SenseWeather: Sensor-Based Weather Monitoring System for Kenya

IST-Africa 2013 Conference Proceedings Paul Cunningham and Miriam Cunningham (Eds) IIMC International Information Management Corporation, 2013 ISBN: 978-1-905824-38-0 SenseWeather: Sensor-Based Weather Monitoring System for Kenya Muthoni MASINDE 1, Antoine BAGULA 2, Muthama NZIOKA 3 1 Department of Information Systems, Central University of Technology, South Africa Email: muthoni@uonbi.ac.ke, muthonimasinde@yahoo.com 2 Department of Computer Science, University of Cape Town, South Africa Email: bagula@cs.uct.ac.za 3 Department of Meteorology, University of Nairobi, Kenya, Email: jmuthama@uonbi.ac.ke Abstract: Like many other developing countries of Africa, Kenya s rain-fed agriculture is the backbone of the economy. As such, the Country s economy and weather are so intertwined that a good season means healthy economy and famine, and death otherwise. The status of weather forecasting in Kenya is rather wanting partly because the data used is collected from very sparse network of professional weather stations. Procuring and maintaining these stations is a costly affair for the Country still struggling to meet basic needs for her population. Developments in sensor technology have resulted in affordable and sustainable weatherboards that can operate as mini-weather stations. Though not as good as professional weather stations, these boards can be used to complement the networks of weather stations, hence improving the accuracy of weather forecasts produced by the Kenya Meteorological Department. We present SenseWeather, a weather monitoring system that we developed using Agriculture Boards from Libelium. In this paper, we describe the design, sensor calibration and deployment of SenseWeather. Keywords: Weather Monitoring, Wireless Sensor Networks, Kenya, sensor calibration, integrated weather forecasts. 1. Introduction The economic case of most African countries is agriculture. Agricultural production and weather are so highly interrelated that a good rainy season means a healthy economy, and failure of the rains means famine and death [An African economic planning Minister during Climate Variability, Water Resources and Agricultural Productivity: Food Security Issues in Tropical Sub-Saharan Africa, 1997], [13]. This is because given accurate, reliable and timely weather forecasts, farmers are able decide on the when, what and how to plant and hence improve production. The Kenya Meteorological Department (KMD) is charged with the sole responsibility of producing and disseminating weather forecasts to all the stakeholders in Kenya and the Eastern Africa Region at large. Like any other such Services in the developing world, KMD is faced with challenges in both the production as well as the dissemination processes of weather forecasts. One of the problems with the weather products production process emanates from the poor coverage by the weather stations. With only 27 operational synoptic stations [2] (a density of 21,495km 2 ), it is a daunting task for KMD to try and ensure high resolution (both spatial and temporal) of the forecasts. Wireless Sensor Networks (WSNs) technology is now maturing and weatherboards that are cheaper than the weather stations could be used to complement the networks of professional weather stations and hence addressing the challenges highlighted in the paragraph above ([1], [10], and [15]). WSNs application areas are diverse; they span several aspects of life such as environmental monitoring, precision agriculture, health and Copyright 2013 The authors www.ist-africa.org/conference2013 Page 1 of 13

