe radar storm, product with a relativelyestimated smaller almost areatwice (4200 (7700 km km)2) getting the area more reported than 250by IMERG mm. to have However, the radar product estimated almost twice (7700 km2) the area reported by IMERG to have received precipitation of more than 250 mm on that day. Moreover, the maximum daily rainfall observed on that day by radar was 2.3 times the maximum daily rainfall observed by IMERG, which was 380 mm. Lastly, the dissipation of the tropical storm was observed on 20 September from both products with quite different spatial patterns. Atmosphere 2021, 12, x FOR PEER REVIEW 8 of 18

received precipitation of more than 250 mm on that day. Moreover, the maximum daily rainfall observed on that day by radar was 2.3 times the maximum daily rainfall observed by IMERG, which was 380 mm. Lastly, the dissipation of the tropical storm was observed Atmosphere 2021, 12, 687 8 of 18 on 20 September from both products with quite different spatial patterns.

FigureFigure 3. Evolution 3. Evolution of the of theTropical Tropical Storm Storm Imelda Imelda after after its landfall on on 17 17 September September 2019 2019 (daily (daily rainfall rainfall≥30 mm≥30 mm are shown) are shown) and spatialand spatial pattern pattern captured captured by by(A ()A IMERG) IMERG Final Final-run,-run, (B)) Stage-IVStage-IV products. products.

A Atime time series series plot plot of anan area area average average of theof the tropical tropical storm storm shows shows the temporal the temporal evolution evolu- tionat at a higher a higher temporal temporal resolution resolution as shown as inshown Figure in4. TheFigure time 4. series The oftime the eventseries showed of the event showedthat the that precipitation the precipitation data during data Tropical during Strom Tropical Imelda Strom had Imelda a bimodal had distribution. a bimodal Thedistribu- first peak was reached at about 11:00 AM on 18 September UTC time zone. The second tion. The first peak was reached at about 11:00 AM on 18 September UTC time zone. The and largest peak came more than 24 h later at 2:00 PM on 19 September UTC time zone second(Figure and4) . largest The second peak peak came was more captured than by 24 all h the later products at 2:00 with PM reasonable on 19 September accuracy inUTC its time zonerise (Figure and fall. 4). However, The second the first peak peak was was captured overestimated by all significantlythe products by with the IMERG reasonable Final- accu- racyrun in product, its rise withand anfall. average However, of 1.5 mmhrthe first−1 overpeak the was study overestimated area, which included significantly an area by the IMERGof around Final 41,100-run kmproduct,2. This waswith approximately an average of 62 1.5 million mmhr cubic−1 over meters the of study water area, in an hourwhich in- cludedspread an across area theof around entire study 41,100 area. km The2. This Early-run was approximately product showed 62 the million best fit amongcubic meters the of waterother in IMERG an hour products spread matching across the both entire of the study peaks ofarea. the The radar Early product.-run The product Late-run showed also the bestshowed fit among a similar the other underestimation IMERG products in both matching of the peaks. both However, of the peaks the highest of the correlation radar product. coefficient of the time series was shown between the radar and the Late-run, with a value The Late-run also showed a similar underestimation in both of the peaks. However, the of 0.98. Moreover, the other two products also showed high correlation coefficients of 0.97 highestand 0.96 correlation for the Early- coefficient and Final-run, of the time respectively. series was A recent shown study between conducted the radar over theand the LateSichuan-run, with Basin a of value China of revealed 0.98. Moreover, that the precipitation the other two deviation products of the also Final-run showed products high corre- lationwas coefficients mainly observed of 0.97 during and moderate0.96 for the precipitation Early- and events Final- (1–10run, respectively. mmhr−1)[67]. A Final-run recent study conductedproducts showedover the better Sichuan detection Basin over of China Early- revealed and Late-run that products the precipitation for light precipitation deviation of the Final(<1-run mmhr products−1) events was [67]. mai Thisnly might observed explain the during slight moderate underestimations precipitation on 17 September events (1–10 −1 mmhrand−1 18) [67] September. Final- whenrun products precipitation showed was lowerbetter than detection 1 mmhr over, and Early the- overestimations and Late-run prod- −1 uctson for the light following precipitation days when (<1 precipitation mmhr−1) events ranged [67] between. This 1.25–10might explain mmhr the. slight underestimations on 17 September and 18 September when precipitation was lower than 1 mmhr−1, and the overestimations on the following days when precipitation ranged between 1.25–10 mmhr−1.

Atmosphere 2021, 12, x FOR PEER REVIEW 9 of 18

Atmosphere 20212021,, 1212,, 687x FOR PEER REVIEW 9 of 18

Figure 4. Time series plot of areal average precipitation estimates captured by the IMERG and FigureStage- IV4. 4. TimeproductsTime series series during plot plot of Tropical ofareal areal average Storm average Imeldaprecipitation precipitation (17–21 estimates September estimates captured 2019). captured by the by IMERG the IMERG and and StageStage-IV-IV products during Tropical Storm Imelda (17–21(17–21 SeptemberSeptember 2019).2019). 3.3. Basic Statistical Indices 3.3. Basic Basic Statistical Statistical Indices Indices FigureFigure5 5 shows shows the the CC CC spatial spatial patterns patterns between between Stage-IV Stage-IV radar radar data data and and each each of of the the threeFigure IMERG 5 shows-GPM satellitethe CC spatialproducts patterns during between Tropical Stage Storm-IV Imelda. radar data The and CC eachspatial of pat-the threethree IMERG-GPMIMERG-GPM satellitesatellite productsproducts during during Tropical Tropical Storm Storm Imelda. Imelda. The The CC C spatialC spatial pattern pat- showstern shows the consistencythe consistency of the of the IMERG-GPM IMERG-GPM products products with with more more detail. detail. As As there there is is highhigh ternspatial shows variability the consistency during of extreme the IMERG events,-GPM the products CC pattern with helps more to detail. evaluate As there the is perfor- high spatialspatial variability variability during during extreme extreme events, events, the the CC CC pattern pattern helps helps to evaluate to evaluate the performance the perfor- ofmance the satellite of the satellite products products and illustrate and illustrate their characteristics. their characteristics. mance of the satellite products and illustrate their characteristics.

