You are on page 1of 26

Agricultural Drought Severity Assessment Using Land Surface Temperature and NDVI In Nakuru region, Kenya.

*Kipterer John Kapoi,1 Omowumi Alabi,2


1

Regional Centre for mapping of resources for development (RCMRD); kkapoi@yahoo.com*


2

Kenya:E-Mails:

African Regional Centre for Space Science and Technology Education in English, Obafemi Awolowo university Campus, Ile Ife, Osun State ; Nigeria: E-Mail: alabi@arcsstee.org

Abstract
This study was focused on Nakuru, a tropical region in the Rift Valley of Kenya, bounded between latitude 0.28N and 1.16S, and longitude 36.27 E and 36.55E. The main aim of this research is to assess the agricultural drought in the high potential region of Kenya with an objective of mapping the agricultural drought severity levels, assessing the precipitation and normalized difference vegetation index deviation over its long term mean average in the region and to generate land surface temperature and emissivity maps to compare the surface temperature proportion during the drought and normal period. The data were obtained from NOAA-AVHRR, LANDSAT TM and ETM+ and was processed with ERDAS Imagine and GIS software of the Environmental Systems Research Institute (ESRI). The land surface temperature was derived using Plancks radioactive principles. The thermal band of Landsat TM was utilized to extract the radiance and brightness temperature. The brightness temperature was combined with surface emissivity to derive the land surface temperature (LST) while NDVI was derived from bands 3 and 4 and its result was divided by the LST to determine the moisture levels. The products were classified into five main classes to reflect the moisture levels. Rainfall and NDVI performance were also processed from NOAA AVHRR and long term mean established and compared with the specific year of student performance. The result of the study revealed that NOAA-AVHRR data offers very useful information in drought monitoring and early warning, LST and NDVI are useful in moisture level mapping that can be used to detect drought and the drought in Nakuru is characterized by both low and high temperatures that exacerbates the crop failure. Key words: Drought, Land Surface temperature (LST), land use classes, emissivity, Vegetation

Introduction The frequent drought experienced in the greater horn of Africa has negatively impacted on the natural habitats, livelihoods and health, and the food production levels in the region. According to the Food and Agricultural Organization (FAO, 2011), the decline in food production associated with the drought conditions in this part of the world is one of the major factors responsible for the increased cost of food and malnutrition experienced by the inhabitants. Other affiliated consequences of the drought situation which include land degradation and soil erosion are most pronounced in the agriculturally potential areas. An example of an agricultural region in the greater horn of Africa that has been subjected to the devastating impact of drought is Nakuru County in Kenya. According to Mubea et al. (2009), Lake Nakuru, the major tourist attraction in this agro-ecological zone has been decreasing in size, with a depth reduction from 2.6m to 1.4m (Morgan, 2009). Furthermore, some of the perennial rivers in the region have become seasonal (Roncoli et al., 2010 and Morgan, 2009). The declined water conditions experienced in the region have also resulted in mass death and migration of birds, notably flamingoes and marabou stocks (Thome, 2009 and Mwenya, 2009). Furthermore, this fertile zone, once accredited with good crop yield, has over time experienced a decline in wheat production (Hezron et al., 2007). !" "

According to Roncoli et al. (2010), the changing weather has resulted in adaptation strategies from wheat production to maize, sweet potatoes, and the cultivation of vegetables. Some severe examples of weather changes associated with the drought situations in Nakuru include adverse changes in temperatures, with extreme prolonged cold weather destroying crops and reducing livestock production. The harsh, foggy, windy and frost moments have also had a negative impact on pollination (Roncoli et al., 2010). These changes which have resulted in decrease in household food production levels have also generated secondary social problems such as conflicts between the crop farmers and livestock keepers. Furthermore, the continual diminution of rainfall in the area resulted in declining forests as a result of pressure from population to meet their energy demands. The region has also experienced reduced investment in farming as farmers have reduced their cultivation to sizeable land in order to minimize massive losses in case of crop failure as a result of drought (Mubea et al., 2009 and Walubengo, 2007). In Kenya, little attention is given to drought assessment in agricultural potential areas such as Nakuru. The impacts of drought, as observed in the previous studies (Hezron et al., 2007; Mubea et al., 2009; Morgan, 2009; Thome, 2009; Mwenya, 2009; Roncoli et al., 2010) cannot be underestimated as it affects the overall food production levels and the government grain reserves. This research is designed to assess the drought levels in the agricultural potential zone of Nakuru using remotely sensed data to determine the drought severity levels based on temperature and vegetation health. Existing Work in the Field and Gap in the State of Knowledge National and local organizations have devised mitigation measures against the severe impacts of drought in Nakuru. For example, the Kenya Wildlife Service (KWS), an authority for wildlife service in Kenya has intervened by drilling boreholes, constructing water troughs and pumping water to the troughs for wildlife in the National parks. Furthermore, the authority, as a result of receding pasture has relocated the large herbivores (e.g. elephants, rhinoceros and buffaloes) to other parks (Mulanda and Macharia, 2009). The urban dwellers in the region have also devised strategic methods of adaptation. In order to cope with the declining purchasing power resulting from increasing cost of food and dwindling food supply to the urban dwellers, urban households have been forced to diversify their livelihoods by introducing urban farming in compounds, along streets, river banks, under power line and any other open spaces in the urban areas (Africa Studies Centre, 2006). A major setback to the present drought monitoring scheme in Nakuru County and other high potential agricultural regions in Kenya is the lack of early drought warning programmes or strategies from the government. Nakuru is unlike the arid and semi arid regions which have adequate early warning projects such as the National Drought Management Authority (NDMA) formerly Arid Land Resource Management Project (ALRMP). ALRMP, a project funded by the World Bank and the Government of Kenya, developed a remote sensing methodology for drought monitoring that relies on NOAA-AVHRR rainfall estimate (RFE) and normalized vegetation index (NDVI) input data. This remote sensing drought monitoring method is supplemented by semi structured questionnaires that are administered by sentinel monitors in the interior rural areas. These questionnaires are designed to capture personal household details of the inhabitants, such as income level, malnutrition data, conditions of livestock, water and pasture accessibility and commodity prices in the markets (ALRMP, 2011). The limitations associated with the remotely sensed drought monitoring scheme of ALRMP is availability of NDVI data after the sensor failed in 2010. Other associated problems include reliable time #" "