medical care, biodiversity mapping and so on. WSNs application in weather monitoring is still relatively immature and most of the boards do not meet the minimum parameters requirement set by the World Meteorological Organization. These are listed in [11] as: (1) precipitation (type and amount); (2) surface air temperature; (3) atmospheric pressure; (4) Wind direction and speed; and (5) relative humidity. Most sensor boards reviewed were found to lack support for (1) and (4). Libelium s Agriculture Board supports all these and more[7]. It is for this reason that this board was selected as the implementation platform for the SenseWeather system presented in this paper. Having been the first (to the best of the authors knowledge) of its kind, the sensor boards were first calibrated in reference to the professional weather stations. Like most other battery-dependent sensor boards, power management was a major bottleneck. 2. Objectives Building upon the advantages provided by wireless sensor networks in terms of cost and the opportunity provided by the underlying technology for filling the visibility gap left by professional weather stations, we present in this paper SenseWeather, a weather monitoring system that we developed using Agriculture sensor boards from Libelium. As the main contribution of the paper and contribution to the deployment of weather monitoring practices in Kenya and the larger Eastern Africa Region, we present the design, sensor calibration and deployment of SenseWeather. This is a complementary solution to the KMD network as well as a tool to be used by small-scale farmers and insurers to mitigate the effects of droughts and other weather-related disasters. 3. Literature Review 3.1 Weather Monitoring for Development While people are hungry, all other development activities are thwarted. The hungry can concentrate little other than their next meal, [14]. The chronic poverty in rural (and some urban) areas of most countries in the developing counties such as Kenya is as a result of several factors. Some of these are; armed conflicts, diseases and droughts. It is the latter that is responsible for most causes of hunger especially in countries such as Ethiopia, Kenya, Lesotho, Swaziland, Zambia, and Zimbabwe as captured by the quote: Of the ten countries with the highest levels of hunger, and of the ten whose scores have actually increased since 1990, nine are in Sub-Saharan Africa in both cases" [17]. There is a link between weather monitoring and droughts; accurate weather monitoring can detect droughts occurrence long before they strike. Weather monitoring systems in place in Kenya and other African countries are fragile; no wonder close to 29% (over 296 million people) of Africa s population was affected by the droughts between 2001 and 2011 [4]. Though not solely responsible for the increase in poverty (population in absolute poverty was estimated to be 44.7% in 1992, 52% in 1997, and 56% by 2002 [5]) in Kenya, tackling droughts through improved weather monitoring is one way of arresting the situation. In Kenya, this is currently implemented using macroinfrastructures based on expensive and well-calibrated weather stations. The stations are then sparsely deployed and managed by KMD in form of relatively small number of fixed locations to provide climate maps for droughts and other natural disasters prediction. This creates a gap that can be addressed through WSNs technology. This technology provides support for low cost weather stations, and when deployed in the environment, they are able to improve coverage. WSNs are relatively cheaper (than conventional weather stations) and they can accurately measure meteorological parameters such as temperature, wind speed/direction, soil moisture, precipitation, atmospheric pressure and relative humidity Copyright 2013 The authors www.ist-africa.org/conference2013 Page 2 of 13

