188 research outputs found
Assessing the suitability of statistical downscaling approaches for seasonal forecasting in Senegal
No full textThis work tests the suitability of statistical downscaling (SD) approaches to generate local seasonal forecasts of daily maximum temperature and precipitation for a set of selected stations in Senegal for the July–August–September season during the period 1979–2000. Two-month lead raw daily maximum temperature and precipitation from the five models included in the ENSEMBLES seasonal hindcast are compared against the corresponding downscaled predictions, which are obtained by applying the analog technique based on two different types of predictors: the direct surface variables and a combination of appropriate upper-air variables. Beyond correcting the large biases of the low-resolution raw model outputs, SD is found to add noteworthy value in terms of forecast association (as measured by interannual correlation), providing thus suitable (i.e. calibrated) predictions at the local-scale needed for practical applications, which means a clear advantage for the end-users of seasonal forecasts over the area of study. Moreover, a recommendation on the adequacy of surface (large-scale) predictors for SD of maximum temperature (precipitation) is also given.This study was supported by the EU projects QWeCI and EUPORIAS, funded by the European Commission through the Seventh Framework Programme for Research under Grant Agreements 243964 and 308291, respectively. TPeer Reviewe
Assessment of model drifts in seasonal forecasting: sensitivity to ensemble size and implications for bias correction
No full textDespite its systematic presence in state‐of‐the‐art seasonal forecasts, the model drift (leadtime‐dependent bias) has been seldom studied to date. To fill this gap, this work analyzes its spatiotemporal distribution, and its sensitivity to the ensemble size in temperature and precipitation forecasts. Our results indicate that model continues to drift well beyond the first month after initialization, leading to significant, highly space‐ and time‐varying drifts over vast regions of the world. Nevertheless, small ensembles (less than 10 members) are enough to robustly estimate the mean model drift and its year‐to‐year fluctuations in skillful regions. Differently, in regions of low model skill, larger ensembles are required to appropriately characterize this interannual variability, which is often larger than the drift itself. This points out a necessity to develop new strategies that allow for efficiently dealing with model drift, especially when bias correcting seasonal forecasts—most of the techniques used to this aim rely on the assumption of stationary model errors. We demonstrate here that the use of moving windows can help to remove not only the mean forecast bias but also the unwanted effects coming out from the drift, which can lead to important intraseasonal biases if it is not properly taken into account. The results from this work can help to identify the nature and causes of some of the systematic errors in current coupled models and can have large implications for a wide community of users who need long, continuous unbiased seasonal forecasts to run their impact models.This study was supported by the EU projects EUPORIAS (EUropean Provision Of Regional Impact Assessment on a Seasonal‐to‐decadal timescales) and SPECS (Seasonal‐to‐decadal climate Prediction for the improvement of the European Climate Services), funded by the European Commission's Seventh Framework Research Programme through Grant Agreements 308291 and 308378, respectively
Extreme precipitation on consecutive days occurs more often in a warming climate
No full textExtreme precipitation occurring on consecutive days may substantially increase the risk of related impacts, but changes in such events have not been studied at a global scale. Here we use a unique global dataset based on in situ observations and multimodel historical and future simulations to analyze the changes in the frequency of extreme precipitation on consecutive days (EPCD). We further disentangle the relative contributions of variations in precipitation intensity and temporal correlation of extreme precipitation to understand the processes that drive the changes in EPCD. Observations and climate model simulations show that the frequency of EPCD is increasing in most land regions, in particular, in North America, Europe, and the Northern Hemisphere high latitudes. These increases are primarily a consequence of increasing precipitation intensity, but changes in the temporal correlation of extreme precipitation regionally amplify or reduce the effects of intensity changes. Changes are larger in simulations with a stronger warming signal, suggesting that further increases in EPCD are expected for the future under continued climate warming.We acknowledge support from the National Key R&D Program of China (2019YFC0409101), Science and Technology Development Plan of Jilin Province (20190201291JC), the Joint Fund of National Natural Science Foundation of China (U19A2023), the Fundamental Research Funds for the Central Universities (2412020FZ002), and 2236 Co-Funded Brain Circulation Scheme2 (CoCirculation2) of TÜBİTAK (121C054). M.G.D. acknowledges support by the Horizon 2020 EUCP project under Grant Agreement 776613 and by the Spanish Ministry for the Economy, Industry and Competitiveness Ramón y Cajal 2017 Grant Reference RYC-2017-22964.Peer Reviewed"Article signat per 28 autors/es: Haibo Du, Markus G. Donat, Shengwei Zong, Lisa V. Alexander, Rodrigo Manzanas, Andries Kruger, Gwangyong Choi, Jim Salinger, Hong S. He, Mai-He Li, Fumiaki Fujibe, Banzragch Nandintsetseg, Shafiqur Rehman, Farhat Abbas, Matilde Rusticucci, Arvind Srivastava, Panmao Zhai, Tanya Lippmann, Ibouraïma Yabi, Michael C. Stambaugh, Shengzhong Wang, Altangerel Batbold, Priscilla Teles de Oliveira, Muhammad Adrees, Wei Hou, Claudio Moises Santos e Silva, Paulo Sergio Lucio, and Zhengfang Wu "Postprint (published version
Biometeorología en el Grupo de Meteorología de Santander
No full textTrabajo presentado al Primer Workshop en Biometeorología Ciudad de Santander celebrado en octubre de 2011.Peer reviewe
Avances en la regionalización estadística de escenarios de cambio climático para precipitación basados en técnicas de aprendizaje automático
No full textA pesar de ser la principal herramienta para estudiar el cambio climático, los modelos globales de clima (GCM) siguen teniendo una resolución espacial limitada y presentan errores sistemáticos considerables con respecto al clima observado. La regionalización estadística pretende resolver este problema aprendiendo relaciones empíricas entre variables de larga escala, bien reproducidas por los GCM (por ejemplo, los vientos sinópticos o el geopotencial), y observaciones locales de la variable en superficie de interés, como la precipitación, objeto de esta tesis. Proponemos una serie de desarrollos novedosos que permiten mejorar la consistencia de los campos regionalizados y producir escenarios regionales plausibles de cambio climático. Los resultados de esta tesis tienen importantes implicaciones para los diferentes sectores que necesitan información fiable de precipitación para llevar a cabo sus evaluaciones de impactos.Even though they are the main tool to study climate change, global climate models (GCMs) still have a limited spatial resolution and exhibit considerable systematic errors with respect to the observed climate. Statistical downscaling aims to solve this issue by learning empirical relationships between large-scale variables, well reproduced by GCMs (such as synoptic winds or geopotential), and local observations of the target surface variable, such as precipitation, the focus of this thesis. We propose a series of novel developments which allow for improving the consistency of the downscaled fields and producing plausible local-to-regional climate change scenarios. The results of this thesis have important implications for the different sectors in need of reliable precipitation information to undertake their impact assessments
On the suitability of the Random Forest technique for the statistical downscaling of climate change projections
No full textRESUMEN: Los Modelos Globales del Clima (GCM, por sus siglas en inglés) son las herramientas utilizadas hoy en día para la simulación del clima en las diferentes escalas temporales (desde 3-5 días vista hasta final de siglo). Debido a ciertas limitaciones físicas y a su alto coste computacional, la resolución espacial de los GCM actuales todavía es insuficiente. Para ayudar a solventar esta limitación se ha desarrollado en las últimas décadas una extensa batería de técnicas de regionalización (o downscaling).
En el marco de la iniciativa europea VALUE (http://www.value-cost.eu/), cuyo objetivo es el de comparar diferentes estrategias de regionalizacion para el estudio del cambio climático (Gutiérrez et al., 2018), han presentado recientemente la intercomparación de métodos de regionalizacion estadística más extensa (más de 50 técnicas) y rigurosa hasta la fecha sobre 86 estaciones repartidas por Europa.
En este Trabajo de Fin de Máster se perfila a random forest como otra opción válida a las técnicas presentes en Gutiérrez et al. (2018). Además se muestra como random forest también es otra opción viable para la regionalización estadística de proyecciones de cambio climático sin necesidad de una fase de selección de variables, obteniendo resultados más realistas que otras técnicas.ABSTRACT: Global Climate Models (GCMs) are the tools used today to simulate climate at different time scales (from 3-5 days seen until the end of the century). Due to certain physical limitations and their high computational cost, the spatial resolution of the current GCMs is still insuficient. In order too solve this limitation, an extensive battery of downscaling techniques has been developed in recent decades.
In the European initiative VALUE (http://www.value-cost.eu/), whose objective is to compare different downscaling strategies for the study of climate change (Gutiérrez et al., 2018), recently presented the intercomparison of methods of statistical downscaling most extensive (more than 50 techniques) and rigorous to date over 86 stations spread across Europe.