series precipitation analysis to quantify pasture through normalized difference vegetation index as well as unclear relationship on the primary and secondary data analysis thresholds. Although this method still provides very effective means of drought monitoring in low agricultural potential districts (i.e. arid and semi arid land of Kenya), the results from these analysis are not reliable in high agricultural potential zones of Kenya such as Nakuru. In the neighboring war torn country Somalia, the existing drought monitoring methods such as Combined Drought Index (CDI), developed by United Nations Food and Agricultural Organization under Somalia Water and Land Information Management System (FAO-SWALIM), based in Nairobi with monitoring field stations in Somalia, has successfully managed to cover a portion of the country, using the few existing rainfall stations managed by the humanitarian organizations and local authorities. This assessment method has had its adequate applications in data scarce environments in Somalia and parts of Kenya (Balint and Mutua, 2011). The major limitation of this method is the spatial coverage considering the fact that the scheme heavily relies on in-situ meteorological data which is poorly distributed and susceptible to gross measurement errors. The main strength of the method applied in this research is its capability to integrate both temperature and vegetation vigor values for detection of moisture levels which is a reflection of moisture contents in the leaf. The method however, is not limited to one form of data, as a variety of remote sensing data with thermal bands and vegetation detection capability exists making it one of the most Versatile option. An example of such data is MODIS, SPOT, NOAA-AVHRR and LANDSAT data among others. Drought There is no clear definition of drought; it only depends on the context and regions. Scientist over time tried to define this phenomenon, but still there is no clear definition. Previous observations observed that there was no clear universal definition of drought because its perception is dependent on water deficit, Nguyen, (2006). Other observations as of Palmer, (1965) defined it as a prolonged and abnormal moisture deficiency, whereas Byun. H, (2010) defines drought as a relative term to normal conditions while water shortage is absolute term for water demand. In an attempt to contextualize to tropical region, SDMC a disaster management Centre in New Delhi, defines it as; a climatic anomaly characterized by deficient supply of moisture resulting either from subnormal rainfall, erratic rainfall distribution, higher water needs or combination of all these factors, SDMC,(2012). According to SDMC, (2012), droughts are however resultant of acute water shortage due to lack of rains over extended period of time affecting various human activities leading to crop failure, un replenished ground water resources, depletion of water in lakes, reservoirs, shortage of drinking water and reduced fodder availability, SDMC,(2012). Drought Categories Wilhite and Glantz (1985) analyzed more than 150 drought definitions and broadly categorized them into four main categories; meteorological, Agricultural, Hydrological and Socio economic droughts. The droughts further are classified in the context of permanent, seasonal, contingent and invisible, J.O Ayoade, (2004). Meteorological Drought: The meteorological drought gas been defines as the degree of dryness specified by deficiencies of precipitation and the duration of the dry spell. (WMO, No.1006, 2006: A. Schuman, 2006). Meteorologist makes distinction between absolute and partial drought. Absolute drought I is said to be a period of 15 consecutive days to none of which is credited with 0.2mm or more of rainfall, whereas as partial is a period of at least 29 consecutive days of which its mean rain$" "

fall does not exceed 0.2mm, J.O.Ayoade, (2004). Agricultural Drought:The Agriculturalist perspective of drought is when moisture storage available through rainfall or soil is insufficient to ensure optimal crop growth as in, J.O Ayoade (2004). Agricultural drought has a centre focus in the precipitation deficits and its impacts. It has great relationship with the meteorological and hydrological characteristics, as observed (A. Ellis, 2010) in the difference between the precipitation and potential evapotranspiration balance. The difference between the actual evapotranspiration and reduced water content in soil and reservoirs levels, as affects plant water demand, is evident in both plant physical and biological growth properties, Parul Chopra, (2006). Hydrological Drought: To hydrologist, drought is as a result of low flow in rivers below a critical threshold discharge. Hydrological drought has been defined in various interrelated versions by many researchers as significant decrease in availability of water in all its forms, i.e. Surface water, stream flow, Lake Reservoir levels, ground water and ground water levels appearing in the land phase of hydrological cycle as in the case of Khana,(2009). Hydrological drought may be the result of long term meteorological droughts that results in drying up of rivers and decline in ground water levels, Rathore, (2004). Descriptions such as those reflected in Tallaksen and Van Lanen, (2004) that refers hydrological drought as sustained and regionally extensive occurrence of below average natural water availability. This definition seems to agree with the fact that the drought phenomenon is closely associated with long term absence of precipitation, and prolonged or increased evapotranspiration. Socio Economic Drought: The socio economic drought as observed in the case of Wilhite and Glantz, (1985), is said to occur as a result of physical water shortage that ends up affecting people at individual scale. The greater demand on commodities than the supply of economic good can best describe this situation. AridLandsResourceManagement Project(ALRMP), Baringo, 2009 describes this impacts of drought situation through the observation on the livestock body conditions, where the livestock body conditions were found to be deteriorating, with decline in milk supply affecting the market prices, whereas in the agro pastoral areas, crop failure cases reported, affecting the market cereal availability and farmers purchasing power. This scenario clearly reflects the creeping drought in the socio economic context. The socio economic droughts reflects the elements meteorological, agricultural and hydrological is drought as it results from the absence or low precipitation, vegetation content reduction for forage, stream flow reduction and declining levels of water as reflected in drying up and diminishing recharge capacities of aquifers. Impacts of Droughts Economic impacts The massive loss of livestock in Baringo and significant loss of rangeland and pasture marked a record observation on stock population changes in Baringo (1983-1985) from onset of the end of severe drought, Homewood et al, (1987). Drought events in many cases impose negative impacts on environment and causes widespread structural damages as observed in, Akhtari et al, (2011). Increased insect infestation, wind erosion, plant and animal and diseases as well as forest and range fires in India are commonly observed impacts in India, Chopra, (2006). These events are common to drought prone areas in the greater horn of Africa, and in Baringo County. The effects of drought are clearly manifested by reduced crop production, loss of agriculture, land degradation, livestock population deaths unemployment and health problems, Murad et al, (2011). The %" "