3.2 Weather Monitoring in Kenya Current Practice Currently, monitoring of climatic/weather variations in Kenya is the mandate of the Kenya Meteorological Department. The Department runs 3 main types of stations that are currently managed by the Climatological Section of the Department (http://www.meteo.go.ke/): 700 rainfall stations, 62 temperature stations and 27 synoptic stations. The latter have the capability to observe and record all the surface meteorological data; rainfall, temperature, wind speed and direction, relative humidity, solar radiation, clouds, atmospheric pressure, sun shine hours, evaporation and visibility. These are therefore very critical in monitoring weather. The 27 stations [2] translate to a density of 21,495km 2 and although this is better than many countries in the SSA (South African 27,751km 2, Tanzania 42,959km 2 and Zambia 50,174km 2 ), it is far from being as good as the coverage in the developed countries (e.g. UK 1,168km 2, Germany 3,607km 2 and USA 6,664km 2 ). The Agrometeorological Section on the other hand manages 13 stations related to agriculture; data is remitted from these stations every 10 days. Apart from the normal meteorological observations, other observations by the Agrometeorological Section include: soil temperature, sunshine duration, radiation, pan evaporation and Potential Evapotranspiration. All this data is stored in semi-automated formats at the Department s Head Quarters in Nairobi. The data is available to interested stakeholders on request. As it is the practice all over the world, KMD uses the data collected to provide five main types of forecasts: Daily for main cities/towns in Kenya, Four-Day, Seven-Day, Monthly and Seasonal. The Four-Day, Seven-Day and Monthly forecasts are in form of downloadable pdf reports summarizing the recent past, current and near-future weather patterns in conceptual terms. On the other hand, seasonal forecasts are more detailed and they are also translated into Kiswahili language. The Department also works hand in hand with both the print and electronic media to disseminate the forecasts. All the above forecasts are freely accessible from the Department s website (http://www.meteo.go.ke). It is not all glossy at KMD; the Department is struggling with a number of challenges during the production, dissemination and application processes of the weather products. First, the sparse network of weather stations poses a huge problem; the data coverage is too coarse to have meaning at local levels. Two, the dissemination channels are not effective and they are not appealing to most stakeholders especially the farmers; they find the information too general and technical. The farmers have therefore continued to rely on their indigenous forecasts that they find more compatible to their local cultures and their format is more specific to their contexts ([12] and [9]). Sensor-based weather stations can be used to beef-up the sparse network of weather stations while the now readily available mobile phones can be utilised in disseminating the forecasts at the local level. Kenya brags of very high penetration levels of mobile phone; during the latest (2009) population sensors [6], it emerged that only 3.6% of households owned at least one computer in comparison with 63.2% of households that owned at least one mobile phone. There are other complexities involved in the whole weather forecasting journey but only the use of WSNs is tackled here; other issues are addressed in ([8], [10] and [16]). 4. Sensor Calibration 4.1 Calibration Model The acceptability of the data provided by the sensors by the meteorologists is only possible if the sensors are calibrated against conventional weather stations. In order to take care of this, calibration of the sensors was carried out using conventional weather stations at KMD; Chiromo Observatory (run by the Department of Meteorology, University of Nairobi) was Copyright 2013 The authors www.ist-africa.org/conference2013 Page 3 of 13

used for the initial experiments. After the calibration exercise, the sensor boards were used to design SenseWeather. Calibration experiments were conducted to evaluate the field readiness of the Libelium s Waspmote[7] platform to support weather forecasting. The calibration exercise was carried out for a period of one year between July 2011 and July 2012. It entailed taking readings from both the sensors and the weather stations in parallel on 24-7 basis. The calibration model followed a 3-steps experimental process with three types of experiments: (1) pilot experiments; (2) explanatory experiments; and (3) confirmatory experiments. A systematic error analysis based on three error types: Mean Absolute percentage Error (MAPE), Mean Error (ME) and Root Square Error (RSE) then followed. We also checked for inherent errors that are inbuilt in the sensor boards. We finally used correlation coefficients and plots such as side-by-side boxplots to run similarity tests between various datasets. With impressive accuracies ranging from 92 to 99 %, this gave us the confidence to move to the next step; system design and implementation. The sensors were programmed to take readings every 30 minutes while the readings from the weather station were taken on hourly basis. That is, at hour t (say 1 GMT), the sensor boards recorded two readings for each of the sensors. For example, with 6 sensor boards, each fitted with 3 sensors (temperature, humidity and pressure), this would result in 12 readings for temperature, 12 readings for humidity and 12 readings for pressure. In order to aggregate these readings for the purpose of comparing them with the respective readings from the weather station, two options were pursued: (A) Option 1: Average All Sensor Readings Taken Within the Hour: Where S s is the aggregated reading for Sensor S; for example, S could be temperature sensor or humidity sensor. S i1 and S i2 are the sensor reading for Sensor S on Sensor Board i. For instance, in the case of five sensor boards, the aggregated reading for temperature sensors would be computed as follows: (B) Option 2: Average Readings Taken Closest to the Hour In this case, only one out of the two sensor reading from each sensor board is considered; the one closest to the weather station readings observation time. That is: In the case, given five sensor boards, the aggregated reading for the temperature sensor would be: 4.2 Calibration Results Before reaching the final decision as to which of the two options (of combining sensor data), further analysis were carried out as explained below. Copyright 2013 The authors www.ist-africa.org/conference2013 Page 4 of 13

