Research
Model of different scales:
Understanding multi-scale resilience options for climate vulnerable Africa
We are developing granular-level agent based model (ABM), zonal-level multi-player microeconomic partial equilibrium model (MME), and national-level computable general equilibrium model (CGE).
The ultimate objective is to understand coupled system dynamics across scales in a manner that allows us to quantify the sensitivity of critical human outcomes (nutritional satisfaction, household economic well-being) to development and climate adaptation strategies in sub-Saharan Africa.
Teff (a crop):
Ethiopia's Policy on the Export Ban of Teff - a staple grain in the country that has increasing global demand
In response to global food price volatility and trends toward increased global food demand, Ethiopian policy makers were forced to adopt strategies such as restricting food exports in order to protect domestic food security. However, these policies can have a disproportionate regional impact on domestic markets and can result in lost revenue from exports. We assess the impact of the ban and of proposed policies to relax the ban, across regions within the country and for various market actors along the teff value chain. Using a partial-equilibrium model developed with a detailed modeling of the agro-economic features of the country, we analyze the direct impacts on export revenue, producers' profits, transport patterns, and consumption across the disaggregated regions in Ethiopia due to changes to its teff export policy. [publication]
Small-scale irrigation at the intersection of Food, Energy, and Water (FEW) system in Ethiopia
I am developing groundwater and surface water hydrologic model in Ethiopia, coupling with a crop yield model, which is integrated to a Food-Energy-Water system model (i.e. multi-player microeconomic partial equilibrium model or MME). We are investigating the influences of irrigation on food security, energy demand, and economic well-being under various irrigation scenarios.
The preliminary results show improved food consumption under with irrigation overall, although there are some uneven distribution of added welfare. The animation shows the changes in crop transport pattern between zones under irrigation. With irrigation, some zone (one on the leftmost) becomes a net importer (food deficit zone) from a net exporter (food surplus zone), meaning consumers in this zone are facing a relatively higher crop price compared to other zones, although the consumption in this zone is actually improved with irrigation.
Crop transport pattern change:
Climate impact on crop yield over years
(0 - crop failure, 1 - no impact):
Climate Yield Factors (CYF) development and projections under climate change scenarios
The climate yield factors (CYFs) represents the impact of water stress on crop yield under different climate conditions. It ranges from 0 to 1, with 1 indicating no water stress and 0 indicating crop failure. The CYFs were computed based on a soil-water balance model, which takes consideration of climate conditions, soil properties, and crop characteristics. The model keeps track of daily soil moisture change, crop root growth, evapotranspiration, groundwater recharge, runoff and baseflow to surface water, and so on.
The animation shows the 25km x 25 km spatially explicit CYFs for maize in each growing season across 1950-2095 in Ethiopia, based on the climate inputs from one ensemble of GFDL CM3 under the RCP 8.5 scenario. Other the Global Circulation Models (GCMs) were also explored.
Incorporating operational climate forecasts into dengue outbreak prediction
We proposed a model that incorporates the state-of-the-art real-time operational climate forecasts from the North American Multi Model Ensemble (NMME) into an outbreak prediction framework, and apply it to the dengue-prone region of Negombo, Sri Lanka. We adapted logistic regression to develop skillful categorical dengue predictions, particularly with a longer forecast window.
Early warning system for dengue outbreak:
Dengue predictions with longer forecast window thanks to climate predictions.
Modeling the role of human mobility on dengue outbreak evolution in Sri Lanka
We modeled the spatial-temporal dynamics of a large dengue outbreak in the Negombo region in Sri Lanka as a function of human mobility patterns, land use and climate data. The analysis was conducted at a 1 km × 1 km spatial scale and a weekly temporal scale. Our results indicate that mobility is a strong indicator for local outbreak clusters. In addition to showing the relationship between mobility and outbreaks, the results of our model can be used to predict where new cases of dengue are likely to occur, and thus be helpful in effectively applying disease prevention and vector control measures.
The animation shows the weekly evolution of the number of dengue cases on a scale from 0 to 8 in each cell of approximately 1km x 1km. The figure title indicates the starting date of that week. We can see that the cases were scattered in the region at first, and gradually increased and clustered mostly in the populated coastal area (ocean marked in blue) around the lake (marked in grey). [preprint]
Dengue cases evolving:
K-means clustering algorithm:
Regionalization of seasonal precipitations using cluster analysis
We applied hierarchical and nonhierarchical (k-means) clustering techniques on a gridded dataset for objective and automatic delineation of homogeneous climate regions regarding seasonal precipitations. We also developed two new objective selection metrics for identifying an optimal number of clusters.