FigureFigure 5.5. Spatial distributions distributions of of the the correlation correlation coefficient coefﬁcient (CC) (CC) between between Stage Stage-IV-IV radar radar data data and and (A) ( AIMERG) IMERG Early Early-,-, (B) Figure 5. Spatial distributions of the correlation coefficient (CC) between Stage-IV radar data and (A) IMERG Early-, (B) (LateB) Late-,-, and and (C) ( CFinal) Final-run-run products products over over 17 17Counties Counties in Texas in Texas (n (=n 192;= 192; hourly hourly from from 14– 14–2121 September September 2019). 2019). Late-, and (C) Final-run products over 17 Counties in Texas (n = 192; hourly from 14–21 September 2019). TheThe spatial patterns patterns of of C CCC show show a higher a higher consistency consistency in the in coastal the coastal areas areasand in and places in placeshit hardestThe hit spatial hardest by the patterns bystorm the of (Figure stormCC show (Figure5). aThe higher5 average). Theconsistency averageCC was in CC almostthe wascoastal the almost areassame the andfor same inthe places three for hittheIMERG hardest three products, IMERG by the products,atstorm 0.574, (Figure 0.600, at 0.574, 5).and The 0.600, 0.595 average andfor the 0.595 CC Early forwas- the, almostLate Early-,-, and the Late-, Finalsame- and runfor Final-runtheproducts, three IMERGproducts,respectively. products, respectively. The atspatial 0.574, The distribution 0.600, spatial and distribution 0.595also shows for the alsosimilar Early shows- , behaviorsLate similar-, and for behaviorsFinal all- runof the products, for IMERG all of respectively.theproducts. IMERG Higher products.The spatialCC values Higher distribution were CC observed values also shows were for all observedsimilar comparisons behaviors for all over comparisons for the all countiesof the over IMERG of theJef- products.countiesferson, Liberty, of Higher Jefferson, and CC Montgomery Liberty,values were and Montgomery observedin the northeastern for inall thecomparisons northeastern part of the over study part the of countiesarea the study(Figure of areaJef- 5). ferson,(FigureGenerally, 5Liberty,) . c Generally,oastal and areas Montgomery coastal with a areas high in with amountthe northeastern a high of precipitation amount part of of precipitationhad the relativelystudy area had higher (Figure relatively corre- 5). Generally,higherlation coefficients correlation coastal areas(CC) coefﬁcients thanwith the a (CC) high areas than amount further the areasof inland. precipitation further Two inland.-thirds had of Two-thirdsrelatively the study higher of area the (abovecorre- study lationarea67%) (above showedcoefficients 67%) a CC (CC) showed higher than than a the CC 0.5areas higher for further all thanIMERG inland. 0.5 products. for Two all IMERG-thirds Though of products. the the study Late- Thoughrunarea product (above the 67%)Late-runshowed showed the product highest a CC showed higher consistency than the highest0.5 of for CC all consistencythroughout IMERG products. ofthe CC study throughout Though area, thethe the FinalLate study--runrun area, productproduct the shoFinal-runshowedwed thea productbetter highest consistency showed consistency a betterin theof consistencyCCregions throughout near in the the the center regions study of neararea,the storm. thethe centerFinal An- earlierrun of the product storm.study showedAn earlier a better study consistency [53] on the same in the region regions during near hurricanethe center Harvey of the storm. found An that earlier the complex study

Atmosphere 2021, 12, x FOR PEER REVIEW 10 of 18

Atmosphere 2021, 12, 687 10 of 18 [53] on the same region during hurricane Harvey found that the complex structure of hurricane Harvey might have significantly affected the performance of the satellite products structurenear the ofcenter hurricane of the hurricane. Harvey might have significantly affected the performance of the satelliteFigure products 6 shows near scatterplots the center of of the IMERG hurricane. Early-, Late- and, Final-run satellite products withFigure Stage6- IVshows radar scatterplots product during of IMERG Tropical Early-, Storm Late- Imelda. and, Final-runThe scatterplot satellite helps products to visu- withally Stage-IVinspect the radar alignment product duringof the individual Tropical Storm pixels Imelda. of the Thetwo scatterplotproducts during helps to the visually event). inspect the alignment of the individual pixels of the two products during the event). Similar Similar to the CC values, all products followed similar patterns, with almost the same to the CC values, all products followed similar patterns, with almost the same linear model linear model fit (R2) with minor variations in the slopes (0.46, 0.43, and 0.44 for Early-, fit (R2) with minor variations in the slopes (0.46, 0.43, and 0.44 for Early-, Late-, and Final- Late-, and Final-run, respectively). All the slopes were found to be significantly different run, respectively). All the slopes were found to be significantly different than zero at a than zero at a 0.01 significance level. There were significant differences between the radar 0.01 significance level. There were significant differences between the radar estimates and estimates and those of the satellite at a pixel scale. Moreover, overestimation of light pre- those of the satellite at a pixel scale. Moreover, overestimation of light precipitation and cipitation and underestimation of heavy precipitation is evident from the scatterplots. underestimation of heavy precipitation is evident from the scatterplots. However, there However, there was a clear improvement in the quality of the data from the Early-run to was a clear improvement in the quality of the data from the Early-run to the Final-run. In the Final-run. In Figure 6, the systematic error is seen as an apparent point cloud, appear- Figure6, the systematic error is seen as an apparent point cloud, appearing as a ceiling in ing as a ceiling in the top part (Figure 6A). That ceiling of the point cloud is much lower the top part (Figure6A). That ceiling of the point cloud is much lower in the Late-run and in the Late-run and disappears in the Final-run. The coefficient of determination (R2) with disappears in the Final-run. The coefficient of determination (R2) with a value around 0.4 showsa value a pooraround fit for0.4 theshows IMERG a poor products. fit for the The IMERG data visualization products. The and data analysis visualization in Figure and6 showsanalysis that in theFigure IMERG 6 shows algorithm that the needs IMERG further algorithm improvements needs further to minimizeimprovements systematic to min- errorsimize for systematic the Early- errors and Late-runfor the Early products.- and Late-run products.

FigureFigure 6. 6.Scatterplot Scatterplot of of IMERG IMERG (A (A) Early-,) Early- (,B (B) Late-,) Late- and, and ( C(C)) Final-run Final-run products, products, and and Stage-IV Stage-IV average average hourly hourly precipitation precipitation datadata from from 14 14 to to 21 21 September September 2019. 2019.

TheThe spatial spatial distribution distribution ofof thethe RMSE (Figure 77)) showsshows an almost similar similar pattern pattern in in all allthe the IMERG IMERG products. products. It is It clearly is clearly seen seen that areas that areas highly highly affected affected by Tropical by Tropical Strom StromImelda Imelda(southeast (southeast coastal coastal regions regions stretching stretching up up to tothe the border border with with Louisiana) Louisiana) show show a a higher higher RMSE,RMSE, with with two two hot hot spots spots in in all all three three products. products. The The RMSE RMSE of of the the products products suggests suggests that that thethe products products were were improved, improved, in in minimizing minimizing the the average average error error compared compared to to a a similar similar study study conductedconducted in in the the same same area area earlier earlier [ 53[53]]. The. The mean mean and and median median RMSE RMSE was was 3.74 3.74 mm mm and and 3.673.67 mm mm,, 3.88 3.88 mmmm and 3.27 3.27 mm, mm, and and 3.15 3.15 mm mm and and 3.42 3.42 mm mm for forthe theEarly Early-,-, Late Late-,-, and Final and - Final-run,run, respectively. respectively. In all In of all the of three the three IMERG IMERG products, products, about about three three quarters quarters (ranging (ranging from from73% 73%to 76%) to 76%) of the of area the experienced area experienced an RMSE an RMSE of less ofthan less 5 thanmm in 5 mmestimating in estimating the Tropical the TropicalStorm Imelda. Storm Imelda. Cui et al. Cui [44] et al.conducted [44] conducted a study a studyover the over central the central and eastern and eastern United United States Statescomparing comparing IMERG IMERG products products with ground with ground-based-based observations observations and found and foundthat the that IMERG the IMERGproducts products showed showed the largest the largest RMSE RMSE values values during during summer summer and and fall fall months, months, caused caused by bysevere severe underestimation. underestimation. They They attributed attributed this this error error to dry biasesbiases ofof thethe IMERG IMERG algorithm algorithm forfor precipitation precipitation estimation estimation of of convective convective systems. systems. Figure8 illustrates the spatial variability of the relative bias (RBIAS) over the study area during Tropical Storm Imelda. The spatial distribution of the RBIAS shows similar patterns for the near-real-time IMERG products (Early and Late). However, the spatial pattern was different for the Final-run product, which had a latency of about 3.5 months (Figure8). The inland areas show much higher positive RBIAS (overestimation) in the Final- run product than the two earlier products, especially in counties such as San Jacinto, Polk,