4.2.1 Error Analysis Summary Table 1 Confirmatory Experiment - Error Analysis Illustration Error Type Option Temperature Humidity Pressure MAPE RMSE Option 1 8.55 12.54 1.47 Option 2 8.53 11.90 1.47 Option 1 1.96 10.94 12.06 Option 2 1.89 10.56 12.10 For temperature and humidity sensors, Option 2 performed better for both MAPE and RMSE. However, though Options 1 and 2 had equal performance (1.47) for pressure sensor using MAPE, Option 1 outperformed Option 2 under RMSE (12.06 versus 12.10). Based on some discrepancies noted for the pressure readings during the experiments (the details are discussed in the Further Work Section), the discrepancy above was ignored and a decision to pick Option 2 as the best way of combining the sensor readings reached. 4.2.2 Correlation Coefficients To further validate the choice of Option 2, the correlation coefficients of the sensor readings with the weather station were computed. Table 2 Confirmatory Experiments - Correlation Coefficients Options Temperature Humidity Pressure Option 1 0.924 0.920 0.723 Option 2 0.940 0.936 0.657 Again, except for the pressure sensor, Option 2 had the highest correlation coefficients. Adjustments were then made at the program-code level to factor in the error rates. For example, AdjustedTempt = OriginalTempt+(OriginalTempt * MAPE); That is, the adjusted temperature reading for each Sensor Board taken at time t is computed by adding a weight factor equivalent to the respective Mean Absolute Percentage Error. In the case of Temperature sensors, this translates to: 8.53%. Graphs (similar to the one below) and boxplots were produced to measure the similarities of the readings Fig. 1: Humidity Sensor Comparison with Station Copyright 2013 The authors www.ist-africa.org/conference2013 Page 5 of 13

5. SenseWeather Design and Implementation 5.1 System Overview SenseWeather is part of a comprehensive system made up of several sub-systems that are linked up together by intelligent agents that are implemented using the Java-based multiagent systems development framework called JADE (Java Agent Development). The subsystems are: (1) Sensor-Based Weather Monitoring System prototype; (2) EDI Monitor which is a FORTRAN program; (3) ANNs Forecasting Tool; (4) IK Fuzzy Sub-System that stores Indigenous Knowledge (IK) drought indicators;(5) Android Mobile Application to input and output IK indicators as well extreme weather events; (6) SMS Gateway that allows members of the public to interact with the entire system and also used to receive weather readings from sensors into the system; and (7) a user-friendly web portal used for both system administration as well for displaying detailed information on droughts and other related details[16]. The integrated system was designed to meet the need for affordable, relevant, sustainable and user-friendly drought early warning system (DEWS) for Sub-Saharan Africa. The DEWS is composed of three elements: (1) Drought Knowledge (2) Drought Monitoring and Prediction; and (3) Drought Communication and Dissemination. This DEWS is currently implemented under a framework called ITIKI; acronym for Information Technology and Indigenous Knowledge with Intelligence is a bridge that integrates indigenous drought forecasting approach into the scientific drought forecasting approach. ITIKI was conceptualised from itiki which is the name used among the Mbeere people (found in the Eastern part of Kenya), to refer to an indigenous bridge made using sticks and was used for decades to go across rivers[16]. Some of the relevant (to this paper) subsystems of the integrated system logic are captured in the figure below: Figure 2: SenseWeather: System Logic Copyright 2013 The authors www.ist-africa.org/conference2013 Page 6 of 13