most common economic impacts in Baringo county (ALRMP,2009) is associated with wasted animal body conditions, reduced milk production, direct loss of browse and pasture, predation of small ruminants by the Hyena and baboons as well as crop failures is common. Environmental Impacts The environmental impacts in Baringo county as per (ALRMP, 2009-2011) is realized through the hydrological effects where the water sources e.g. water pans, rivers dry up and the reporting of reduced levels of water by flowing springs and drying up of the available boreholes. The loss of biodiversity, natural habitat, degradation of landscape, increased soil erosion leading to permanent loss of land productivity and the loss of wetlands impacts negatively on plant and animal species and the ecosystem. These key elements induce migration of host communities as of the case of Baringo County (ALRMP, 2009) where the pastoral and agro-pastoral communities migrate to search for better and quality pasture and water. Social Impacts The typical social impact as a result of drought stress includes the conflicts and public safety, health and nutrition affecting quality of life, population migration and increased poverty. For the case of the study area, the malnutrition levels has been reported to be declining as stated in ALRMP, 2009 bulletins, that resulted from unavailability of essential commodities in markets and declining milk production by livestock. The loss of human lives through protracted drought impacts occasioned by increased heat stress and declining purchasing power in Kenyas arid and semi arid counties has been a key cause of water and management conflicts among the pastoral and agro-pastoral livelihood zones. Agricultural Drought Indices Normalized Difference Vegetation Index (NDVI) The NDVI is a dimensionless variable. The index can be used to provide information for agriculture and vegetation health situation. This information is useful in determining water stress levels in vegetation and estimation of crop yield (Penuelas et al, 1993 and is useful in drought assessment (Tucker, 1980).It provides information of vegetation health that can be used as a means of monitoring changes in vegetation over time. The healthy vegetation absorbs most of the visible light that it receives and reflects a large proportion of the near infra red light. Unhealthy or sparse vegetation reflects more visible light and less near infra red, Frantzova (2010). Research by Thomas, et al (2004) observed that, NDVI has limited capability for estimating vegetation water condition as it is affected by other variables. Ceccato et al (2002b) in Thomas, T.J et al, (2004), summarized the limitations of using NDVI as ; different plant species has their own relationship of chlorophyll content and vegetation water condition, and a decrease in chlorophyll content does not imply a decrease in vegetation water condition, whereas decrease in vegetation water condition does not imply a decrease in chlorophyll content. Another limitation observed by Tsegaye Tadesse (1998) is that, in cases of extended periods of cloud coverage, the NDVI values tend to be depressed giving a false impression of water stress or drought condition. To remove this effect, the temperature condition (TCI) is used. The TCI is derived from brightness temperature (BT) and its algorithm is similar to that of NDVI vegetation response. The combination of VCI/TCI is also used to estimate vegetation stress. Normalized Difference Water index (NDWI) The NDWI is a dimensionless product whose value ranges between -1 to +1 depending on the leaf wa&" "

