Deep Learning for Vegetation Health Forecasting: A Case Study in Kenya

Thomas Lees,Gabriel Tseng,Steven Reece,Simon Dadson,Clement Atzberger

doi:10.3390/rs14030698

Abstract

East Africa has experienced a number of devastating droughts in recent decades, including the 2010/2011 drought. The National Drought Management Authority in Kenya relies on real-time information from MODIS satellites to monitor and respond to emerging drought conditions in the arid and semi-arid lands of Kenya. Providing accurate and timely information on vegetation conditions and health—and its probable near-term future evolution—is essential for minimising the risk of drought conditions evolving into disasters as the country’s herders directly rely on the conditions of grasslands. Methods from the field of machine learning are increasingly being used in hydrology, meteorology, and climatology. One particular method that has shown promise for rainfall-runoff modelling is the Long Short Term Memory (LSTM) network. In this study, we seek to test two LSTM architectures for vegetation health forecasting. We find that these models provide sufficiently accurate forecasts to be useful for drought monitoring and forecasting purposes, showing competitive performances with lower resolution ensemble methods and improved performances over a shallow neural network and a persistence baseline.

Highlights

Academic Editor: Aleixandre VergerDrought is estimated to be one of the world’s most costly hazards, accounting for22% of damage from natural disasters [1]
We have explored the efficacy of two Long Short Term Memory (LSTM)-based architectures for vegetation health forecasting
The LSTMs are competitive with the published results for the ensemble of models developed by the National Drought Management Authority (NDMA) [31], while having significantly greater spatial resolution

Summary

Introduction

22% of damage from natural disasters [1]. Droughts impact social and natural environments around the world [2,3]. Horn of Africa Drought [5], the 2013–2014 California drought [6], the 2015–2017 Southern. African Drought [7,8], the 2005 Amazon Drought [9], and the 2003 European Drought [10]). Vegetation health is a key drought indicator, and forms an important component of many drought early warning systems (EWS), as reviewed in Rembold et al [12]. The European ASAP system provides timely information about possible crop production anomalies based on a time series of satellite-based biophysical indicators for food insecure areas [12]

Methods

Results

Conclusion