Atmosphere 2021, 12, 687 11 of 18

and Fort Bend, where the positive RBIAS was more than 75%. This might be attributed to the tendency of IMERG Final-run products to overestimate precipitation measurements in inland areas during extreme storm events [68,69]. With the Early- and Late-run products, the coastal areas show a higher positive RBIAS and the inlands show a lower negative RBIAS. Overall, the Early- and Late-run products underestimated the tropical storm, with mean RBIAS values of −4% and −11%, and medians of −15% and −18%, respectively. However, the Final-run product showed contrasting results by providing a mean RBIAS of 41% and a median RBIAS of 20%. This suggested that the Final-run product overestimated the tropical storm by a large margin. However, when we follow the spatial distribution of the RBIAS (Figure8C), it is very clear that the Final-run product signiﬁcantly overestimated precipitation in less-hit areas. In contrast, it underestimated precipitation in the hard-hit areas. For all three products, about 50% of the areas experienced RBIAS between −30% and 30%. Hence, it can be said that the performance of the IMERG Final-run product was not as good as the Early- and Late-run products in capturing the storm event. In more than 30% of the study area, the Final-run product indicated an overestimation of the storm event by more than 50%. However, the Late and Early product estimates showed an overestimation of more than 50%, in 8% and 12% of the study area, respectively. This suggest that there was a signiﬁcant problem of overestimation in the Final product, especially in the outskirts of the study area (Figure8C). The areas with an acceptable range of RBIAS (i.e., between −10% and 10%) in all the three IMERG products were about 16% to 17% of the total area. This outcome follows similar ones from earlier studies, where GPM products underestimated Atmosphere 2021, 12, x FOR PEER REVIEWprecipitation measurements during extreme events [35,53,68,69]. Table2 summarizes11 theof 18

performance of the three IMERG-GPM products based on basic statistical indices for the study period.

Atmosphere 2021, 12, x FOR PEER REVIEW 12 of 18 FigureFigure 7.7. Spatial distribution of RSME for the IMERG (A)) Early-,Early-, ((BB)) Late-,Late-, andand ((CC)) Final-runFinal-run hourlyhourly productsproducts duringduring TropicalTropical StormStorm Imelda.Imelda.

Figure 8 illustrates the spatial variability of the relative bias (RBIAS) over the study area during Tropical Storm Imelda. The spatial distribution of the RBIAS shows similar patterns for the near-real-time IMERG products (Early and Late). However, the spatial pattern was different for the Final-run product, which had a latency of about 3.5 months (Figure 8). The inland areas show much higher positive RBIAS (overestimation) in the Final-run product than the two earlier products, especially in counties such as San Jacinto, Polk, and Fort Bend, where the positive RBIAS was more than 75%. This might be attributed to the tendency of IMERG Final-run products to overestimate precipitation measurements in inland areas during extreme storm events [68,69]. With the Early- and Late- run products, the coastal areas show a higher positive RBIAS and the inlands show a lower negative RBIAS. Overall, the Early- and Late-run products underestimated the tropical storm, with mean RBIAS values of −4% and −11%, and medians of −15% and −18%, Figure 8. Spatial distribution of RBIAS for the IMERG (A) Early-, (B) Late-, and (C) Final-run products during Tropical Figure 8. Spatial distributionrespectively. of RBIAS for However, the IMERG the (A Final) Early-run-, (B )produ Late-,ct and showed (C) Final contrasting-run products results during by Tropicalproviding a Storm Imelda. Storm Imelda. mean RBIAS of 41% and a median RBIAS of 20%. This suggested that the Final-run product overestimated the tropical storm by a large margin. However, when we follow the spatialThe distribution Kling–Gupta of modelthe RBIAS efficiency (Figure coefficient 8C), it is (KGE) very clearshows that a spatial the Final distribution-run product pat- significantlytern similar to overestimated the other basic precipitation statistical indices in less above.-hit areas. The earlyIn contrast, and late it products underestimated outper- precipitationformed the final in the product, hard-hit especially areas. For in all the three places products, that were about less 50% affected of the byareas the experi- storm encedevents RBIAS(low precipitation), between −30% as andshown 30%. in FigureHence, 9. it The can averagebe said valuesthat the of performance the KGE show of thatthe IMERGthe late productFinal-run was product found was to be not the as bestgood model as the for Early capturing- and Late the-run event, products with an in average captur- ingof 0.39, the stormfollowed event. by Inthe more Early than and 30% Final of products, the study with area, KGE the Final values-run of product 0.36 and indicated 0.16, re- anspectively. overestimation Around of one the-third storm of theevent study by areamore showed than 50%. a KGE However, of more thanthe Late 0.5 in and the Early productand Late estimatesproducts, showedwhereas an only overestimation one-quarter was of more indicated than in50%, the Finalin 8% product. and 12% of the study area, respectively. This suggest that there was a significant problem of overestimation in the Final product, especially in the outskirts of the study area (Figure 8C). The areas with an acceptable range of RBIAS (i.e., between −10% and 10%) in all the three IMERG products were about 16% to 17% of the total area. This outcome follows similar ones from earlier studies, where GPM products underestimated precipitation measurements during extreme events [35,53,68,69]. Table 2 summarizes the performance of the three IMERG- GPM products based on basic statistical indices for the study period.

Figure 9. Spatial distribution of KGE for the IMERG (A) Early-, (B) Late-, and (C) Final-run products during Tropical Storm Imelda.

Table 2. The areal average of the basic statistical indices of the satellite products.

Study Period Product CC RMSE RBIAS (%) KGE Early 0.574 3.74 −4.0 0.36 14–21 September 2019 Late 0.600 3.66 −11.00 0.39

Final 0.595 3.88 41.00 0.16

3.4. Probabilistic Statistical Indices In the POD analysis (Figure 10), Final-run was found to have the best results, with the highest POD values out of the three GPM products. The spatial distribution of the POD in all three products reveals that, as expected, the POD was found to be higher in counties that registered higher precipitation. The average POD for all three products was around 0.90, with a higher median of 0.92. For the Final-run product, the POD was observed to be greater than 0.80 in about 95% of the study area. However, the short study period and small study area (8 days with 192 observations) could have skewed the POD values. Moreover, the timeframe included in the study covers an extreme event that can be easily detected by both

Atmosphere 2021, 12, x FOR PEER REVIEW 12 of 18

Atmosphere 2021, 12, 687 12 of 18

Table 2. The areal average of the basic statistical indices of the satellite products.

Study Period Product CC RMSE RBIAS (%) KGE 14–21 Early 0.574 3.74 −4.0 0.36 − Figure 8. Spatial distribution ofSeptember RBIAS for the IMERGLate (A) Early- 0.600, (B) Late-, and (C 3.66) Final-run products11.00 during Tropical 0.39 Final 0.595 3.88 41.00 0.16 Storm Imelda. 2019

TheThe Kling Kling–Gupta–Gupta model model efficiency efficiency coefficient coefficient (KGE) (KGE) shows shows a spatial a spatial distribution distribution pat- patterntern similar similar to the to other the other basic basic statistical statistical indices indices above. above. The early The and early late and products late products outper- outperformedformed the final the finalproduct, product, especially especially in the in the places places that that were were less less affected affected by by the storm storm eventsevents (low precipitation), precipitation), as as shown shown in in Figure Figure 9.9 .The The average average values values of ofthe the KGE KGE show show that thatthe late the product late product was found was found to be the to be best the model best modelfor capturing for capturing the event, the with event, an withaverage an averageof 0.39, followed of 0.39, followed by the Early by the and Early Final and products, Final products, with KGE with values KGE of values 0.36 and of 0.36 0.16, and re- 0.16,spectively. respectively. Around Around one-third one-third of the study of the area study showed area showeda KGE of a more KGE than of more 0.5 in than the0.5 Early in theand Early Late andproducts, Late products, whereas only whereas one- onlyquarter one-quarter was indicated was indicated in the Final in the product. Final product.