ter content, vegetation type and cover. High values of NDWI correspond to high vegetation water content and high vegetation fraction cover and vice versa, whereas its anomalies are in standard deviation units commonly ranging from -4 to +4, Joint Research Council (JRC, 2011). The NDWI anomalies are more dependent of time series available to calculate the mean values and the standard deviations. To achieve better results, the period should be long enough to characterize the area study area. The indices are calculated as normalized difference vegetation index but the red band is replaced by the short wave infra red band, (SWIR) canal of (1580- 1750um) of SPOT VGT. NDWI is very efficient in the domain of stress because it is sensitive to soil moisture content, vegetation cover and leaf moisture content, Tychon et al (2007). Some of the noted weaknesses of NDWI are in its susceptibility to soil background effects on partial vegetation cover. Drought and water stress are not the only factors that can cause a decrease in the NDWI values or anomalies. The Change in land cover or pest and diseases can also be responsible for such variations of the signal, Gao et al (1996). To enhance the information obtained from the index, the indicator must be used jointly with other indicator that gives more information on precipitation and soil moisture to determine the vegetation response in case of drought investigations Gao, et al, (1996), JRC (2011). MODIS NDWI has been used to detect and monitor the moisture conditions of vegetation canopies, Xiao et al (2002), and has been tested as drought indicator (Gu, et al, (2007) and found that the values exhibited quicker response to drought condition than NDVI. Water Supplying Vegetation Index (WSVI) Water supplying vegetation index is an indicator based on the relationship between normalized difference Vegetation Index and surface temperature. This method provides effective method of estimating surface moisture condition Luke, et al, (2001).WSVI is an index that combine both aspects of vegetation and temperature as observed byXiao, et al, (1995). In the classifications of WSVI, Higher values are indicative of greater moisture amounts and in this situation canopy temperature are lower and NDVI values are higher, whereas the lower values is and indicative of drought. WSVI values ranges from a value of -4 for extreme drought to +4 for highly moist condition, Luke, et al, (2001). Jiyuan, et al, () observed in his applications that, WSVI takes into account the effects of vegetation reflection in red, near infra red and thermal band. It was found that, the method is effective where NDVI is greater than 0.3. Jain et al (2010), observed that, in drought condition, NDVI values derived from satellite data will fall below normal while the crop canopy temperature will rise above normal both of these effects are related to available water supply, and combining both would yield a sensitive measure of drought condition. The visual output of WSVI was compared with PDSI in Luke et al (2001), showed similarities in estimates of general moisture conditions. Normalized Difference Drought Index (NDDI) The Normalized Difference Drought Index (NDDI) is another strong drought index. The NDDI combines information from both the NDWI and NDVI data derived from the satellite data Gu, et al, (2007). NDDI is found to be more responsive and have wider dynamic range values than a simple NDVI and NDWI differencing through drought periods, Charat, et al, (2006), Liu Cheng-lin , et al, 2008).

'" "

Charat, et al, (2006) observed that, the NDVI and NDWI values decreases with decreasing slope gradient of cumulative rainfall while rapid increase of NDDI values during dry months of the years, thus more sensitive to water content than NDVI and making it better index for drought identification. In studies undertaken using MODIS NDVI and NDWI over great plains of United States, it was found that NDDI had stronger response to summer drought conditions than a simple difference between NDVI and NDWI and is therefore more sensitive indicator of drought in grass land than NDVI alone. It was also observed that, NDDI values increased during summer condition s which demonstrated an additional indicator for large grassland drought monitoring, Gu, Y. et al, (2007). Land surface temperature (LST) Land Surface temperature is how the surface of the earth would feel to touch in a particular location. Monitoring of land surface temperature enables critical assessment of the influence and how the surface is influenced by weather and climate Patters. It is sometimes referred to as the surface skin temperature of the earth. LST is a very important variable required for a wide variety of applications for instance climatological, hydrological, agricultural, biochemical and change detection studies, Prasanjit Dash (2005). The LST as a climatic variable is related to surface energy balance and integrated thermal state of atmosphere, Jin, (1999).it acts as an indicator of climate change due to its upward terrestrial radiation influencing sensible latent heat flux exchange of air. Yin, (2007) Land surface temperature therefore can provide information about surface physical properties and climate which plays a role in environmental processes, Javed et al, (2008). The LST research shows that land surface temperature varies with surface soil water content and vegetation cover, Weng et al (2003) that the higher latent heat exchange is found with vegetated areas while the sensible heat exchange found in sparsely vegetated and urban areas. The land surface temperature is sensitive to vegetation and soil moisture and it can therefore be used to detect land use, land cover changes, Javed et al, (2008). LST validation is difficult because derived quantity is representative of the whole pixel, while point temperature measurement covers a short distance; hence a field validation is possible for homogeneous areas e.g. dense vegetation, desert and others, Dash, (2005).Dash, (2005) observed problems that are associated with LST and summarized them as follows; i. ii. The surface emitted radiance is altered by atmosphere before reaching the top of Atmosphere (TOA) sensors. Radiance measurement by sensor are made in one direction which is not necessarily representative for upper hemisphere, hence angular characterization of emissivity is difficult depending on anisotropy and Separation of temperature from surface radiance is unfeasible because of under determination for sensor with spectral channels.

iii.

Land surface temperature estimation methods There are three main methods of estimating LST. i. Single channel method. In this method, Top of Atmosphere (TOA) radiance is directly compared with radiative transfer calculations of known land surface emissivitys and land surface temperature are derived. This method is accurate but it needs exact atmospheric information. (" "

ii.

iii.

Split Window Technique (SWT)/Multi channel method. This is based on differential absorption in two spectral channels within 10 12!m atmospheric window and land surface temperature is related to this measurements. Multi angle Method. This method is similar to SWT, but the differential absorption is due to different atmospheric slant path-lengths when the same target is observed under different viewing angles in the same spectral range.

Land surface emissivity and its retrieval methods Emissivity is defined as the radiation efficiency of a real world surface as compared to a blackbody radiator. A black body is the hypothetical object that absorbs all radiation that falls on it. It is also defined as the ratio of emittance from a body to that of blackbody (a perfect emitter) at the same temperature. Sobrino et al (2004) developed and proposed an algorithm to compute emissivity (") of mixed pixels composed of bares soil and vegetation. Data and Methods The data type to be used is Landsat 5 and 7 satellite image from USGS for 2000, 2010, 2011 with varied spatial resolutions of 15m at panchromatic, 30m at Visible and near and mid Infra red and 120m for TM and 60m for ETM+ at Thermal infra red and NOAA-RFE(Rainfall) data of 8km spatial resolution. A stepwise processing chain was established based on the inputs and flow involved in the extraction of quantities in the satellite image. The processing tools that will be utilized in this research will be the use of ESRI ArcGIS products, Leica Geosystems products; ERDAS IMAGINE and Microsoft suite. The output results for the NDVI data generated in 30m spatial resolution of Landsat TM bands 3 and 4 will be resampled to 120m cell resolution to overlay with the Land surface temperature output generated from band 6 of the same image with 120m spatial resolution to enable cell statistics analysis in the assessment. Water bodies urban, artificial surfaces and bare rocks layer will be developed from the globe cover land use classes and will be used as a mask in the final NDVI and LST layers for analysis, so that results will not be biased as this land use classes will affect the result. The processing flow below outlines the major processing steps to be undertaken and the final output of the project as agricultural drought map. Processing flow

