Gerald Corzo Perez | IHE Delft (original) (raw)

Papers by Gerald Corzo Perez

EGU General Assembly Conference Abstracts, Apr 1, 2018

Currently, large-scale drought monitoring is carried out through the implementation of different ... more Currently, large-scale drought monitoring is carried out through the implementation of different drought indicators that use both remote sensing and ground-based data. Among the most promising are those that use remote sensing data variables, mainly soil moisture (SM) and evaporation (E), because of their link with agricultural drought impacts. Such drought indicators are for instance, the Standardized Soil Moisture Index (SSMI) and the Standardized Precipitation Evaporation Index (SPEI), respectively. In the last decade, new formulations on drought indicators based on E, have been proposed, but not yet implemented in large-scale drought monitoring. These include for example, Evapotranspiration Deficit Index (EDI), Evaporative Demand Drought Index (EDDI) and Standardized Evapotranspiration Deficit Index (SEDI). This research aims to conduct a comparative analysis of the spatial characteristics of drought calculated by using two drought indicators based on E: SPEI and SEDI. These characteristics include drought area and number of clusters in drought obtained through the application of Non-Contiguous Drought Area (NCDA) and Contiguous Drought Area (NCDA) approaches. Global Land Evaporation Amsterdam Model (GLEAM) data is used to compute both drought indicators. Results indicate that there is a difference between the shape of the area in drought and the number of groups that compose it, this when using different drought indicator. This study suggests that SEDI presents an improved representation of the overall water availability and therefore could be used for more detailed analysis.

AGU Fall Meeting Abstracts, Dec 1, 2011

Abstract Large-scale models are often applied to study past drought (forced with global reanalysi... more Abstract Large-scale models are often applied to study past drought (forced with global reanalysis datasets) and to assess future drought (using downscaled, bias-corrected forcing from climate models). The EU project WATer and global CHange (WATCH) provides a 0.5 o degree global dataset of meteorological forcing (ie WATCH Forcing Data, WFD), which was used as input for a suite of global hydrological models (GHMs) and land surface models (LSMs). Ten GHMs and LSMs have been run for the second half of the 20th C and seven ...

Hydrology Research, Mar 26, 2021

Precipitation data are useful for the management of water resources as well as flood and drought ... more Precipitation data are useful for the management of water resources as well as flood and drought events. However, precipitation monitoring is sparse and often unreliable in regions with complicated geomorphology. Subsequently, the spatial variability of the precipitation distribution is frequently represented incorrectly. Satellite precipitation data provide an attractive supplement to ground observations. However, satellite data involve errors due to the complexity of the retrieval algorithms and/or the presence of obstacles that affect the infrared observation capability. This work presents a methodology that combines satellite and ground observations leading to improved spatiotemporal mapping and analysis of precipitation. The applied methodology is based on space-time regression kriging. The case study refers to the island of Crete, Greece, for the time period of 2010-2018. Precipitation data from 53 stations are used in combination with satellite images for the reference period. This work introduces an improved spatiotemporal approach for precipitation mapping.

Science of The Total Environment, May 1, 2020

h i g h l i g h t s An analysis on the characterisation of drought dynamics is presented. Locatio... more h i g h l i g h t s An analysis on the characterisation of drought dynamics is presented. Location, spatial paths, direction, and rotation of droughts were calculated. Spatial patterns of the most severe droughts were analysed. Calculated droughts were compared with documented information.

Stochastic Environmental Research and Risk Assessment, Apr 19, 2023

Journal of Hydrology, Dec 1, 2020

A spatiotemporal object-based rainfall analysis method is used to evaluate the hydrological respo... more A spatiotemporal object-based rainfall analysis method is used to evaluate the hydrological response of two systematic satellite error sources for storm estimation in the Capivari catchment, Brazil. This method is called Spatiotemporal Contiguous Object-based Rainfall Analysis (ST-CORA) specifically evaluates the error structure of satellite-based rainfall products using a 3D pattern clustering algorithm. Errors due to location and magnitude in the Near Real-time (NRT) CMORPH product are subtracted by adjusting the shift and the intensity distribution with respect to a storm object obtained from gauge-adjusted weather radar. Synthetic scenarios of each error source are used as forcing for hourly calibrated distributed hydrological 'wflow-sbm' model to evaluate the main sources of systematic errors in the hydrological response. Two types of storm events in the study area are evaluated: short-lived and a long-lived storm. The results indicate that the spatiotemporal characteristics obtained by ST-CORA clearly reflect the main source of errors of the CMORPH storm detection. It is found that location is the main source of error for the short-lived storm event, while volume is the main source in the longlived storm event. The subtraction of both errors leads to an important reduction of the simulated streamflow in the catchment. The method applied can be useful in bias correction schemes for satellite estimations especially for extreme precipitation events. Recently, satellite-based rainfall products (SRP) have widely used in hydrological applications for overcoming the lack of spatial representation of rain-gauges [e.g.] Artan et al. (2007), Nikolopoulos et al. (2013), Stisen and Sandholt (2010), Su et al. (2008). A variety of global-scale satellite-derived rainfall products are freely and openly. Sun et al. (2018) presented a full description of operational and nonoperational SRP. Despite numerous advances, satellite products are subject to several systematic and random errors from multiple sources [e.g] Hu et al.