FigureFigure 9.9. SpatialSpatial distributiondistribution of KGEKGE forfor thethe IMERGIMERG ((AA)) Early-,Early-, ((BB)) Late-,Late-, andand ((CC)) Final-runFinal-run productsproducts duringduring TropicalTropical StormStorm Imelda.Imelda.

Table 2.3.4. The Probabilistic areal average Statistical of the basic Indices statistical indices of the satellite products. Study Period ProductIn the POD analysisCC (Figure 10), Final-runRMSE was foundRBIAS to have (%) the best results,KGE with the highestEarly POD values0.574 out of the three GPM3.74 products. The spatial−4.0 distribution of the0.36 POD in all three products reveals that, as expected, the POD was found to be higher in counties 14–21 September 2019 Late 0.600 3.66 −11.00 0.39 that registered higher precipitation. The average POD for all three products was around Final 0.595 3.88 41.00 0.16 0.90, with a higher median of 0.92. For the Final-run product, the POD was observed to be greater than 0.80 in about 95% of the study area. However, the short study period and small study3.4. Probabilistic area (8 days Statistical with 192 Indices observations) could have skewed the POD values. Moreover, the timeframeIn the POD included analysis in (Figure the study 10), covers Final-run an extreme was found event to thathave can the bebest easily results, detected with the by bothhighest radar POD and values satellite out products.of the three Hence, GPM toproducts. normalize The the spatial presence distribution of the extreme of the POD event, in threeall three days products before landfallreveals that, and as after expected, dissipation the POD were was included found into thebe higher analysis. in counties that registeredIn Figure higher 11, theprecipitation. false alarm The ratio average (FAR) followedPOD for similarall three patterns products to thosewas around of the POD. 0.90, Lowerwith a higher FAR values median were of 0.92. registered For the in Final the- areasrun product, heavily the affected POD was by the obse storm,rved to and be highergreater FARthan values0.80 inwere about seen 95% in of the the periphery study area. of theHowever, study area. the short All three study IMERG period products and small showed study aarea similar (8 days spatial with pattern. 192 observations) The Early- and could Late-run have products skewed the showed POD a values. relatively Moreover, better FAR, the withtimeframe an average included value in the of 0.40,study however, covers an the extreme Final-run event product that can registered be easily detected a FAR value by both of 0.44. The Early product showed its advantage in FAR by registering a lower median, of about 0.35, compared to 0.38 and 0.42 in the Late- and Final-run products, respectively. Nearly two-thirds (67%) of the study area for the Early-run product showed a FAR of less than 0.4, whereas, only 40% of the areas in the Final-run showed a FAR of less than 0.4. Generally, the FAR by all the IMERG products was found to be very high, especially when considering the analysis was performed for a short period (8 days) and in an extreme event condition. Atmosphere 2021, 12, x FOR PEER REVIEW 13 of 18

radar and satellite products. Hence, to normalize the presence of the extreme event, three days before landfall and after dissipation were included in the analysis.

Atmosphere 2021, 12, x FOR PEER REVIEW 13 of 18

radar and satellite products. Hence, to normalize the presence of the extreme event, three Atmosphere 2021, 12, 687 13 of 18 days before landfall and after dissipation were included in the analysis. Figure 10. Spatial distribution of POD for the IMERG (A) Early-, (B) Late-, and (C) Final-run products during Tropical Storm Imelda.

In Figure 11, the false alarm ratio (FAR) followed similar patterns to those of the POD. Lower FAR values were registered in the areas heavily affected by the storm, and higher FAR values were seen in the periphery of the study area. All three IMERG products showed a similar spatial pattern. The Early- and Late-run products showed a relatively better FAR, with an average value of 0.40, however, the Final-run product registered a FAR value of 0.44. The Early product showed its advantage in FAR by registering a lower median, of about 0.35, compared to 0.38 and 0.42 in the Late- and Final-run products, respectively. Nearly two-thirds (67%) of the study area for the Early-run product showed a FAR of less than 0.4, whereas, only 40% of the areas in the Final-run showed a FAR of less than 0.4. Generally, the FAR by all the IMERG products was found to be very high, especially when considering the analysis was performed for a short period (8 days) and in an FigureFigure 10.10. SpatialSpatial distributiondistribution of PODPOD forfor thethe IMERGIMERG ((AA)) Early-,Early-, ((BB)) Late-,Late-, andand ((CC)) Final-runFinal-run productsproducts duringduring TropicalTropical extreme event condition. StormStorm Imelda.Imelda.

In Figure 11, the false alarm ratio (FAR) followed similar patterns to those of the POD. Lower FAR values were registered in the areas heavily affected by the storm, and higher FAR values were seen in the periphery of the study area. All three IMERG products showed a similar spatial pattern. The Early- and Late-run products showed a relatively better FAR, with an average value of 0.40, however, the Final-run product registered a FAR value of 0.44. The Early product showed its advantage in FAR by registering a lower median, of about 0.35, compared to 0.38 and 0.42 in the Late- and Final-run products, respectively. Nearly two-thirds (67%) of the study area for the Early-run product showed a FAR of less than 0.4, whereas, only 40% of the areas in the Final-run showed a FAR of less than 0.4. Generally, the FAR by all the IMERG products was found to be very high, especially when considering the analysis was performed for a short period (8 days) and in an extreme event condition. FigureFigure 11.11. SpatialSpatial distributiondistribution ofof FARFAR forfor thethe IMERGIMERG ((AA)) Early-,Early-, (B(B)) Late-,Late-, andand (C(C)) Final-runFinal-run productsproducts duringduring TropicalTropical StormStorm Imelda. Imelda.

TheThe spatialspatial distributiondistribution ofof thethe criticalcritical successsuccess indexindex (CSI)(CSI) againagain supportedsupported thatthat thethe near-real-timenear-real-time IMERG IMERG products products (Early (Early and and Late) Late) outperformed outperformed the the Final-run, Final-run, especially especially in thein the areas areas where where the the storm storm was was most most severe severe (Figure (Figure 12). 12). The The Early-run Early- productrun product showed showed the bestthe best performance performance in terms in terms of CSI, of CSI, with with an average an average CSI ofCSI 0.57 of and0.57 aand median a median of 0.60. of The0.60. Final-runThe Final product-run product was found was found to have to thehave lowest the lowest CSI, having CSI, having an average an average value ofvalue 0.53 of and 0.53 a medianand a median of 0.55. of To 0.55. put To this put into this perspective, into perspective, in the in Early-run the Early product,-run product, about about 20% 20% of the of Atmosphere 2021, 12, x FOR PEER REVIEW 14 of 18 studythe study area area showed showed a CSI a of CSI more of more than 0.7,than but 0.7, only but 5% only of the5% areaof the recorded area recorded a similar a similar CSI in theCSI Final-run in the Final product.-run product.

Figure 11. Spatial distribution of FAR for the IMERG (A) Early-, (B) Late-, and (C) Final-run products during Tropical Storm Imelda.

The spatial distribution of the critical success index (CSI) again supported that the near-real-time IMERG products (Early and Late) outperformed the Final-run, especially in the areas where the storm was most severe (Figure 12). The Early-run product showed the best performance in terms of CSI, with an average CSI of 0.57 and a median of 0.60. The Final-run product was found to have the lowest CSI, having an average value of 0.53 and a median of 0.55. To put this into perspective, in the Early-run product, about 20% of the study area showed a CSI of more than 0.7, but only 5% of the area recorded a similar CSI in the Final-run product. FigureFigure 12. SpatialSpatial distribution distribution of of CSI CSI for for the the IMERG IMERG (A)( EarlyA) Early-,-, (B) Late (B) Late-,-, and ( andC) Final (C)- Final-runrun products products during during Tropical Tropical Storm StormImelda. Imelda.