In this Master's Thesis, random forest is outlined as another valid option to the techniques present in Gutiérrez et al. (2018). It is also shown that random forest is also another viable option for the statistical downscaling of climate change projections without the need for a phase of selection of variables, obtaining more realistic results than other techniques.Máster en Ciencia de Dato
Modeling streamflow using multiple precipitation products in a topographically complex catchment
No full textPrecipitation is of primary importance in hydrological modeling and streamflow prediction. However, lack of gauge stations for long-term precipitation data, particularly in the data-scarce Chitral River Basin (CRB) of Pakistan and other parts in the developing world, is a hindrance to understand surface water hydrology. Therefore, this study aims to assess different sources of precipitation data for streamflow prediction in the CRB. A modified version of the conceptual and semi-distributed hydrological model Hydrologiska Byråns Vattenbalansavdelning (HBV) known as HBV-light is used in this study to model streamflow by forcing it with precipitation inputs of different Precipitation Products (PPs). These PPs include APHRODITE (V1101, V1801R1), CHIRPS V2.0, CPC-Global, ERA5, GPCC V.2018 (V2), GPCP-1DD V1.2, PERSIANN, CHRS CCS, CHRS CDR and TRMM (3B42V7). The model was calibrated and validated for two periods (1995–2005 and 2007–2013, respectively), and showed good performance during both periods. Prior to assessing the performance of these PPs to simulate observed streamflow, they were assessed against gauged precipitation. Results of this study showed that APHRODITE-based precipitation performed better than other precipitation products in the simulation of precipitation characteristics in the study region. Multiple efficiency evaluation metrics including KGE, NSE, and PBIAS were employed to assess streamflow prediction capability of different products. Results indicated that APHRODITE outperformed all other PPs (KGE 0.89) in terms of simulating observed streamflow in the CRB. The CPC Global precipitation product (KGE 0.71) was found to be the least suitable product for hydrological modeling in the CRB. This study provides useful guidance for the selection and application of gridded precipitation products for long-term continuous streamflow prediction in the CRB.Full Tex
Modeling implications of climate induced streamflow changes on the fish species of the Soan River, Pakistan
No full textClimate change has significantly impacted the hydrological cycle in the rivers of Pakistan and the streamflow regimes of different rivers have witnessed noticeable flow alteration. These changes in streamflow affect aquatic biodiversity (e.g., freshwater fish species) and the productivity of freshwater ecosystems. This study, therefore, evaluates the streamflow alterations and their impacts on the fish species at the upstream and downstream of the Soan River in Pakistan. The hydrological model HBV light was calibrated and validated for both up- (Chirah) and down- (Dhoke Pathan) stream gauged stations. The model was then forced with an ensemble of NEX-GDDP Global Climate Models (GCMs) to simulate historic and future streamflow. Afterwards, changes in the streamflow characteristics for two future periods against a historic one were assessed. Different ecologically relevant streamflow indices were used. The base flow at upstream and downstream was projected to increase under both the RCP 4.5 and RCP 8.5 emission scenarios for mid and end of century periods. Under the RCP 8.5, base flow was projected to increase with more inter-annual variability at upstream station, and changes depicted less variability for downstream under both emission scenarios. A number of high flow pulses were projected to decrease under both RCP 4.5 and RCP 8.5 in both future periods for upstream and downstream; however, high variability was depicted for upstream than downstream area. The duration of high flow pulses was also projected to increase under the RCP 8.5 for end of century period for both upstream and downstream. These projections of streamflow characteristics may have a marked (positive) influence on the habitat and ecological conditions for the fish species of the Soan River.Full Tex
Efecto del tipo de corte y la temperatura de almacenamiento en la actividad metabólica en manzanas "red delicious" y "cripps pink" mínimamente procesadas en fresco
No full textMemoria para optar al Título Profesional de Ingeniero Agrónomo mención FruticulturaSe investigó el efecto del corte (cascos, cubos y rodajas) y de la temperatura de
almacenamiento (5 y 8 ºC) sobre la actividad metabólica en manzanas mínimamente
procesadas en fresco (MPF). Para llevar esto a cabo, se realizaron dos ensayos con dos
variedades de manzanas (‘Red Delicious’ y ‘Cripps Pink’). Los trozos de manzanas fueron
envasados en tarrinas de polietileno con tapa (200 g) y almacenados en cámaras de
refrigeración durante 8 días. Se evaluó la tasa respiratoria, luminosidad (L), croma (C*),
tono (Hab), firmeza, concentración de sólidos solubles totales (SST), pH, acidez titulable y
parámetros sensoriales.
En el Ensayo I (Red Delicious), el cortado provocó un aumento significativo en la tasa
respiratoria de todos los tratamientos estudiados, este incremento fue más notorio a 8 ºC
mostrando tasas superiores a 30 mgCO2kg-1h-1 al día 8; a 5 ºC las tasas respiratorias
alcanzaron valores entorno a los 20 mgCO2kg-1h-1, destacando la importancia de la
temperatura de conservación sobre la actividad metabólica de los productos MPF. No se
observaron diferencias significativas en el avance del pardeamiento de la fruta almacenada,
sin embargo el tratamiento cascos a 8 ºC presentó la mayor disminución en los parámetros
L (17%) y Hab (14%) respectivamente. La pérdida de firmeza se incrementó con el periodo
de almacenamiento, siendo mayor en cascos a 5 y 8 ºC.