A sufficient data length can play an important role in a proper estimation drought index, leading... more A sufficient data length can play an important role in a proper estimation drought index, leading to a better appraisal for drought risk reduction. The South Central Region of Vietnam is one of drought prone areas but it has poor data conditions. A collection of meteorological data in the study area during a period of 38 years 1977–2014 found out a fact that there existed missing values in 10 out of 30 collected rainfall stations and 4 out of 13 collected temperature stations. Therefore, this study aims at evaluating the influence of three different infilling techniques (Inverse Distance Weighting, Multi-Linear Regression, and Artificial Neural Network) on 1-month Standardized Precipitation Evapotranspiration Index (SPEI1) drought indicator for the given region. The performance on rainfall and temperature infilling indicated that ANN technique achieved lower errors between observations and predictions than others. Infilled rainfall and temperature generated from different infilling techniques were then combined with the available data to calculate SPEI1. The results showed that infilling techniques seem to create the SPEI1 time series which have higher number of drought events, in comparison with that time series containing the observation values only. Otherwise, drought index seems to be insensitive to different infilling techniques.

EGU General Assembly Conference Abstracts, Apr 1, 2019

Soil Science, Jun 1, 2016

The formulation of a model that can reliably simulate the temporal groundwater level fluctuations... more The formulation of a model that can reliably simulate the temporal groundwater level fluctuations of an aquifer is important for effective water resource management and for the prevention of possible desertification effects. Mires Basin at the island of Crete, Greece, is part of a major watershed with significantly reduced groundwater resources because of overexploitation during the past 30 years. In this work, the interannual variability of groundwater level is modeled with a discrete time autoregressive exogenous variable (ARX) model that is based on physical grounds (soilwater balance equation). Precipitation surplus is used as an exogenous variable in the ARX model. Three new modified versions of the original form of the ARX model are proposed and investigated: the first considers a larger time scale; the second considers a larger time delay in terms of the groundwater level input; and the third considers the groundwater level difference between the last two hydrological years, which is incorporated in the model as a third input variable. Modeling results for the time series of the spatially averaged groundwater level show very good agreement, after an initial adaptation period, with measured data. Among the three modified versions of the original ARX model considered in this work, the third model version shows significantly better agreement with measured data.

EPiC series in engineering, Sep 20, 2018

Due to the underlying characteristics of drought, monitoring of its spatio-temporal development i... more Due to the underlying characteristics of drought, monitoring of its spatio-temporal development is difficult. Last decades, drought monitoring have been increasingly developed, however, including its spatio-temporal dynamics is still a challenge. This study proposes a method to monitor drought by tracking its spatial extent. A methodology to build drought trajectories is introduced, which is put in the framework of machine learning (ML) for drought prediction. Steps for trajectories calculation are (1) spatial areas computation, (2) centroids localization, and (3) centroids linkage. The spatiotemporal analysis performed here follows the Contiguous Drought Area (CDA) analysis. The methodology is illustrated using grid data from the Standardized Precipitation Evaporation Index (SPEI) Global Drought Monitor over India (1901-2013), as an example. Results show regions where drought with considerable coverage tend to occur, and suggest possible concurrent routes. Tracks of six of the most severe reported droughts were analysed. In all of them, areas overlap considerably over time, which suggest that drought remains in the same region for a period of time. Years with the largest drought areas were 2000 and 2002, which coincide with documented information presented. Further research is under development to setup the ML model to predict the track of drought.

EGU General Assembly Conference Abstracts, Apr 1, 2017

Urban flood mostly is pluvial flood, caused by high rainfall intensities combine... more Urban flood mostly is pluvial flood, caused by high rainfall intensities combined with the unsuitable drainage system and land cover. Because of the heterogeneous drainage system and the storm distribution dynamics, urban floods are rapid and spontaneous in space and time. Flood risk analysis was created to understand and assess the flood behavior, manage and mitigate the flood damage. However, flood risk assessments recently only have focused on the spatial distribution of the flood while temporal flood evolution in urban area is still an open question. This research aims to provide a spatio-temporal analysis of the urban flood by implementing the simplified model-based representation of flood evolution/development in space and time (spatiotemporal patterns) including time to flood the manhole, the location, spatial and temporal sequences of flood.To specify, 3 precipitation distribution patterns were collected from rainfall incidents in 2008, 2019, 2020, then extrapolated to create 21 scenarios following the return periods (i.e. 1.25-year, 2-year, 2.5-year, 5-year, 10-year, 20-year, 50-year). In each scenario, the dynamics of flood was estimated using the urban drainage system by SWMM5. Case study (Do Lo, Yen Nghia, Ha Dong, Ha Noi, Viet Nam) was divided into 115 sub-catchtments based on the Digital Elevation Model (DEM) and the drainage system map of this area. Land cover was created based on the LANDSAT images. Domestic waste water distribution was included in the model. The model is validated with the extreme events in 2020 and 2022.A spatio-temporal risk map was generated to show the flooding spatio-temporal sequences and non-flood region. Flood evolution on the time scale was shown in this map. The rate of flood change diagrams shows the flood responses from urban areas which vary from 1 to 107 mins in different scenarios.Rain gauge distribution sensitivity is examined under ranges of rain gauge distribution combinations in term of space and time.