The Peirce skill score (PSS) helps by providing insights into the ability of satellite products to separate the actual occurrences of precipitation from the no-occurrence instances. PSS shows the overall robustness of the system to accurately predict the tropical storm event. The Early- and Late- products again outperformed the Final-run in the PSS analysis (Figure 13), in a similar way as for FAR and CSI. The spatial distribution of PSS from the three products showed a PSS of more than 0.70 in case of the Early and Late products, however, only 36% of the area was found to have a PSS higher than 0.70 in the Final-run product. Table 3 summarizes the probabilistic statistical indices of the performance of the Early-, Late-, and Final-run products. A comparison of the summarized data with an earlier study [44] conducted over the central United States showed similar values of POD (~0.91), slightly lower values of FAR (~0.33), and slightly higher values of CSI (~0.62) for fall months (September, October and November).

Figure 13. Spatial distribution of PSS for the IMERG (A) Early-, (B) Late-, and (C) Final-run products during Tropical Storm Imelda.

Table 3. The areal average of the probabilistic statistical indices of the satellite products.

Study Period Product POD FAR CSI PSS Early 0.87 0.38 0.57 0.68 14–21 September 2019 Late 0.90 0.41 0.56 0.68 Final 0.91 0.44 0.53 0.65

4. Conclusions This study focused on the performance of the GPM-IMERG- Early-, Late-, and Final- run satellite precipitation products during the Tropical Storm Imelda. Categorized as a tropical storm at midday on 17 September by the National Hurricane Center (NHC), Imelda

Atmosphere 2021, 12, x FOR PEER REVIEW 14 of 18

Figure 12. Spatial distribution of CSI for the IMERG (A) Early-, (B) Late-, and (C) Final-run products during Tropical Storm Atmosphere 2021, 12, 687 14 of 18 Imelda.

The Peirce skill score (PSS) helps by providing insights into the ability of satellite The Peirce skill score (PSS) helps by providing insights into the ability of satellite prod- products to separate the actual occurrences of precipitation from the no-occurrence in- ucts to separate the actual occurrences of precipitation from the no-occurrence instances. PSSstances. shows PSS the shows overall the robustnessoverall robustness of the system of the system to accurately to accurately predict predict the tropical the tropical storm event.storm event. The Early- The andEarly Late-- and products Late- products again outperformed again outperformed the Final-run the Final in the-run PSS in analysisthe PSS (Figureanalysis 13 (Figure), in a similar13), in a way similar as for way FAR as andfor FAR CSI. Theand spatialCSI. The distribution spatial distribution of PSS from of PSS the threefrom the products three showedproducts a showed PSS of morea PSS than of more 0.70 inthan case 0.70 of in the case Early of andthe Early Late products,and Late however,products, onlyhowever, 36% ofonly the 36% area of was the found area was to have found a PSSto hav highere a PSS than higher 0.70 inthan the 0.70 Final-run in the product.Final-run Tableproduct.3 summarizes Table 3 summarizes the probabilistic the probabilistic statistical statistical indices of indices the performance of the perfor- of themance Early-, of the Late-, Early and-, Late Final-run-, and Final products.-run products. A comparison A comparison of the summarized of the summarized data with data an earlierwith an study earlier [44 study] conducted [44] conducted over the over central the United central StatesUnited showed States showed similar valuessimilar of values POD (~0.91),of POD slightly(~0.91), lowerslightly values lower of values FAR (~0.33), of FAR and (~0.33), slightly and higher slightly values higher of CSIvalues (~0.62) of CSI for fall(~0.62) months for fall (September, months (September, October and October November). and November).

FigureFigure 13.13. Spatial distribution of PSS for the IMERG (A) Early-,Early-, ((B)) Late-,Late-, andand ((CC)) Final-runFinal-run productsproducts duringduring TropicalTropical StormStorm Imelda.Imelda.

Table 3. TheTable areal 3. averageThe areal of averagethe probabilistic of the probabilistic statistical indices statistical of the indices satellite of the products. satellite products. Study Period Product POD FAR CSI PSS Study Period Product POD FAR CSI PSS Early 0.87 0.38 0.57 0.68 14–21 Early 0.87 0.38 0.57 0.68 14–21 September 2019 Late 0.90 0.41 0.56 0.68 September Late 0.90 0.41 0.56 0.68 Final2019 Final0.91 0.910.44 0.440.53 0.530.65 0.65

4. Conclusions This study focused on the performance of the GPM-IMERG-GPM-IMERG- Early-,Early-, Late-,Late-, and Final-Final- run satellite satellite precipitation precipitation products products during during the theTropical Tropical Storm Storm Imelda. Imelda. Categorized Categorized as a trop- as aical tropical storm stormat midd atay midday on 17 September on 17 September by the byNational the National Hurricane Hurricane Center Center(NHC), (NHC),Imelda Imelda made landfall with massive downpours during 17–19 September 2019. The spatial distribution and temporal evolution of the storm during 17–20 September showed that the storm caused two major peak precipitations; one on 18 September and one on 19 September. ◦ ◦ The quality of IMERG products at hourly and 0.1 × 0.1 temporal resolutions during the storm was evaluated using Stage-IV radar measurements as a reference. The hourly Stage-IV radar product was calibrated and adjusted using rain gauge networks in the surrounding study area. The timeframe of the data was adjusted to include three days before the landfall and three days after the dissipation to better capture the temporal and spatial variability of the event. Generally, areas that received a higher amount of precipitation were found to have relatively good statistical metrics by all three of the IMERG GPM products. Maps of the CC showed that the alignment of the IMERG and radar was not as good as other studies have shown, with an average of 0.60 for the best product [53,62,70]. Moreover, the coefﬁcient of determination (R2) for all three IMERG products was just 0.4. When it comes to RMSE, three- quarters (ranging from 73% to 76%) of the area experienced an RMSE of less than 5 mm in all three IMERG products. The IMERG GPM Final-run product showed a huge overestimation Atmosphere 2021, 12, 687 15 of 18

of the storm, which was confirmed by basic statistical indices. The spatial distribution of the RBIAS indicated similar patterns in the near-real-time IMERG products (Early and Late). However, the gauge-adjusted IMERG product (Final-run) showed the very high overestimation and underestimation across the affected region varied by location (inland and coastal areas). Significant underestimation was found in the hard-hit coastal areas, whereas significant overestimation was recorded in the relatively less-hit inland areas. This summarizes the weakness of satellite-based precipitation products; it has been reported in the literature for a long time that the IMERG products underestimate precipitations during extreme (heavy) events and overestimate average to low events [70–75]. It is also notable that during summer and fall months, IMERG products have been found to show dry biases for convective systems [44]. The Early-run showed a much better RBIAS relative to the Final-run product. However, when considering the overall performance of the IMERG products, it was poor, because areas with an acceptable range of RBIAS (i.e., between −10% and 10%) in all the three IMERG products were only about 16% to 17% of the total area. Probabilistic statistical indices showed that all three products were successful in capturing the precipitation event with adequate accuracy (average POD ≈ 0.9). FAR index results were less strong, with the Early- and Late-run products having almost 67% of the areas with a FAR <0.4; however, only 40% of the areas in the Final-run product showed a FAR <0.4. PSS and CSI scores showed that the IMERG GPM Final-run product was outperformed by the near-real-time IMERG products (Early and Late) in capturing the entire storm event with better accuracy. Overall, between the products, the Early-run was found to be better than the Late- and Final-run products for capturing the storm precipitation. Thus, it can be opined that the processing algorithm that used the monthly gauge analysis to develop the Final-run failed to capture extreme events that occurred for more than a few days. Furthermore, better precipitation estimates for all stages of the GPM products will be crucial in hydrological analysis and modeling during extreme events, such as hurricanes, tornadoes, and tropical storms. This conclusion is supported by earlier studies on IMERG-GPM products during extreme precipitation events [35,53,68,69].

