Drought Monitoring and Prediction With Higher Temporal Resolution Satellite Images
No Thumbnail Available
Date
2013-03
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Attributed to climatic change and uncertainty of weather conditions, drought has become a recurrent phenomenon in many parts of the world and particularly in eastern African region as most of its agriculture is dependent up on rainfall. In most areas of the region drought monitoring and prediction is done based on meteorological data which is very tedious and time consuming to gather and even the stations from where the data are collected are not evenly distributed and it is impossible to get data representing all the areas. And this issue brought us the need for better tools for monitoring and predicting drought in the area. Our study focuses on monitoring and prediction of agricultural drought. Studies has indicated integration of remotely sensed data such as NDVI images with some other relevant environmental attributes that influence vegetation condition increases the capability of predicting future vegetation conditions. The purpose of our study is constructing models, using higher temporal resolution data (dekadal data), by integrating Satellite vegetation, Climate, Oceanic and Biophysical data, for case of Ethiopia. The models are then used in monitoring and prediction of drought with higher accuracy. A 24 year historical data from the years 1983 to 2006, with spatial resolution of 8 km, for the selected attributes were used in constructing the model. The models were validated on test data and we got up to 99% prediction accuracy but when the predicted period is getting far the prediction accuracy decreased. The models make dekadal predictions and requires only for 10-days data to make the predictions this gives us the advantage of getting early update on drought situation. The models were implemented for the years 1984 and 2002 where severe drought occurred in Ethiopia. And we got similar results as at the time. Our work was compared with a recent study by Getachew et al. [19] where they built models, for Ethiopia case, using the same type of data as we used but with lesser temporal resolution (monthly average data). The comparison was done to check if using dekadal data brings a significant increase in prediction accuracy of drought over using monthly average data. The comparison confirmed that using data with higher temporal resolution, dekadal data, gives us higher prediction accuracy. So decision makers and other stake holders in the area are encouraged to try this model as it gives us greater prediction accuracy and early update on drought condition.
Key words: Drought monitoring, Drought prediction, Dekadal data, Monthly average data, Model
Description
Keywords
Drought Monitoring, Drought Prediction, Dekadal Data, Monthly Average Data, Model