5.2 SenseWeather Design In order to design, implement and test the SenseWeather, the following Sensor nodes/boards and accessories were identified and procured from Libelium Comunicaciones Distribuidas S.L. (http://www.libelium.com/): Fig. 3: Actual Size of Agriculture Sensor Board After the rigorous sensor calibration exercise, the calibrated sensor boards were used to design a weather monitoring system prototype. Two types of deployments were set up: (A) Sensor Boards Next to Weather Stations Here, sensor boards were placed within the Observatory Units of selected weather stations in Kenya. The boards individually send readings to a remote database via an SMS Gateway. The sensors included are those for measuring temperature, relative humidity and atmospheric pressure. In a few of the locations, rainfall, wind speed, wind direction, and soil moisture sensors were installed. Aggregation of multiple sensor readings was performed using Option 2 described under the Calibration Section. Apart from monitoring weather, this set up sought to further validate the calibration decisions. (B) Stand-Alone Sensor Boards In order to deploy the sensors in the rural areas especially in Mbeere and Bunyore on which SenseWeather deployments are targeted, stand-alone sensors mounted with temperature, relative humidity and atmospheric sensors were used. The sensors boards were placed inside traditional granaries which provided environment almost similar to the one supported by the Stevenson Screens. Fig. 4: A Traditional Granary in one of the Mbeeres Testing Location Copyright 2013 The authors www.ist-africa.org/conference2013 Page 7 of 13

Fig. 5: Wireless Sensor Boards Program code that put into consideration the calibration weights reached at during the calibration exercise was loaded on to each of the sensor boards; readings were then taken every 30 minutes. To minimise the cost of sending SMS, the sensors send the readings to the database (using the GSM/GPRS module) via the SMS gateway on hourly basis. For backup purposes, each board also saves all (every 30 minutes) the readings in a Secure Digital (SD) card. The activity of the sensors is monitored from a web interface; a sample is shown in Fig. 6. 5.3 Sensors Monitoring Interface All data from the sensors is accessible from the web portal from where the system administrator is able to monitor the sensors behaviour. A sensor board that does not send readings for a period of one hour is deemed to have failed due to either (or all) of the following reasons: Depleted airtime: The SIM cards installed on the sensor boards were always topped up with adequate airtime for sending the weather readings in form of text messages. However, in isolated cases, the airtime got depleted and therefore such a sensor board would not send the readings. Simply topping up the SIM card remotely (using services such Sambaza by Safaricom) would solve this. Low Battery Power: Based on the calibration results, a sensor board s battery would be replaced (with a fully charged one) as soon as the BatteryLevel value was below 40%. However, there are situations where some boards failed long before the battery power level went below 50%. Failure of the GPRS module: the GPRS installed on the sensor boards caused most of the failures; it would fail to connect to the network, causing the program code on the sensor to halt. This was the most difficult problem to solve because it required re-starting (reprogramming sometimes) the sensor boards and hence interfering with the synchronisation of the readings. This also meant that only a technical person (with the knowledge of programming the sensor boards) would solve this problem. For instance, any time the stand-alone sensors installed in Mbeere and Bunyore failed, they had to be physically delivered to the administrator based in Nairobi to reset them; this would take at least a day. Copyright 2013 The authors www.ist-africa.org/conference2013 Page 8 of 13

5.4 Sample Output from the Sensors Figure 6: Sensor Boards Monitoring Interface The example below is of readings from temperature sensors installed on sensor boards installed in Mbeere. The readings were taken between 15 July and 16 July 2012 Fig. 7: Sample Temperature Readings from Sensor Boards Located in Mbeere Below are relative humidity readings from sensors located at KMD taken between 21 July and 25 July 2012. Copyright 2013 The authors www.ist-africa.org/conference2013 Page 9 of 13