Author Contributions: Conceptualization, S.S., D.G. and H.O.S.; methodology, S.S. and D.G.; soft- ware, D.G.; validation, S.S., D.G. and H.O.S.; formal analysis, S.S.; investigation, D.G.; resources, H.O.S.; data curation, D.G.; writing—original draft preparation, S.S.; writing—review and editing, S.S. and H.O.S.; visualization, S.S. and D.G.; supervision, H.O.S. All authors have read and agreed to the published version of the manuscript. Funding: This research was partially funded by U.S. Army Research Office (Grant W912HZ-14- P-0160) and a grant from UTSA (University of Texas at San Antonio) Open Cloud Institute (OCI) entitled “Prediction of Crash Incidence and Severity on Texas Roadways”. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: The data presented in this study are available in http://pmm.nasa.gov/ data-access/downloads/gpm/ (accessed on 26 May 2021) and http://data.eol.ucar.edu/ (accessed on 26 May 2021). Acknowledgments: The aforementioned funding and supports are committedly acknowledged. Conflicts of Interest: The authors declare no conflict of interest.

References 1. NOAA National Centers for Environmental Information (NCEI) U.S. Billion-Dollar Weather and Climate Disasters. 2020. Available online: https://www.ncdc.noaa.gov/billions/ (accessed on 22 March 2020). 2. Smith, A.B. U.S. Billion-Dollar Weather and Climate Disasters, 1980–Present (NCEI Accession 0209268); NOAA National Centers for Environmental Information 2020. Available online: https://doi.org/10.25921/stkw-7w73 (accessed on 22 March 2020). 3. Schuetz, R.A. Texas Leads Country in Disasters. 18 April 2019. Available online: https://www.houstonchronicle.com/business/ real-estate/article/Texas-leads-country-in-disasters-13777748.php (accessed on 22 March 2020). Atmosphere 2021, 12, 687 16 of 18

4. Besse, G.; Sakib, S.; Gang, D.; Hayes, D. Performance characterization of a novel marsh shoreline protection technology: The Wave Suppression and Sediment Collection (WSSC) system. Ocean Eng. 2018, 154, 216–225. [CrossRef] 5. Sakib, S. Design Optimization and Field Performance Evaluation of the Wave Suppression and Sediment Collection (WSSC) System: Computational Fluid Dynamics (CFD) Modeling, Surface Elevation Table (SET) Survey, and Marker Clay Study; University of Louisiana at Lafayette: Lafayette, Louisiana, 2017. 6. Sakib, S.; Gang, D.; Besse, G.; Tang, B.-B.; McCoy, N.; Hayes, D. Laboratory study and mathematical modeling of a novel marsh shoreline protection technology for sand collection. Appl. Ocean Res. 2018, 76, 22–33. [CrossRef] 7. Landreneau, D.; Brazzell, J.; Mogged, T.; Erickson, R.; Warthen, S. Tropical Storm Imelda. Tropical Weather 2019. Available online: https://www.weather.gov/lch/2019Imelda (accessed on 22 March 2020). 8. Bui, H.T.; Ishidaira, H.; Shaowei, N. Evaluation of the use of global satellite–gauge and satellite-only precipitation products in stream flow simulations. Appl. Water Sci. 2019, 9, 53. [CrossRef] 9. Prakash, S.; Mitra, A.K.; AghaKouchak, A.; Liu, Z.; Norouzi, H.; Pai, D.S. A preliminary assessment of GPM-based multi-satellite precipitation estimates over a monsoon dominated region. J. Hydrol. 2018, 556, 865–876. [CrossRef] 10. Liu, Z. Comparison of Integrated Multisatellite Retrievals for GPM (IMERG) and TRMM Multisatellite Precipitation Analysis (TMPA) Monthly Precipitation Products: Initial Results. J. Hydrometeorol. 2016, 17, 777–790. [CrossRef] 11. Guo, H.; Chen, S.; Bao, A.; Hu, J.; Gebregiorgis, A.S.; Xue, X.; Zhang, X. Inter-Comparison of High-Resolution Satellite Precipitation Products over Central Asia. Remote Sens. 2015, 7, 7181–7211. [CrossRef] 12. Wang, W.; Lu, H.; Zhao, T.; Jiang, L.; Shi, J. Evaluation and Comparison of Daily Rainfall From Latest GPM and TRMM Products Over the Mekong River Basin. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2017, 10, 2540–2549. [CrossRef] 13. Wang, Z.; Zhong, R.; Lai, C.; Chen, J. Evaluation of the GPM IMERG satellite-based precipitation products and the hydrological utility. Atmos. Res. 2017, 196, 151–163. [CrossRef] 14. Afshari, S.; Tavakoly, A.A.; Rajib, M.A.; Zheng, X.; Follum, M.L.; Omranian, E.; Fekete, B.M. Comparison of new generation low-complexity flood inundation mapping tools with a hydrodynamic model. J. Hydrol. 2018, 556, 539–556. [CrossRef] 15. Omranian, E.; Sharif, H. Evaluation of the Global Precipitation Measurement (GPM) Satellite Rainfall Products over the Lower Colorado River Basin, Texas. JAWRA J. Am. Water Resour. Assoc. 2018, 54, 882–898. [CrossRef] 16. Gilewski, P.; Nawalany, M. Inter-Comparison of Rain-Gauge, Radar, and Satellite (IMERG GPM) Precipitation Estimates Performance for Rainfall-Runoff Modeling in a Mountainous Catchment in Poland. Water 2018, 10, 1665. [CrossRef] 17. Tapiador, F.J.; Turk, F.J.; Petersen, W.; Hou, A.Y.; García-Ortega, E.; Machado, L.A.T.; Angelis, C.F.; Salio, P.; Kidd, C.; Huffman, G.J.; et al. Global precipitation measurement: Methods, datasets and applications. Atmos. Res. 2012, 104–105, 70–97. [CrossRef] 18. Li, Z.; Yang, D.; Hong, Y. Multi-scale evaluation of high-resolution multi-sensor blended global precipitation products over the Yangtze River. J. Hydrol. 2013, 500, 157–169. [CrossRef] 19. Michaelides, S.; Levizzani, V.; Anagnostou, E.; Bauer, P.; Kasparis, T.; Lane, J. Precipitation: Measurement, remote sensing, climatology and modeling. Atmos. Res. 2009, 94, 512–533. [CrossRef] 20. Guo, H.; Chen, S.; Bao, A.; Behrangi, A.; Hong, Y.; Ndayisaba, F.; Hu, J.; Stepanian, P.M. Early assessment of Integrated Multi-satellite Retrievals for Global Precipitation Measurement over China. Atmos. Res. 2016, 176, 121–133. [CrossRef] 21. Aksoy, A.; Dowell, D.C.; Snyder, C. A Multicase Comparative Assessment of the Ensemble Kalman Filter for Assimilation of Radar Observations. Part I: Storm-Scale Analyses. Mon. Weather. Rev. 2009, 137, 1805–1824. [CrossRef] 22. Berne, A.; Krajewski, W. Radar for hydrology: Unfulfilled promise or unrecognized potential? Adv. Water Resour. 2013, 51, 357–366. [CrossRef] 23. Gabella, M.; Morin, E.; Notarpietro, R.; Michaelides, S. Winter precipitation fields in the Southeastern Mediterranean area as seen by the Ku-band spaceborne weather radar and two C-band ground-based radars. Atmos. Res. 2013, 119, 120–130. [CrossRef] 24. Germann, U.; Galli, G.; Boscacci, M.; Bolliger, M. Radar precipitation measurement in a mountainous region. Q. J. R. Meteorol. Soc. 2006, 132, 1669–1692. [CrossRef] 25. Heinselman, P.L.; Priegnitz, D.L.; Manross, K.L.; Smith, T.M.; Adams, R.W. Rapid Sampling of Severe Storms by the National Weather Radar Testbed Phased Array Radar. Weather. Forecast. 2008, 23, 808–824. [CrossRef] 26. Krajewski, W.F.; Villarini, G.; Smith, J.A. RADAR-Rainfall Uncertainties: Where are We after Thirty Years of Effort? Bull. Am. Meteorol. Soc. 2010, 91, 87–94. [CrossRef] 27. McKee, J.L.; Binns, A.D. A review of gauge–radar merging methods for quantitative precipitation estimation in hydrology. Can. Water Resour. J. Rev. Can. Ressour. Hydr. 2016, 41, 186–203. [CrossRef] 28. Szturc, J.; O´sródka, K.; Jurczyk, A. Quality index scheme for quantitative uncertainty characterization of radar-based precipitation. Meteorol. Appl. 2011, 18, 407–420. [CrossRef] 29. Sharifi, E.; Steinacker, R.; Saghafian, B. Assessment of GPM-IMERG and Other Precipitation Products against Gauge Data under Different Topographic and Climatic Conditions in Iran: Preliminary Results. Remote Sens. 2016, 8, 135. [CrossRef] 30. Cecinati, F.; Wani, O.; Rico-Ramirez, M.A. Comparing Approaches to Deal With Non-Gaussianity of Rainfall Data in Kriging-Based Radar-Gauge Rainfall Merging. Water Resour. Res. 2017, 53, 8999–9018. [CrossRef] 31. Gabella, M.; Amitai, E. Radar rainfall estimates in an alpine environment using different gage-adjustment techniques. Phys. Chem. Earth, Part B Hydrol. Oceans Atmos. 2000, 25, 927–931. [CrossRef] 32. Keblouti, M.; Ouerdachi, L.; Berhail, S. The use of weather radar for rainfall-runoff modeling, case of Seybouse watershed (Algeria). Arab. J. Geosci. 2013, 8, 1–11. [CrossRef] Atmosphere 2021, 12, 687 17 of 18

