Tesi etd-05152024-003218
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
AHMED, JEMAL SEID
URN
etd-05152024-003218
Titolo
Rainfall Analysis and Prediction for Agricultural Decision-Making in Ethiopia
Settore scientifico disciplinare
FIS/06
Corso di studi
Istituto di Scienze della Vita - PH.D. IN AGROBIODIVERSITY
Commissione
relatore Prof. DELL'ACQUA, MATTEO
Parole chiave
- CHIRPS
- Climate variability
- ERA5
- Ethiopia
- Precipitation extremes
- Rainfall estimation
- Seasonal forecasting verification
- Validation
Data inizio appello
23/10/2024;
Disponibilità
parziale
Riassunto analitico
Ethiopia, a country where agriculture is the backbone of the economy, relies heavily on rainfall for its agricultural activities. The majority of its population depends on small-scale, rain-fed farming, which is highly vulnerable to fluctuations in weather patterns. Over the years, Ethiopia has experienced significant climatic variability, including recurrent droughts and rainfall patterns, leading to severe socio-economic consequences such as food shortages and loss of livelihoods. Given this context, accurate rainfall estimation and forecasting are critical for effective agricultural planning and disaster management. Improved rainfall prediction datasets and models can help taking the most effective decision to mitigate the impacts of weather extremes and enhance the resilience of agricultural practices, thereby supporting sustainable development and food security in the country. This is particularly relevant in the seasonal forecast range, since it can foster better-informed actions. This PhD thesis aims to enhance the understanding and prediction of rainfall patterns in Ethiopia by evaluating rainfall estimation products and seasonal forecasting methods to support better agricultural decision-making.
We start by looking at existing monitoring products, to understand how the climate is changing. We use observations and reanalyses products: the latter have the advantage of merging observations, where and when they are available, with short-range model forecasts. They provide a homogenous and consistent dataset covering the whole globe and many decades. But they have the disadvantage that they have a coarse resolution.
More specifically, we assess how good a state-of-the-art reanalysis, the fifth generation ECMWF atmospheric reanalysis of the global climate (ERA5), generated by the European Centre for Medium-Range Weather Forecasts (ECMWF), is for Ethiopia. The assessment is performed by comparing ERA5 against an existing gridded dataset of Climate Hazards Group InfraRed Precipitation With Station Data (CHIRPS) and against local observations. Once assessed its quality, we use ERA5 to investigate trends in a key variable extreme indices which are critical for Ethiopian farmers, specifically precipitation extremes. We then look at predicting precipitation, more specifically months ahead. We have chosen this very challenging forecast range because farmers have clearly express the need to access to forecasts issued months ahead, so that they can decide what crop and varieties to plant. The monthly forecasts that we used are generated by the ECMWF, which are accessible either directly from ECMWF or via the European Copernicus Climate Change Service (C3S). We also consider other probabilistic forecast covering the monthly time range, and access their accuracy and reliability over Ethiopia to support the agricultural decision-making.
Prior to this research, the accuracy and reliability of ERA5 reanalyses data and C3S model operational seasonal forecasts were not well understood for Ethiopia. This study addresses these gaps by providing new insights into the performance of these tools in monitoring ongoing climate change and predicting precipitation, particularly extreme events. In addition, this thesis we evaluate a the seasonal forecast based of the demand driven approach by accessing farmers needs for climate information service to identify existing gap and challenges to enhances the relevance and applicability of the forecasting models, ensuring they meet the real-world needs of the agricultural community.
This study addresses these gaps by providing new insights into the performance of these tools in monitoring ongoing climate change and predicting precipitation, particularly extreme events. ERA5, produced by the ECMWF, was assessed for its accuracy in representing rainfall across Ethiopia's diverse landscapes. The evaluation included comparisons with CHIRPS satellite-based rainfall estimates and local ground observations. The findings indicate that while ERA5 performs well at coarser resolutions, it faces challenges in accurately representing rainfall at finer resolution in complex terrains and high-altitude regions. In contrast, CHIRPS outperforms ERA5 in these areas due to its integration of local observations. Additionally, ERA5 has limitations in simulating extreme precipitation events accurately, particularly in regions with complex terrain, suggesting a need for enhanced calibration for localized settings. The study also utilized ERA5 to assess climate trends in precipitation, revealing significant variability and trends in precipitation extremes. We found out that ERA5 has a better capability in detecting the long term trend of precipitation extremes in most part the climate homogeneous regions of Ethiopia. This analysis is critical for understanding the impacts of climate change on Ethiopian agriculture and provides valuable insights into long-term climatic trends.
We conducted group discussions with farmers and carried out household surveys to understand their needs regarding agroclimate information services. Our findings indicate that farmers prioritize rainfall distribution and temperature forecast, followed by total rainfall amount and onset. It was clear that farmers require timely information to make decisions and allocate resources for farming. The majority of farmers prefer to receive information with a 1-month lead time, followed by 2 months in advance. This information is crucial for strategic agricultural decision-making, including crop selection, variety, and planning for agricultural inputs. Once we identify the need of the farmer, we evaluated the predictive capabilities of seasonal forecasting models, specifically the C3S models at different lead time. Prior to this research, forecast verification using the C3S models was not extensively studied, particularly in the context of Ethiopia's diverse climate. This thesis contributes by identifying which models work best for which seasons and at what lead times. The evaluation showed significant variations in forecast skill across different seasons, with models from ECMWF, UKMO, and CMCC exhibiting high skill during winter and late autumn. These reliable forecasts can meet farmers' needs, particularly in informing decisions about crop variety selection, land preparation, and planting schedules. However, they also suffer from the predictability barrier in spring, as the models like DWD, NCEP, and METEO FRANCE, which affect forecasts issued before spring aiming to predict weather for the forthcoming spring and summer months.
