87 Figure 20 Magnification of the continuous wavelet cross-spectrum of XWT and wavelet coherence scalogram for precipitation and PWV in HV1. a) Observed XWT, (b) Observed coherence, (d) Modeled XWT (d) Modeled coherence. 80 Table 9 Main statistical indicators of the studied stations for observed and modeled PWV and precipitation, in 2014.
Importance of the study
Objectives
Outline of the doctoral thesis and study area
Similarly, analyzing specific intense rain events revealed the existence of PWV peaks that preceded intense rain events by 11 hours, except at the Amazon station, where the recorded peak was lower. When analyzing specific intense rain events, the model showed similar behavior, with delays of less than 3 hours.
- Introduction
- Study area and data
- Materials and Methods
- P, PWV and SST statistical metrics
- Relationships between intense rainfall rates, PWV and SST
- Spatial analysis
- Results and discussion
- Monthly analysis
- Statistical evaluation of RID, PWV and SST
- Functional relationships between rainfall, PWV and SST
- Spatial relationships
- Conclusions and discussion
A statistical analysis of the investigated variables (RID, PWV and SST) is presented in section 3.1 to find the model for the maximum precipitation values. Then, the selection of the best model and the confidence intervals were found according to the model selection method.
Introduction
The troposphere creates variable delays (ZTD for Zenit Total Delay) that can be quantified in GNSS analysis (Bevis et al. Finally, researchers in Argentina's Mendoza Valley (~769 m.a.s.l.) detected an increase in PWV before severe storms, but not a definitive lag ( Calori et al., 2016).
Geographical Location of the Measurement Network and Methods
- GNSS-ZTD Data
- Meteorological Data
- Calculating ZTD from GNSS Data
- Quality Control and Data Synchronization
- Estimation of the PWV from the ZTD
- Conditioning of the Resulting PWV and Rain Data for Harmonic Analysis
- Harmonic Analysis by Descriptive Statistics and Wavelets
- Convection Analysis Using Satellite Images
- Meteorological Anomalies of the Event’s Precipitation Threshold
The constant κ depends on the vertical integral of vapor density and temperature and has an accuracy of 2% (Brenot et al., 2014). In other words, it is an important tool to use to analyze the lead and lag of the set of interest. On the other hand, their sample has distinct peaks, as in the case of rain.
The mathematical steps of how to obtain the statistical significance of the wavelet coherence WTC are very well explained in (Ge, 2008).
Results
- Meteorological Description of the Stations’ Location
- Continuous and Discrete Wavelet Lead-Lag Analysis
- Analysis of Convective Clouds Using Satellite Images
- Analysis of Meteorological Anomalies
Continuous wavelet analysis of available rainfall and PWV data was performed for each station. The phase content of rain and PWV can be determined by the direction of the arrows on the plot. The Wavelet Toolbox (Mathworks, 2021) was very useful to decompose each of the datasets of interest into rain and PWV.
The sign of the lag obtained confirms that the peak preceding the rain is from the PWV.
Discussion
In Figure 17b, a fairly regular behavior with respect to wind speed is observed, with the exception of the coastal station, where a clear increase in wind speed is seen in the presence of a precipitation event, possibly due to the effect of the sea breeze that they can generate significant winds (Mapes et al., 2003). Finally, when analyzing Figure 17c, it can be observed that a generalized behavior is characterized by a decrease in temperature as a result of the rain event, a phenomenon explained by the subsequent evaporation of precipitation and, therefore, the absorption of latent heat from the environment (Mapes et al., 2003; Houze, 2014; Yepes et al., 2020; Ruiz-Hernández et al., 2021). This influence of the diurnal cycle favors the recharge of PWV before the occurrence of an intense rainfall event, and even if it is not an intense rain event, there are preferred hours for PWV to recharge so that it is discharged during the hours of most likely rainfall, in accordance with (Meza et al., 2020).
The effect of the diurnal cycle is ubiquitous for PWV in the coastal and Andean regions, as shown in Appendix A and Figure 7, with characteristic periods of 12 and 24 h reported (Torri et al., 2019).
Conclusions
This fact shows the importance of PWV dynamics for the occurrence of intense rain events, in accordance with previous investigations carried out in other coastal areas such as Portugal (Benevides et al., 2015), Italy (Bonafoni & Biondi, 2016), and China (Yao et al., 2017; Zhao et al., 2020). These types of studies in high-altitude areas have been little reported (Meza et al., 2020) and show the importance not only of the occurrence of PWV peaks, but also of other factors such as topography in the occurrence of intense precipitation events. The smallest delay (about 6 hours) was recorded in the Amazon area, which coincides with previous research (Adams et al., 2013b; Sapucci et al., 2019).
In Figure 9b, a fairly regular behavior with respect to wind speed is observed, except at the coastal station, where a marked increase in wind speed is observed in the presence of a rainfall event, possibly due to the effect of the sea breeze which could generate significant wind gusts (Mapes et al., 2003).