33. Sikorska, A.; Seibert, J. Value of different precipitation data for flood prediction in an alpine catchment: A Bayesian approach. J. Hydrol. 2018, 556, 961–971. [CrossRef] 34. Gebregiorgis, A.S.; Kirstetter, P.; Hong, Y.E.; Gourley, J.J.; Huffman, G.J.; Petersen, W.A.; Xue, X.; Schwaller, M.R. To What Extent is the Day 1 GPM IMERG Satellite Precipitation Estimate Improved as Compared to TRMM TMPA-RT? J. Geophys. Res. Atmos. 2018, 123, 1694–17077. [CrossRef] 35. Sunilkumar, K.; Yatagai, A.; Masuda, M. Preliminary Evaluation of GPM-IMERG Rainfall Estimates Over Three Distinct Climate Zones With APHRODITE. Earth Space Sci. 2019, 6, 1321–1335. [CrossRef] 36. AghaKouchak, A.; Behrangi, A.; Sorooshian, S.; Hsu, K.; Amitai, E. Evaluation of satellite-retrieved extreme precipitation rates across the central United States. J. Geophys. Res. Space Phys. 2011, 116, D02115. [CrossRef] 37. Mehran, A.; AghaKouchak, A. Capabilities of satellite precipitation datasets to estimate heavy precipitation rates at different temporal accumulations. Hydrol. Process. 2014, 28, 2262–2270. [CrossRef] 38. Romine, G.S.; Schwartz, C.S.; Snyder, C.; Anderson, J.L.; Weisman, M.L. Model Bias in a Continuously Cycled Assimilation System and Its Influence on Convection-Permitting Forecasts. Mon. Weather. Rev. 2013, 141, 1263–1284. [CrossRef] 39. Smalley, M.; L’Ecuyer, T.; Lebsock, M.; Haynes, J. A Comparison of Precipitation Occurrence from the NCEP Stage IV QPE Product and the CloudSat Cloud Profiling Radar. J. Hydrometeorol. 2014, 15, 444–458. [CrossRef] 40. Chen, J.; Wang, Z.; Wu, X.; Chen, X.; Lai, C.; Zeng, Z.; Li, J. Accuracy evaluation of GPM multi-satellite precipitation products in the hydrological application over alpine and gorge regions with sparse rain gauge network. Hydrol. Res. 2019, 50, 1710–1729. [CrossRef] 41. Gosset, M.; Alcoba, M.; Roca, R.; Cloché, S.; Urbani, G. Evaluation of TAPEER daily estimates and other GPM-era products against dense gauge networks in West Africa, analysing ground reference uncertainty. Q. J. R. Meteorol. Soc. 2018, 144, 255–269. [CrossRef] 42. Sungmin, O.; Foelsche, U.; Kirchengast, G.; Fuchsberger, J.; Tan, J.; Petersen, W.A. Evaluation of GPM IMERG Early, Late, and Final rainfall estimates using WegenerNet gauge data in southeastern Austria. Hydrol. Earth Syst. Sci. 2017, 21, 6559–6572. 43. Zubieta, R.; Getirana, A.; Espinoza, J.C.; Lavado-Casimiro, W.; Aragon, L. Hydrological modeling of the Peruvian–Ecuadorian Amazon Basin using GPM-IMERG satellite-based precipitation dataset. Hydrol. Earth Syst. Sci. 2017, 21, 3543–3555. [CrossRef] 44. Cui, W.; Dong, X.; Xi, B.; Feng, Z.; Fan, J. Can the GPM IMERG Final Product Accurately Represent MCSs’ Precipitation Characteristics over the Central and Eastern United States? J. Hydrometeorol. 2020, 21, 39–57. [CrossRef] 45. Chen, F.; Li, X. Evaluation of IMERG and TRMM 3B43 Monthly Precipitation Products over Mainland China. Remote Sens. 2016, 8, 472. [CrossRef] 46. Islam, M.A. Statistical comparison of satellite-retrieved precipitation products with rain gauge observations over Bangladesh. Int. J. Remote Sens. 2018, 39, 2906–2936. [CrossRef] 47. Krishna, U.V.M.; Das, S.K.; Deshpande, S.M.; Doiphode, S.L.; Pandithurai, G. The assessment of Global Precipitation Measurement estimates over the Indian subcontinent. Earth Space Sci. 2017, 4, 540–553. [CrossRef] 48. Bhuiyan, M.A.E.; Yang, F.; Biswas, N.K.; Rahat, S.H.; Neelam, T.J. Machine Learning-Based Error Modeling to Improve GPM IMERG Precipitation Product over the Brahmaputra River Basin. Forecasting 2020, 2, 14. [CrossRef] 49. Bhuiyan, M.A.E.; Anagnostou, E.N.; Kirstetter, P.-E. A Nonparametric Statistical Technique for Modeling Overland TMI (2A12) Rainfall Retrieval Error. IEEE Geosci. Remote Sens. Lett. 2017, 14, 1898–1902. [CrossRef] 50. Biemans, H.; Hutjes, R.W.A.; Kabat, P.; Strengers, B.J.; Gerten, D.; Rost, S. Effects of Precipitation Uncertainty on Discharge Calculations for Main River Basins. J. Hydrometeorol. 2009, 10, 1011–1025. [CrossRef] 51. Bhuiyan, M.A.E.; Nikolopoulos, E.I.; Anagnostou, E.N. Machine Learning—Based Blending of Satellite and Reanalysis Precipita- tion Datasets: A Multiregional Tropical Complex Terrain Evaluation. J. Hydrometeorol. 2019, 20, 2147–2161. [CrossRef] 52. Bhuiyan, M.A.E.; Nikolopoulos, E.I.; Anagnostou, E.N.; Polcher, J.; Albergel, C.; Dutra, E.; Fink, G.; la Torre, A.M.; Munier, S. Assessment of precipitation error propagation in multi-model global water resource reanalysis. Hydrol. Earth Syst. Sci. 2019, 23, 1973–1994. [CrossRef] 53. Omranian, E.; Sharif, H.; Tavakoly, A. How Well Can Global Precipitation Measurement (GPM) Capture Hurricanes? Case Study: Hurricane Harvey. Remote Sens. 2018, 10, 1150. [CrossRef] 54. Billah, K.; Adegbite, Q.; Sharif, H.; Dessouky, S.; Simcic, L. Analysis of Intersection Traffic Safety in the City of San Antonio, 2013–2017. Sustainability 2021, 13, 5296. [CrossRef] 55. Hinojosa, J.; Green, J.; Estrada, F.; Herrera, J.; Mata, T.; Phan, D.; Pasha, A.T.; Matta, A.; Johnson, D.; Kapoor, V. Determining the primary sources of fecal pollution using microbial source tracking assays combined with land-use information in the Edwards Aquifer. Water Res. 2020, 184, 116211. [CrossRef] 56. Kapoor, V.; Gupta, I.; Pasha, A.B.M.T.; Phan, D. Real-Time Quantitative PCR Measurements of Fecal Indicator Bacteria and Human-Associated Source Tracking Markers in a Texas River following Hurricane Harvey. Environ. Sci. Technol. Lett. 2018, 5, 322–328. [CrossRef] 57. Morshed, S.A.; Arafat, M.; Mokhtarimousavi, S.; Khan, S.S.; Amine, K. 8R Resilience Model: A stakeholder-centered approach of disaster resilience for transportation infrastructure and network. Transp. Eng. 2021, 4, 100058. [CrossRef] 58. Morshed, S.A.; Arafat, M.; Ahmed, A.; Saha, R. Discovering the Commuters’ Assessments on Disaster Resilience of Transportation Infrastructure. In Proceedings of the International Conference on Transportation and Development 2020, Seattle, WA, USA, 26–29 May 2020; pp. 23–34. Atmosphere 2021, 12, 687 18 of 18