Fig 1:

)" "

Processing chain Normalized Differnce Vegetation Index extraction The NDVI is a dimensionless variable. The index can be used to provide information for agriculture that can be used to determine water stress levels in vegetation and estimation of crop yield. Penuelas et al, (1993) and is useful in drought assessment (Tucker, 1980). NDVI is expressed as follows: ; (1) This is similarly Landsat TM, TM 3 (0.66 !m, red band and TM4, (0.83 !m near infra red band) respectively, are used to compute vegetation indices and NDVI, therefore based on these concept, the same can be expressed in LANDSAT TM and ETM+ data as; ; (2) Atmospheric correction based on image data In order to obtain accurate NDVI values that are more representative, the top of the atmosphere values (TOA) has to be corrected and this can be computed by use of the algorithm developed by Chavez (1996) as provided below in, Quinqin et al, (2010);
:

Chavez, (1996) (3)

Where; Lsat = radiance at sensor, d =Earth-Sun distance, Eo = Spectral solar irradiance on top of the atmosphere, #z =Solar Zenith Angle, Tz =Atmospheric transmissivity between sun and surface Lp = irradiance resulted from interactions of the electromagnetic radiance with the atmospheric components (molecules and aerosols) that can be obtained as; Lp = (Lmin - L1%) (4) Where; Lmin is irradiance that corresponds to digital count value for the sum of all pixels with digital counts lower or equal to this value of 0.01% of all the pixels from the image and is expressed as; ; (5) Where Tz (Atmospheric transmissivity) of TM 3 and TM 4 is 0.85 and 0.91 respectively The spectral solar irradiance for the Landsat TM and ETM+ quantities to be applied was sourced from in, Quinqin et al, (2010) as shown in the table below;

TM and ETM+ Solar Irradiance (Eo) (Wm-2x!m Band 1 2 3 4 5 7

*" "

TM4 (Markham &Barker, 1986) TM 5(Neckel & Labs (1984) ETM+ (Igbal (1983)

195.8

182.8

155.9

104.5

21.91

7.457

195.7 1969

182.9 1840

155.7 1551

104.7 1044

21.93 225.7

7.452 82.07

Table 1 Solar Irradiance (Quinqin et al, (2010) Radiance and Temperature brightness retrieval The radiance is computed using the algorithm; L$= Gain* DN +Bias (Landsat7 Science user data Handbook Chapter 11, (2002); (6) Where: L$ is Radiance, DN is digital Numbers values recorded, Gain is (L!max L!min)/255 (slope of the response function). Bias is the Lmin (intercept response function). L$max is the highest and L$min is the lowest radiance measured at detector, (saturation in MWcm-2r-1). In the Landsat 5 metadata, L$max and L$min were obtained from metadata to be 15.303 and 1.238 respectively. The Sensor calibrations constants for the Landsat TM and ETM+ is expressed in the table below TM and ETM+ Thermal Band Calibration Constants Constant 1-K1 (Wm-2sr 1 !m) ETM+ (Markham &Barker 1986) TM (Irish, 2000) 666.09 607.76 Constant 2-K2 (Kelvin) 1282.71 1260.56

Table 2 Thermal Band Calibration; NASA, Quinqin et al, (2010) The radiance brightness temperature was thereafter extracted based on the sensor algorithm available in the Landsat Handbook in the equation 7; : (7) Where, TB is the sensor brightness temperature in Kelvin (K), K1 is calibration constant 1 equal to 666.09 Wm-2sr -1!m, K2 calibration Constant 2 equal to 1282.71 watts/m2sr !m, L$ is spectral radiance expressed in Watts/m2sr !m. Refer Table 3: Thermal Band Calibrations.

!+" "

Land surface Emissivity Estimation Sobrino et al (2004) developed and proposed an algorithm to compute emissivity (") of mixed pixels composed of bares soil and vegetation as expressed in the equation; " = "v PV + "s (1- PV) + %E; Sobrino et al, (2004); (8) Where "v and "s is emissivities of vegetation (0.99) and soil (0.97) respectively, %E is the effect of geometric distribution of natural surfaces and internal reflection with plain surface assumed to be of negligible unit and heterogeneous and rough surfaces e.g. forest among others takes a value of 0.55. Sobriono et al, (2004).The term PV is vegetation proportion obtained according to Carlson and Ripleys (1997) as expressed in the equation below;
;

Carlson and Ripleys (1997); (9)

Where; NDVI max = 0.5 and NDVI min = 0.2 Land Surface Temperature retrieval The brightness values obtained was therefore converted to land surface temperature. The algorithm for conversion applied is as shown below; ; (10) Where: LST = Land surface temperature, ! = Wavelength of emitted radiance for which the peak response and the average of limiting wavelength is spectral radiance. (!=11.5!m) Markham and Barker, (1985) will be used, & = hc/', where ' is Stefan Boltzmann constant, h=Planck constant and c=speed of light in a vacuum, TB = Sensor Brightness temperature and ln is the natural logarithm to base10 (alog) " = surface emissivity. Drought Assessment using WSVI Index The Water supplying vegetation index (WSVI) is one of the indices that were developed to combine the NDVI and the temperature (Land surface temperature) to detect the moisture condition, Luke et al (2001). The Expression for this index was developed by Xiao et al, (1995) and is expressed as shown below ; Xiao et al, (1995), (11) Where, LST is Land surface temperature. NOAA-AVHRR RFE Performance in the period of study The NOAA-AVHRR rainfall estimates will be downloaded and the long term mean average of ten years will be established for the area of study. The precipitation performance for the years of study will be compared with the long term mean average performance to assess deviations from normal trends. Graphical and image trends will be derived to establish the rainfall changes from the long term mean. !!" "