This thesis contribute to the understanding and application of rainfall estimation and forecasting products for agricultural decision-making in Ethiopia, which was not not extensively studied before. By assessing the performance of ERA5 and existing, top-quality seasonal forecasting models, it provides valuable insights into their strengths and limitations. Looking into the future, we suggest that improved model calibration, integration of localized observational data, and the use of advanced technologies like AI and machine learning could further enhance forecast accuracy and reduce their uncertainty, and improve the quality of reanalyses. Future research should focus on long-term validation studies, integration of localized data, and development of hybrid models that combine multiple data sources. These efforts will further support sustainable agricultural practices and resilience to climatic variability in Ethiopia.
We start by looking at existing monitoring products, to understand how the climate is changing. We use observations and reanalyses products: the latter have the advantage of merging observations, where and when they are available, with short-range model forecasts. They provide a homogenous and consistent dataset covering the whole globe and many decades. But they have the disadvantage that they have a coarse resolution.
More specifically, we assess how good a state-of-the-art reanalysis, the fifth generation ECMWF atmospheric reanalysis of the global climate (ERA5), generated by the European Centre for Medium-Range Weather Forecasts (ECMWF), is for Ethiopia. The assessment is performed by comparing ERA5 against an existing gridded dataset of Climate Hazards Group InfraRed Precipitation With Station Data (CHIRPS) and against local observations. Once assessed its quality, we use ERA5 to investigate trends in a key variable extreme indices which are critical for Ethiopian farmers, specifically precipitation extremes. We then look at predicting precipitation, more specifically months ahead. We have chosen this very challenging forecast range because farmers have clearly express the need to access to forecasts issued months ahead, so that they can decide what crop and varieties to plant. The monthly forecasts that we used are generated by the ECMWF, which are accessible either directly from ECMWF or via the European Copernicus Climate Change Service (C3S). We also consider other probabilistic forecast covering the monthly time range, and access their accuracy and reliability over Ethiopia to support the agricultural decision-making.
Prior to this research, the accuracy and reliability of ERA5 reanalyses data and C3S model operational seasonal forecasts were not well understood for Ethiopia. This study addresses these gaps by providing new insights into the performance of these tools in monitoring ongoing climate change and predicting precipitation, particularly extreme events. In addition, this thesis we evaluate a the seasonal forecast based of the demand driven approach by accessing farmers needs for climate information service to identify existing gap and challenges to enhances the relevance and applicability of the forecasting models, ensuring they meet the real-world needs of the agricultural community.
This study addresses these gaps by providing new insights into the performance of these tools in monitoring ongoing climate change and predicting precipitation, particularly extreme events. ERA5, produced by the ECMWF, was assessed for its accuracy in representing rainfall across Ethiopia's diverse landscapes. The evaluation included comparisons with CHIRPS satellite-based rainfall estimates and local ground observations. The findings indicate that while ERA5 performs well at coarser resolutions, it faces challenges in accurately representing rainfall at finer resolution in complex terrains and high-altitude regions. In contrast, CHIRPS outperforms ERA5 in these areas due to its integration of local observations. Additionally, ERA5 has limitations in simulating extreme precipitation events accurately, particularly in regions with complex terrain, suggesting a need for enhanced calibration for localized settings. The study also utilized ERA5 to assess climate trends in precipitation, revealing significant variability and trends in precipitation extremes. We found out that ERA5 has a better capability in detecting the long term trend of precipitation extremes in most part the climate homogeneous regions of Ethiopia. This analysis is critical for understanding the impacts of climate change on Ethiopian agriculture and provides valuable insights into long-term climatic trends.
We conducted group discussions with farmers and carried out household surveys to understand their needs regarding agroclimate information services. Our findings indicate that farmers prioritize rainfall distribution and temperature forecast, followed by total rainfall amount and onset. It was clear that farmers require timely information to make decisions and allocate resources for farming. The majority of farmers prefer to receive information with a 1-month lead time, followed by 2 months in advance. This information is crucial for strategic agricultural decision-making, including crop selection, variety, and planning for agricultural inputs. Once we identify the need of the farmer, we evaluated the predictive capabilities of seasonal forecasting models, specifically the C3S models at different lead time. Prior to this research, forecast verification using the C3S models was not extensively studied, particularly in the context of Ethiopia's diverse climate. This thesis contributes by identifying which models work best for which seasons and at what lead times. The evaluation showed significant variations in forecast skill across different seasons, with models from ECMWF, UKMO, and CMCC exhibiting high skill during winter and late autumn. These reliable forecasts can meet farmers' needs, particularly in informing decisions about crop variety selection, land preparation, and planting schedules. However, they also suffer from the predictability barrier in spring, as the models like DWD, NCEP, and METEO FRANCE, which affect forecasts issued before spring aiming to predict weather for the forthcoming spring and summer months.
This thesis contribute to the understanding and application of rainfall estimation and forecasting products for agricultural decision-making in Ethiopia, which was not not extensively studied before. By assessing the performance of ERA5 and existing, top-quality seasonal forecasting models, it provides valuable insights into their strengths and limitations. Looking into the future, we suggest that improved model calibration, integration of localized observational data, and the use of advanced technologies like AI and machine learning could further enhance forecast accuracy and reduce their uncertainty, and improve the quality of reanalyses. Future research should focus on long-term validation studies, integration of localized data, and development of hybrid models that combine multiple data sources. These efforts will further support sustainable agricultural practices and resilience to climatic variability in Ethiopia.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|