Machine learning and specifically deep learning has been applied in solving nume... more Machine learning and specifically deep learning has been applied in solving numerous hydrology related problems in the past. Furthermore, extensive research has been done on the evaluation and comparison of performances of different Machine learning techniques applied in solving hydrology related problems. In this research, the possible reasons behind these performance variations are being assessed. The performance of recently introduced deep learning techniques for rainfall-runoff modelling are being evaluated by looking in to the possible modelling set-up and training procedures. Therefore, model set-up and training procedures such as: normalization techniques, input variable selection (feature selection), sampling techniques, model complexity, optimization techniques and random initialization of weights are being examined closely in order to improve the performances of different deep learning techniques for rainfall-runoff modelling. As a result, this study is trying to answer whether these factors have significant effect on the model accuracy. The experiments are being conducted on different deep learning models such as: LSTMs, GRUs and MLPs as well as non-deep learning models such as: XGBoost, Random Forest, Linear Regression and Na&#239;ve models. Deep learning frameworks including TensorFlow and Keras are being implemented on Python. For better generalization, study areas from three different climatic zones namely: Bagmati catchment in Nepal, Yuna catchment in Dominican Republic and Magdalena catchment in Colombia are chosen to implement this experimental research. Additionally, in situ meteorological and stream flow data are being used for the rainfall-runoff modelling. The preliminary model results show that model performances in case of Bagmati catchment are higher as compared to the other catchments. The LSTMs and MLPs are performing good with NSE values of 0.71 and 0.72 respectively. Most importantly, the linear regression model was outperforming the other models with NSE up to 0.75 in case of considering 6 days lagged rainfall input. This implies the relationship between daily rainfall and runoff data from Bagmati catchment may not be as complex. On the contrary, the 3-hourly data from Yuna catchment shows results with lower values for the performance metrics. This may be an indication of more complex relationships within the Yuna catchment. This research provides key elements of the modelling process, especially in setting up and training deep learning models for rainfall-runoff modelling. The comparative analysis performed here, provides a basis of performance variations on different basins. This work contributes to the experiences in understanding machine learning requirements for different types of river basins.

Hydrological forecasting is of global importance, especially with the spotted in... more Hydrological forecasting is of global importance, especially with the spotted increasing trend of flood-related disasters as seen in the last two decades.&#160; The causative rainfall events of these extreme events are primarily analysed in a one-dimensional method. However, through an object-based approach, more data on these rainfall fields can be generated and studied to link them to the hydrological response observed. Through an object-based methodology ST-CORA, features from rate of change of rain intensity in space and time can be extracted by simple visual inspection. Every side of an object provides time variations that can be used as images that contain features not easy to extract. In general, rainfall events in previous studies have used aggregated information, like the duration, area, volume, maximum intensity, and the centroid. In this work, more information is captured that describes the spatial and temporal properties of the event. The main objective of this research is to use these 3D objects and their features with a deep learning model to produce a 15-day hydrological probabilistic forecast for flood prediction. A calibrated version of a large-scale hydrological model (MGB) is used to study an Amazon subbasin. The model is forced with the 50-member perturbed forecast from the TIGGE dataset for the period 2006 to 2014 (from ECMWF). The purpose of using the large-scale model is to better capture the spatio-temporal characteristics over a wider area in an effort to reduce the uncertainty in the analysis. For data-driven models, there is a need for sufficiently large databases, in this case for both the causative rainfall events and the observed hydrological responses. As such, the first two steps relate to the data generation. The first database is developed from the daily streamflow which is generated from the calibrated hydrological model at specific locations of interest with the known higher performance metrics. Second, the ST-CORA methodology is applied to extract the features from the rainfall events in order to develop a database of the rainfall objects. Third, an analysis on the statistics of the features of the objects to understand the rainfall which occurs within the study area. The final part of the research involves the effective use of these features and objects with a deep learning model. From the average annual rainfall from 2001 to 2020, three distinct precipitation patterns are observed. For the streamflow, the subbasin shows a relatively fast response which is captured within a 15-day window. A convolutional LSTM deep learning model is developed to handle 3D rainfall objects as sequences of images representing space time sequences. The outcome of this research contributes to the end-to-end deep learning model which receives the forecasted rainfall as objects and generates a corresponding hydrograph at the area of interest for which it has been trained. A potential contribution of this Conv-LSTM network is that it may provide an efficient and automated approach for streamflow forecasting in basins where there is known complexity and non-linearity, which is especially useful for early warning systems.