The figure shows a synthetic k-means clustering conceptualization based on a two-dimensional dataset (X, Y). The diamonds and the crosses represent the data points and the centroids, respectively. (a) Scatterplot of the data. Data points assigned to the closest centroid (labeled with the same index as its assigned centroid): (b) randomly assigned centroids to initiate the algorithm, (c) recalculated centroids and reassigned data points, and (d) recalculated centroids and the same reassigned results (convergence). [publication]
Seasonal precipitation predictions in western Ethiopia
We proposed advancing local-level seasonal precipitation predictions by first conditioning on regional-level predictions, as defined through objective cluster analysis, for western Ethiopia.
Sea surface temperatures (SSTs) in the equatorial Pacific Ocean representing the well-known El Niño–Southern Oscillation (ENSO) phenomenon are considered a primary indicator of Jun-Sep seasonal total (JJAS) precipitation variability, with El Niño/La Niña often associated with deficit/excess of JJAS precipitation across the study region.
The figure shows the justifiable regions for selecting SSTs as predictors, including equatorial Pacific (EP), North Indian Ocean (NI), South Indian Ocean (SI), and equatorial/South Atlantic Ocean (AT). We developed a program that automates the predictor selection processes. [publication]
Predictor regions identified:
Framework of incorporating climate prediction in agricultural decision-making:
Integrating Climate Prediction and Regionalization into an Agro-economic Model to Guide Agricultural Planning
We connected climate prediction to agricultural planning using an agro-economic model - an interdisciplinary approach to access the ex-ante value of the prediction in terms of various economic indices. Hence, the climate predictive information was tailored and can be communicated in a manner relevant to policy makers.
The figure shows the concept of our methodology. When climate prediction is given to farmers, they can decide to take alternative actions from business as usual. Based on the action they chose, the model can simulate the associated net benefits such as GDP, calorie consumption, and poverty rate given the realized climate condition. [publication]
Predicting seasonal Blue Nile river flow of summer 2018, 2019, 2020, ...
Ethiopia will soon begin filling the reservoir of the largest hydropower dam in Africa, the Grand Ethiopian Renaissance Dam, located on the Blue Nile River. It has brought concerns from the downstream riparian countries, particularly regarding the amount of water that will be reduced due to the filling. In this context, we ask a simple forecast question: what is the status of seasonal forecasts of Blue Nile River flow? If a wetter year is expected during the main rainfall and streamflow season for the filling, there will be fewer acute impacts on the downstream countries.
Eight statistical models (four previously published and four created by us) and eight dynamical models from the North American Multi-Model Ensemble (NMME) were used to forecast the 2018 high rainfall season (June-September) and high river flow season (July-October). As a result, average to above average Blue Nile River flow was expected in 2018 as shown in the figure. Details of this work can be found on the blog with a follow-up update. We continued to predict in 2019 (blog) and will continue to issue predictions in coming years. [publication]
2020 Blue Nile peak season forecasts (full blog):
Projection using wavelet signal:
Projection of hydroclimatic time-series with identified decadal signals
We evaluated interannual and decadal-scale streamflow variability into the forthcoming Grand Ethiopian Renaissance Dam reservoir. Simulations of probabilistic streamflow via wavelet analysis are produced to define the propensity towards wetter or drier conditions during the filling stage of the reservoir.
The figure shows the historical and projected monthly precipitations, highlighting the embedded low-frequency signal which is aggregated from two decadal-scale signals detected via wavelet analysis. [publication 1] [publication 2]
Evaluating impoundment strategies for the Grand Ethiopian Renaissance Dam with implications for downstream riparian countries
The dam, located just upstream of the border with Sudan, is the first dam ever to be constructed directly on the main stem of the Blue Nile and will become the largest dam in Africa. Filling the reservoir will have clear implications on downstream flows in Sudan and Egypt. We explored various filling policies and future climate states simultaneously through a linked set of models to infer potential streamflow reductions near Sudan’s Gezira Scheme, one of the largest irrigated areas in the world, and at Lake Nasser behind Egypt’s High Aswan Dam. [publication]
The Nile Basin and the Grand Ethiopian Renaissance Dam:
The map shows the Nile basin with the forthcoming dam (GERD). The box plots show the percent change in annual average streamflow during 2017-2032 at Gezira Scheme (upper right) and Lake Nasser (lower right) for each filling policy across climate projections. All values are relative to the conditions under no water retainment and no climate change.