Fig. 8: Sample Humidity Readings from Sensor Boards Located at KMD The readings below show an example where one sensor stopped responding due to GPRS failure. 5.5 SMS Gateway Fig. 9: Sample Sensor Readings Showing a GPRS failure This is a simple Java-based SMS Gateway made up of one main class (SmsSender.java); it is supported by two other classes; (ComputerSmsData.java and SerialToGsm.java) and one xml file (configs.xml). To have it run continuously, smssender.java implements threads. It receives messages from the sensors, decodes them into the various components and then uploads the data to the database, which is located in a remote site. The SMS Gateway runs on a computer that is permanently connected to the Internet to allow for routing of the readings. It is currently installed in the Server Room in the School of Computing and Informatics, University of Nairobi, Kenya. 5.6 Estimated Cost of ITIKI Implementation 5.6.1 Overview The salient features of SenseWeather require that its implementation is people-driven (especially the small-scale farmers). As such, anchoring this within a local (to the implementation region/area) community based organisation, church, non-governmental organisation and so on is recommended. The Mbeere implementation pilot mentioned in Copyright 2013 The authors www.ist-africa.org/conference2013 Page 10 of 13

this paper was anchored within a Community Based Organisation. The estimated implementation cost below is therefore based on this fact. 5.6.2 Equipment Cost The major cost is associated with the sensor boards; we used Agriculture and Agriculture PRO sensor boards from Libellium. In 2011, it would cost about 2,500 Euros to acquire a complete weather meters for measuring all the necessary weather parameters (including soil moisture) for an area of about 200km 2. At the time of writing this paper, there were strong indicators that the cost of such meters was going down and the number of suppliers was going up. Below is a sample quotation issues in May 2011: Given the critical role of precipitation in SenseWeather, we also recommend that the density of the weather meters be increased using about 10 conventional (manual) rain gauges for the area covered by the sensors boards. Other equipment costs were associated with android phone; about 5 phones (at 80 Euros each). Finally, a computer server and an Internet modem (to receive sensor readings and drought forecasts input/output) cost us about 500 Euros. Copyright 2013 The authors www.ist-africa.org/conference2013 Page 11 of 13

6. Conclusion and Future Work In this paper, we have described the design and development of SenseWeather; an application that uses sensor-based weather station to complement the sparse network of professional weather stations in Kenya. The system is designed to integrate weather readings from the stations with parameters (temperature, pressure, relative humidity, precipitation, soil moisture, wind speed and wind direction) recorded by the sensors. Given that the use of WSNs in weather monitoring is relatively new, the design of SenseWeather was preceded by a calibration exercise that was meant to compute the accuracy of the sensor boards. We acknowledge that monitoring weather alone cannot eradicate droughts in SSA; access to relevant and accurate information on impending droughts in timely fashion and comprehensible formats however could go a long way in assisting all the stakeholders plan for and mitigate effects of the droughts. This precisely is the contribution of SenseWeather. The presence of sensors in SenseWeather enables capture of micro weather data and hence, improved prediction accuracy. Despite challenges such as battery power management and instability of some of the sensor boards functionalities (especially the GPRS module), the system prototype currently undergoing user testing has great potential as a complementary tool for weather and eventually drought monitoring in Kenya and the rest of the countries the Sub-Saharan Africa. SenseWeather is a prototype with lots of extensions in-waiting; more sensor boards need to be acquired and installed. This will improve the current poor coverage by conventional weather stations. The coverage could also be improved by installing mobile weather sensors, which could be mounted on mobile objects such as long distance buses that travel to the cities from remote villages. The latter requires careful design work involving both computer scientists and meteorologists. Further, before the output from SenseWeather is made public, the system prototype needs to be anchored into a formal institution. Such an institution will be responsible for putting necessary measures in place before/after issuing weather alerts. It will also take liability for any consequences arising from the forecasts/alerts issued. Finally, apart from securing all the resources needed to roll out SenseWeather, this institution should acquire the necessary registrations that are needed before recognition (such as acquiring International Civil Aviation Organisation identities) of new weather stations by World Meteorological Organisation. This is especially so for the stand-alone sensor-based stations that must first be registered before their readings are used. The pressure sensor we used seemed too sensitive compared to the one used at the weather station (Kew-type station barometer). This made it very difficult to compute a very suitable correction factor for this sensor. Experiments to use the barometer that comes with the Automatic Weather Station (AWS) should be performed and used to resolve this. References [1] BAGULA, A., ZENNARO M., INGGS G., SCOTT S. AND GASCON D., 2012. Ubiquitous Sensor Networking for Development (USN4D): An Application to Pollution Monitoring. Sensors 12, 391-414.. [2] EAC, SECRETARIAT. 2008. Enhancing Capacities of the Meteorological Services in Support of Sustainable Development in the East African Community Region Focusing on Data Processing and Forecasting Systems.. [3] EIKO, Y. AND BACON, J. 2006. A survey of Wireless Sensor Network technologies: research trends and middleware's role.. [4] GEOFFREY, T., SUE, A. AND FREDERIC, M. 2011. World Disasters Report Focus on Hunger and Malnutrition. International Federation of Red Cross and Red Crescent Societies. [5] GOVERNMENT OF KENYA. 2005. Millennium Development Goals In Kenya: Needs & Costs. [6] KENYA NATIONAL BUREAU OF STATISTICS. 2009. The Kenya Census 2009: Population and Housing Census Highlights.. [7] LIBELIUM, COMUNICACIONES, DISTRIBUIDAS, S.,L. 2010. Agriculture Board Technical Guide.. Copyright 2013 The authors www.ist-africa.org/conference2013 Page 12 of 13