59. Pasha, A.B.M.T.; Hinojosa, J.; Phan, D.; Lopez, A.; Kapoor, V. Detection of human fecal pollution in environmental waters using human mitochondrial DNA and correlation with general and human-associated fecal genetic markers. J. Water Heal. 2019, 18, 8–18. [CrossRef] 60. Anjum, M.N.; Ahmad, I.; Ding, Y.; Shangguan, D.; Zaman, M.; Ijaz, M.W.; Sarwar, K.; Han, H.; Yang, M. Assessment of IMERG- V06 Precipitation Product over Different Hydro-Climatic Regimes in the Tianshan Mountains, North-Western China. Remote Sens. 2019, 11, 2314. [CrossRef] 61. Moazami, S.; Najaﬁ, M. A comprehensive evaluation of GPM-IMERG V06 and MRMS with hourly ground-based precipitation observations across Canada. J. Hydrol. 2021, 594, 125929. [CrossRef] 62. Furl, C.; Ghebreyesus, D.; Sharif, H.O. Assessment of the Performance of Satellite-Based Precipitation Products for Flood Events across Diverse Spatial Scales Using GSSHA Modeling System. Geosciences 2018, 8, 191. [CrossRef] 63. Nielsen-Gammon, J.W.; Zhang, F.; Odins, A.M.; Myoung, B. Extreme Rainfall in Texas: Patterns and Predictability. Phys. Geogr. 2005, 26, 340–364. [CrossRef] 64. Lin, Y.; Mitchell, K.E. The NCEP stage II/IV hourly precipitation analyses: Development and applications. In Proceedings of the 19th Conference Hydrology, American Meteorological Society, San Diego, CA, USA, 9–13 January 2005. 65. Gupta, H.V.; Kling, H.; Yilmaz, K.K.; Martinez, G.F. Decomposition of the mean squared error and NSE performance criteria: Implications for improving hydrological modelling. J. Hydrol. 2009, 377, 80–91. [CrossRef] 66. Kling, H.; Fuchs, M.; Paulin, M. Runoff conditions in the upper Danube basin under an ensemble of climate change scenarios. J. Hydrol. 2012, 424–425, 264–277. [CrossRef] 67. Tang, S.; Li, R.; He, J.; Wang, H.; Fan, X.; Yao, S. Comparative Evaluation of the GPM IMERG Early, Late, and Final Hourly Precipitation Products Using the CMPA Data over Sichuan Basin of China. Water 2020, 12, 554. [CrossRef] 68. Asong, Z.E.; Razavi, S.; Wheater, H.S.; Wong, J.S. Evaluation of Integrated Multisatellite Retrievals for GPM (IMERG) over Southern Canada against Ground Precipitation Observations: A Preliminary Assessment. J. Hydrometeorol. 2017, 18, 1033–1050. [CrossRef] 69. He, Z.; Yang, L.; Tian, F.; Ni, G.; Hou, A.; Lu, H. Intercomparisons of Rainfall Estimates from TRMM and GPM Multisatellite Products over the Upper Mekong River Basin. J. Hydrometeorol. 2017, 18, 413–430. [CrossRef] 70. Alsumaiti, T.S.; Hussein, K.; Ghebreyesus, D.T.; Sharif, H.O. Performance of the CMORPH and GPM IMERG Products over the United Arab Emirates. Remote Sens. 2020, 12, 1426. [CrossRef] 71. Bajracharya, S.R.; Palash, W.; Shrestha, M.S.; Khadgi, V.R.; Duo, C.; Das, P.J.; Dorji, C. Systematic Evaluation of Satellite-Based Rainfall Products over the Brahmaputra Basin for Hydrological Applications. Adv. Meteorol. 2015, 2015, 398687. [CrossRef] 72. Liechti, T.C.; Matos, G.D.S.C.; Boillat, J.; Schleiss, A.J. Comparison and evaluation of satellite derived precipitation products for hydrological modeling of the Zambezi River Basin. Hydrol. Earth Syst. Sci. 2012, 16, 489–500. [CrossRef] 73. Milewski, A.; Elkadiri, R.; Durham, M. Assessment and Comparison of TMPA Satellite Precipitation Products in Varying Climatic and Topographic Regimes in Morocco. Remote Sens. 2015, 7, 5697–5717. [CrossRef] 74. Wehbe, Y.; Ghebreyesus, D.; Temimi, M.; Milewski, A.; Al Mandous, A. Assessment of the consistency among global precipitation products over the United Arab Emirates. J. Hydrol. Reg. Stud. 2017, 12, 122–135. [CrossRef] 75. Wehbe, Y.; Temimi, M.; Ghebreyesus, D.T.; Milewski, A.; Norouzi, H.; Ibrahim, E. Consistency of precipitation products over the Arabian Peninsula and interactions with soil moisture and water storage. Hydrol. Sci. J. 2018, 63, 408–425. [CrossRef]