El mayor grado de corte y la temperatura de almacenamiento a 8 °C provocaron un
aumento significativo de la tasa respiratoria. La conservación a bajas temperaturas, en esta
experiencia, no fue efectiva en el control de los diferentes problemas causados por el
procesamiento. Por lo que se aconseja la utilización de métodos alternativos para
complementar la refrigeración, como antipardeantes, sales de calcio, EAM, etc.
En el Ensayo II (Cripps Pink), los cubos conservados a 8 ºC obtuvieron tasas respiratorias
más altas, registrando 46 mgCO2kg-1h-1 al día 8; y mayor grado de pardeamiento con
disminución de los parámetros L (6%) y Hab (4%). La firmeza disminuyó a través del
tiempo, sin presentar diferencias significativas entre tratamientos.
Cripps Pink mantiene la calidad del producto durante casi todo el ensayo, siendo una
opción para el desarrollo de manzanas MPF. No obstante se recomienda el uso de métodos
alternativos para complementar la refrigeración, con el fin de extender la vida útil y periodo
de comercialización a un tiempo mayor a 8 días
Automated wildfire season detection at a global scale: Application for the development of a predictive system of fire activity
No full textRESUMEN: En la primera parte de este trabajo describimos un procedimiento de aprendizaje automático no supervisado basado en la técnica de Gaussian Mixtures con el objetivo de determinar la estación de fuegos a escala global a partir de datos de satélite de área quemada a una resolución espacial de 0.5◦ . Nuestro método permite la identificación de ciclos anuales de tipo unimodal y multimodal, así como la determinación del inicio, fin y momento de máxima actividad de incendios, con la ventaja adicional de proporcionar un procedimiento totalmente automatizado que puede ser utilizado a múltiples escalas espacio-temporales. La caracterización de la estación de fuegos aquí presentada desvela un patrón inequívoco y espacialmente coherente, que es consistente con estudios previos sobre la estacionalidad de incendios. Nuestro método puede ser fácilmente adaptado por el usuario mediante el ajuste de unos pocos parámetros sencillos para adecuarlo a bases de datos de incendios de distinta naturaleza, extensión geográfica y resolución espacio-temporal. A continuación, se parte de la zonificación proporcionada por las Gaussian mixtures para el desarrollo de modelos predictivos de área quemada —durante la estación de fuegos principal— a nivel de clúster, tomando como única información predictora una serie de índices que representan los patrones de teleconexión climática más relevantes a escala global. Para ello se consideran modelos lineales, random forest y k-vecinos cercanos, en cuyo ajuste se aplican técnicas de validación cruzada. Nuestros resultados muestran que, cuando se consideran como predictores aquellos índices que están más fuertemente correlacionados con el área quemada, incluso los modelos lineales más simples son capaces de proporcionar predicciones de área quemada fiables en determinadas zonas del planeta. Este trabajo abre la puerta para el futuro desarrollo e implementación de un sistema operativo de alerta temprana de incendios en base a modelos climáticos de predicción estacional. Todos los análisis realizados son totalmente reproducibles a través de los datos post-procesados, scripts y notebooks que están disponibles en un repositorio público.ABSTRACT: In the first part of this project, we describe an unsupervised machine learning procedure based on Gaussian mixtures in order to determine the fire season at a global scale, using remotely sensed data of burned area at a 0.5◦ spatial resolution. Our results allow the identification of unimodal and multimodal annual cycles as well as the start, end and timing of bulk fire activity, with the added advantage of providing a fully automated procedure that can be used at multiple temporal and spatial scales. The fire season characterization presented unveils an unambiguous, spatially coherent pattern, consistent with previous studies on fire seasonality. Our method can be easily tuned by the user through the manipulation of a few simple parameters in order to accommodate fire databases of varying nature, geographical extent and spatial and temporal resolutions. Next, using the fire season definition given by the Gaussian mixtures, predictive models of burned area are developed at a global scale using a set of the most relevant climate teleconection indices as predictors. We consider linear models, random forests and k-nearest neighbours, fitted following a cross-validation setup. Our results show that, when only the most correlated indices with the fire intensity are considered as predictors, even the simplest linear models are able to give accurate predictions in certain parts of the world. This study paves the way for the implementation of an operational early-warning wildfire system based on seasonal forecasting climate models. All the analyses undertaken are fully reproducible through the post-processed data, scripts and notebooks available through a dedicated open repository.Máster en Ciencia de Dato
- …