The Study Area This research is focused on Nakuru County, a region in Kenya bounded between latitude 0.28N and 1.16S, and longitude 36.27 E and 36.55E. Nakuru has a land mass of 7,495 km2 (CRA, 2012), with a population of 1,603325 and growth rate of 3.4% per annum (CBS, 2009). The fertility of Nakuru can be traced to the geology of the area. Nakuru lies within the Great Rift Valley which is characterized by volcanic activities and volcanic land formations. According to studies by Wegulo et al. (2010), the soils in this study area are molic andosols that developed from volcanic ashes and pyroclastic rocks from recent volcanoes. These soils are well drained, deep to moderately deep firm clay loam with humic top soil of high fertility.The study area has two agro-ecological zones that are classified as Upper Highlands (UH2) with altitude ranging from 2580m to 2800m above the mean sea level. Nakuru has an average mean temperature of 12C to 13.7C with an annual average rainfall of 1100 mm to 1400 mm per annum. The altitude of the Lower Highlands (LH3) range between 1890m and 2190m above the sea level while the mean average temperature varies from 15C to 17.5C with an annual rainfall average of 810 mm to 1100 mm. The map of the study area is shown in

Figure 1. Results and Analysis Agricultural drought severity levels. The data used to generate the water supplying vegetation Index to determine the moisture levels, was the land surface temperature and normalized difference vegetation Index from Landsat Image. The !#" "

moisture levels were derived from the NDVI and land surface analysis. The index values were reclassified into five (5) major vegetation moisture levels classes and applied to all analysis.

Class 1 2 3 4 5

Index (-0.00157) - (-0.00133) (-0.00133) - (+0.0084) (+0.0084) - (+0.0155) (+0.0155) - (+0.0243) > (+0.0243)

Severity Levels Very Low Moisture Low Moisture Moderate High Moisture Very High Moisture

Table: Moisture Classification The results indicates that about 39.71% of the study area in the year 2000 experienced moisture deficit, the vegetation moisture levels ranged from low to very low, which is an indication of poor rainfall performance well below the long term mean average, with poor vegetation health implying severe drought. The forested areas and areas with higher vegetation density around lakes had high to very high moisture levels, with areas that are normally of high moisture having moderate moisture levels, a clear indication of vegetation stress. The year 2010 indicated very adequate moisture levels in the vegetated area. Low to very low moisture levels was 7.03% with high o very high moisture having 67.36% coverage and moderate levels of 25.61%.This indicates that, the years rainfall and vegetation performance adequate enough and better harvest would be realized. The rainfall and vegetation performance is confirmed by its performance in the year as shown in figure 4.2b (Graph of Rainfall and Vegetation 2010). The rainfall and vegetation health was well and above its normal long term mean average. In the year 2011, he results indicates that, regions of low to very low moisture levels covered 12.24%, with high to very high moisture levels taking 59.99% of the area with moderate moisture levels of 37.77% Coverage. This indicated that the years rainfall and vegetation performance was very stable and pasture and browse for livestock as well as crops were adequate and doing well. Precipitation and NDVI deviation over the long term average The data utilized to achieve this result was derived from NOAA-AVHRR. The satellite images has a resolution of 8km by 8km.A long term mean average for the study area (1981 to 2008) was derived and used for comparison for rainfall and vegetation performance in the specific interest years of 2000,

!$" "

2010 and 2011.The graphical analysis was derived and analyzed in simple Microsoft excel as shown below.

Figure 4.2.1: Graph of Rainfall and Vegetation 2000 The rainfall performance for the first agronomic season in the county normally referred to as long rains season in Kenya, (March to May) recorded rainfall that was below long term mean average, and the vegetation health was poor as it fell below the long term average mean and failed to recover over time. The vegetation health improved gradually with improved rains that were close to the long term normal average especially between second dekad of June through to first dekad of august. Short rains season is usually expected to start on August through to November realized improved rainfall performance that was above long term mean average in the third dekad of October and the rest of November. December also recorded above normal rainfall on its first and second dekad. The vegetation health in the second season was well above normal in some instances and close to normal indicating that, the pasture, browse and crops performed well in the season. The region experienced low rainfall in its first season, implying that most rain fed Agricultural crop production failed as a result of prolonged deficits of rainfall. This is evident by precipitation that was below the normal long term average. This is evident by the vegetation health situation that fell to a low of 0.29 against its long term normal average of 0.41. Rainfall and vegetation, 2010 The graph below shows the rainfall performance in the year 2010.The senor failed to provide vegetation data from the second dekad of March, however, SPOT vegetation data for the year was used to provide the remaining average data.