[8] MASINDE, M. AND BAGULA, A. 2011. A Framework for Integrating Indigenous Knowledge With Wireless Sensors in Predicting Droughts in Africa. In Indigenous Knowledge Technology Conference 2011, 2-4 November, N. BIDWELL AND W. HEIKE, Eds.. [9] MASINDE, M. AND BAGULA, A. 2012. ITIKI: bridge between African indigenous knowledge and modern science of drought prediction. Knowledge Management for Development Journal In Press, 1-19.. [10] MASINDE, M., BAGULA, A. AND MUTHAMA, N. 2012. The Role of ICTs in Downscaling and Upscaling Integrated Weather Forecasts for Farmers in Sub-Saharan Africa. In The Fifth International Conference on Information and Communication Technologies and Development, March 12-15, M. BEST L., Z. ELLEN, D. JONATHAN, G. BEKI AND M. GARY, Eds. ACM Digital Library, Atlanta, GA, USA, 122. [11] PLUMMER, N., TERRY, A. AND JOSÉ, A., LOPEZ. 2010. Guidelines on Climate Observation Networks and Systems.. [12] UNITED, NATIONS, DEVELOPMENT, PROGRAMME. 2000. Coping with Drought in Sub-Saharan Africa: Better Use of Climate Information.. [13] VIRJI, H., CORY, F., AMY, F. AND MAYURI, S. 1997. Climate Variability, Water Resources and Agricultural Productivity: Food Security Issues in Tropical Sub-Saharan Africa.. [14] WORLD FOOD, PROGRAMME. 2006. Millennium Development Goal 1: Eradicate Extreme Hunger and Poverty. 2012,. [15] ZENNARO, M., BAGULA, A. AND BJORN, P. 2008. Wireless Sensor Networks: a great opportunity for researchers in Developing Countries. In International Symposium on Wireless Communications and Information Technology in Developing Countries, 6-7 October, P. KRITZINGER AND P.S. ANA, Eds. www.uct.ac.za, Pretoria, South Africa. [16] MASINDE, M., 2012, ITIKI: Bridge between African Indigenous Knowledge and Modern Science on Drought Prediction', PhD thesis, University of Cape Town [17] IFPRI, 2011-last update, 2011 GHI - Facts and Findings: Sub-Saharan Africa [Homepage of International Food Policy Research Institute], [Online]. Available: [May 20, 2012]. http://www.ifpri.org/publication/2011-ghi-facts-and-findings-sub-saharan-africa Copyright 2013 The authors www.ist-africa.org/conference2013 Page 13 of 13