Introduction
Disasters caused by extreme rainfall are still considered the deadliest global natural hazards (Pielke et al., 2013) despite improvements in the understanding of rainfall phenomena and the use of many forecasting tools. When small spatial features of the region are difficult to represent well, such as in mountainous areas of high-slope areas, negligible biases have been reported in NWP simulations of precipitation (Junquas et al., 2018; Ruiz-Hernández et al., 2021). NWP models have focused on the study of PWV derived from the Global Navigation Satellite System (GNSS), as it produces a significant improvement in the prediction of precipitation and its intense events when directly assimilated into the forecast model (Shoji et al., 2009 ; Risanto et al., 2021).
It is clear that the study of PWV is a key factor for precipitation estimation.
Data, area of study and methodology
- GNSS network data of tropospheric delays and in-situ meteorological data
- WRF modeled data
- Quality control of in situ-data of ZTD and precipitation
- Estimation of the PWV from ZTD
- Wavelet based bi-variate analysis of PWV and Precipitation Data: observed
- Lead-lag analysis of the intense precipitation
- Statistical analysis of the model behavior for seasonal and diurnal cycles
This is the case with the PWV where some data are not filled in with respect to the pronounced precipitation peaks. However, no erroneous data or outliers were identified in any of the modeled WRF series. Also, these libraries apply the Morlet wavelet for the calculation of the XWT, which ensures a good balance location both in time and frequency.
If the strings are lagging behind each other, the absolute value of the angle must be greater than π/2.
Results
- Statistical indicators
- Harmonic relationship between modeled and observed PWV and intense
- General performance of the WRF model for the estimation of precipitation and
As in the observed case, the modeled precipitation cross spectrum follows the most intense precipitation events. However, in the case of the diurnal cycle, one can see the first discrepancy between the model and the observed PWV data for the case of HV1 (Fig. 22c). Finally, in the case of HMA ( Fig. 24d ) at the daytime level, any other discrepancy between modeled and observed values is hardly significant, apart from the apparent bias of the data.
In the case of HV1, the appropriate location of the data is appreciated (Figure 25a), with better.
Discussion
Regarding the seasonal behavior of the model, in the case of PWV, HV1 has an adequate reproduction of the seasonal values appreciated (Fig. 22a). When performing the harmonic analysis, it can be clearly seen how the observed values of the correlation and the coherence of the observed. When analyzing the behavior of PWV in HMA, one of the strengths of harmonic analysis is seen, which allows us to analyze the consistency and correlation between the observed and modeled data, regardless of systematic errors such as bias.
Accurate estimation of heavy precipitation events is one of the most difficult NWP problems to solve (Herman & . Schumacher, 2016).
Conclusions
In the same way the data resolution determines the periods found, a better temporal resolution could analyze shorter periods with better detail. Therefore, the methods for analyzing and discussing the results given in this research can be very useful as a way to explain the main deviations in the PWV estimate, especially in locating the presence of atmospheric periods of charge and discharge that depend on the diurnal cycle of the atmosphere and precipitation and in the area , which could be as complex as the one studied in this research. Several studies focus on the proper modeling of precipitation related to topography or statistical strategies to improve model evaluation (Heredia et al., 2019; Mourre et al., 2016); or focus on PWV modeling (González et al., 2013; Qian et al., 2020).
Even if the model could not be changed, the subsequent addition and/or correction of these reasonable interactions, such as fill and discharge periods, could be a valuable strategy in the future part of the PWV estimation process and thus in the precipitation estimation.
Synthesis
However, it is important to note that when this phenomenon occurs, the hourly difference or lag between PWV and rain is 11 h, consistent with the phase analysis of the covariant periods of PWV and rain being in opposition. To check whether these interactions are correctly assimilated in models such as WRF, Chapter 4 analyzes the relationship between PWV and precipitation and the NWP model of Heredia et al. Despite the good performance of the model, which is able to reproduce quite well the climatological behavior of precipitation in the studied areas, the estimation of PWV presents errors, first in the estimation of the true magnitude, mainly the bias deviation, as described in the Station Antisana.
Apparently, the WRF model considers the behavior of PWV and rain to be in phase or with a minimal lag (less than 3 h, as shown in Fig. 24 ), rather than the results obtained with the observed data, which claim the opposite.
Future research
A GPS network for tropospheric tomography within the framework of the Mediterranean Hydrometeorological Observatory Cévennes-Vivarais (Southeastern France). Investigation of the relationship between water vapor field development and rain rate based on 5 years of measurements at a mid-latitude location. Validation of the WRF model for estimating precipitable water vapor at the ali observatory on the Tibetan Plateau.
Harmonic analysis of the relationship between GNSS precipitable water vapor and heavy precipitation over the northwestern equatorial coast, Andes, and Amazon regions.
María Sheila Fabiola Serrano Vincenti Appendix H. Continuous wavelet cross-spectrum XWT and wavelet coherence scalogram of precipitation in HMA for the entire year 2014. She graduated as a physicist and a master's degree in physics from the Department of Physics of the National Polytechnic School in Ecuador. An empirical model for precipitation maxima conditioned to tropospheric water vapor over the eastern Pacific.
Assessing the impact of higher temperatures in mortality risk indices in Ecuador until 2070 due to climate change, Frontiers in Earth Science, section Atmospheric Science 10.3389/feart.