Figure 4.2.2: Graph of Rainfall and Vegetation 2010 The rainfall season started very early this year, as January received rainfall high above normal long term average. The performance for the first agronomic season (March to May) was very good. The first dekad of March received the higher rainfall of 127.22 mm against its long term mean average of !%" "

25.09mm, with the first dekad of January receiving exceptionally low of 1.38mm against its long term mean average of 9.72mm. The first season from the results received good rainfall and normal planting is expected to have taken place in time. The vegetation health in the first season was well above long term normal average, indicating that the pasture or forage and crops were performing well. The rainfall in the second season (August to November) started in time and performed well above the long term normal average throughout with the end of the season; November realizing dwindling levels of rainfall below its long term normal average. The vegetation performed well throughout the seasons. The vegetation continued varying well above the long term mean average throughout. This implies that the year realized abundant crop harvest in the region. Rainfall and vegetation, 2011 The graph below shows the rainfall performance in the year 2011.The vegetation data for the year was not available as a result of NOAA-AVHRR sensor malfunction

Figure 4.2.3: Graph of Rainfall and Vegetation 2011

Rainfall performance at the beginning of the season, (March to May) started very well, with first and second dekads of March receiving rainfall above its long term normal average. The season ended with dwindling precipitation that gradually picked soon after. The second season (August to November) started well with first dekad of the beginning season receiving rainfall above long term normal average. The remaining months received rainfall that was well distributed fluctuating within the normal and above the normal mean, an indication of good performance of crops, pasture and browse for livestock. Land Surface Temperature (LST) and Emissivities The information obtained from the analysis was re-classified into five major classes using statistical natural jenks method in ArcGIS software. The range was slightly modified to capture more information that would lead to meaningful detection of feature quantities that are likely erroneous in the analysis. This decision was made based on the climatological temperature data of the study area which is known to be between 12C to 13.7C. The result obtained was used as base information and applied uniformly to the subsequent years study. The table below shows the results of the classification. !&" "

Class 1 2 2 4 5

Temperature Range (oC) 0.0 - 12.82 12.82 - 25.64 25.64 - 32.06 32.06 - 37.82 >37.82

Class Level Very Low Low Medium High Very High

Table Temperature Classification The results obtained after applying the classification range was generated as shown in the map below, and the chart alongside was generated to show class spatial coverage in percentage 2000. The land mass under high temperature (32.06oC - 37.82oC) and very high temperature (>37.82 oC), is 27.59% and 21.69 %, which makes up to 49.28% which is very large area. This result indicates that most of the crops and other vegetation (pasture) were water stressed, or under drought situation. About 21.98% of the vegetation in farms that are near or close to forested areas, were within the medium temperature range of category of 25.64 oC - 32.06 oC, implying that, some crop produce could have chances of survival, though they may not adequately reach their maximum output. The very low temperature class with a range of between; 0.0 oC - 12.82 oC was identified in pixels contaminated by clouds and areas of no data. The Land surface emissivity used to obtain the land surface temperatures was generated, and the emissivity variation over land indicated low of 0.8389 and high of 0.9944. The emissivity variation over dense vegetation was at 0.9711, Water Bodies, 0.9894, while mixed Vegetation (Agricultural land) had 0.9811.The table below (table 4.3b) shows more details on the features standard deviations, minimum and maximum recorded values, while and the Map (figure 4.3b) shows detailed spatial variations. In the year 2010, the results reveal that about 0.8% and 5.78% of the land was within low and very low temperatures, and 64.63% (25.64 oC to 32.06 oC) was in medium, while areas with high and very high temperatures had 24.86% and 3.93% respectively. Compared to the year of 2000, this year significantly improved, as the land mass that was within low and very low temperatures in 2000 was about 49.28% compared to this year which was 28.79%, a difference of about 20.49%. The areas under low (12.82 oC - 25.64 oC) and very low (0.0 oC - 12.82 oC) temperature are low The land surface emissivities in this year showed variations with water remaining highly emissive followed by agricultural and forested areas. The table (table 4.3.3) and map below shows the detailed variations and the standard deviation over the feature class characteristics.The emissivity variations in this year showed clear distinction between water bodies, agricultural land, and dense forested areas. The temperature variation in this year reveals that, low, medium and high temperatures dominated most of the parts of the study area, with medium temperatures, (25.64 oC to 32.06 oC) realized in covering about 39.2% of the area. Low temperature between (12.82 oC - 25.64 oC), covered significant portion of the area with about 30.92% coverage.

!'" "

From the rainfall performance data, the first season of this year started early and very well above the normal average and performed steadily throughout to the next season, this year, appears to be normal as the rainfall fluctuated within normal and above the normal in most cases. The land surface emissivity revealed very strong relationship supporting the vegetation performance. The average emissivity of this year was at 0.9765 for agricultural land same as the emissivity realized in the previous year where the rainfall and vegetation performed well above the normal, whereas the dense vegetation was at 0.9702 compared to 0.9704 the previous year (2010) .The map below shows the surface emissivity variations over space, whereas the table shows average emissivitys over land use classes. CONCLUSIONS AND RECOMMENDATIONS The Combination of NOAA AVHRR rainfall estimates and normalized difference vegetation index, provides very useful information for a drought monitoring and early warning system. This is indicated by the graphical analysis that indicated by the continued decline in vegetation index precipitation over its long term mean average. The use of temperature and vegetation index provides adequate means for mapping drought extend over the agricultural fields, though the method of deriving the emissivities and land surface temperature is highly technical for quick results though scripts can be developed to derive the data easily. The existing methods take into consideration a number of scientific factors that may not be professionally friendly. Use of time series data offers potentials to establishing long term mean average for vegetation moisture levels that will provide adequate monitoring for effective early warning system to the farmers. The land surface temperature provides adequate indication of moisture levels on agricultural land near dense vegetation cover such as forests which is an indicative of adequate moisture in plants, nevertheless, the produce is uncertain due to extreme high and low temperatures experienced by the vegetation. The drought period indicates that the extreme low and high and very high temperatures dominates the land (study area) causing more destructions to the crops and livestock production. The values obtained in the (Figure 4.3.2) for the year 2000 reveals this result, which was converges with other observed researches in the same study area such as that of Roncoli et al,( 2010) where she observed that during drought in Nakuru, prolonged cold weather destroying crops and reducing livestock production, harsh foggy, windy and frost moments affecting pollinators is experienced. Time series analysis for vegetation moisture monitoring is recommended for these studies in order to establish long term mean average that can be used for monitoring and provision of timely warnings to the farmers soon the moisture level performance constantly remains below average.

References Africa Studies Centre, (2006). Information Sheet: Farming as a livelihood Source from urban dwellers; Results from a research project in Nakuru, Kenya. Retrieved on 26th July, 2012 www.fao.org/fileadmin/templates/FCIT/pdf/african_studies_centre_info_sheet.pdf

!(" "

Akhtari, R. Bandarabadi, B. (2001). Spatio temporal patterns of drought in North East of Iran. Options Mediterranean Series A, No. 80; pp 71-77 www.iamz.ciheam.org/medroplan/a-80OPTIONS/session1/(069-78)08Akhtarpsi.pdf Andrew, E. and Lenart, M. (2010). Introducing moisture balance drought index. South West Climate Outlook, USA Arid lands resource management project, (ALRMP, 2009, November 30th); Drought Monthly Bulletin Baringo County; Ministry of State for Development of Northern Kenya and other Arid Lands. Government of Kenya. Ayoade, J. (2004). Introduction to climatology for tropics. (Eds.) First published by John Wiley in 1983: ISBN 978029-373-6, Reprinted by Sam Adex Printers Felele rab, Ibadan, 2004 Balint, Z. Mutua F. (2010). Drought monitoring with combined drought index. Food and Agricultural Organization of the United Nations project on Somalia Water and Land Information Management System (SWALIM) Bruce, C. Hilder, D.(2004).Pre-processing Methodology for Application to Landsat TM/ETM+ Imaging of the wet tropics. Cooperative Research Centre for Tropical Rainforest, CRC, cairns (44pp), link: www.rainforest-crc.jcu.edu.au Brunsell, N and Gillies, R.(2002). Incorporating Surface emissivity into a thermal atmospheric correction. Photogrammetric Engineering and Remote Sensing, Vol 68, No, 12, pp1263-1269 Chander, G. Markham, B. Helder, D.(2009).Summary of Radiometric Calibration Coefficient for Landsat MSS, TM,ETM+ and EO-1ALI Sensors. Remote Sensing of Environment (113) 893-903 Charat,M. Wattanakij, N. Kamchai,vT. Mongkolsawat, K. Chuyakhai, D. (2009). Exploration of spatio-temporal drought patterns using satellite derived indices for crop management in Northeastern Thailand. Retrieved on 12nd January 2011 from: www.a-a-r-s.org/acrs/proceeding/ACRS2009/papers/oralpresentation/T-2804.pdf Chavez, P. (1996). Image based atmospheric corrections revisited and improved. Photogrammetric engineering and remote sensing. Vol.62, No.9. pp1025 1036 Cheng-lin, L. Jian-jun, W. (2008). Crop drought monitoring using MODIS NDDI over mid-territory of China. Geosciences and Remote Sensing Symposium. IGARSS 2008.IEE International. pp 883-886. doi: 10.1109/IGARSS.2008.4779491.E-ISBN: 978-1-14244-2808-3 Chopra, P. (2006). Drought risk assessment using remote sensing and GIS: A case study of Gujarat, Unpublished MS thesis of India, Joint Master of Science awarded by ITC, Netherlands and IIRS, India Commission on revenue allocation, (CRA, 2009). Nakuru County.Government of Kenya fact sheets. Retrieved on 28th May 2012 from: www.crakenya.org/wp-content/uploads/downloads/2012/04/12/2-FORMULAFORSHARING-REVENUE-AMONG-COUNTY-GOVERMENTS-26TH-2012.pdf !)" "

Dash, P. (2005). Land surface temperature and emissivity retrieval from satellite measurements. A PhD dissertation submitted to Institute of Meteorology and Climate Research Forschungszentrum, Karlsruhe, University of Karlsruhe, Germany Frantozova, A. (2010). Remote sensing, GIS and Disaster management. 3rd International Conference on Cartography and GIS, 15-20 June 2010, Nessebar, Bulgaria Gao, M. QIN, Z. Zhang, H. Lu, L. Zhou, X. and Yang, X. (2008).Remote sensing of Agro-droughts in Guangdong province of China using MODIS satellite data. Sensors, 8, 4687-4708. Doi: 10.3390/s8084687 ISSN 1424-8220 Gu, Y. Jesslyn, F. Brown, J. Verdin, P. Wardlow, B. (2007). A five year Analysis of MODIS NDVI and NDWI for grassland drought assessment over the central great plains of United States: Geophysical Research Letters, Vol 34:L06407, doi: 10.1029/2006GL029127 Homewood, K. and Lewis, J.(1987). Impacts of drought on pastoral livestock in Baringo, Kenya 1983-85. Journal of Applied Ecology Vol.24, No2.1987, pp 615 -631 Jain, S. Keshri, R. Goswami, A. Sakar, A. (2010). Application of Meteorological and Vegetation Indices for evaluation of Drought Impact: A case study for Rajasthan, India: National Hazards, (540: pp643-656: DOI 10.1007/s11069-009-9493-x Javed, M. Yogesh, K. and Bharath , B. (2008). Estimation of land surface temperature over Delhi using Landsat 7 ETM+. J.Ind. Geophys.Union Vol 12, No.3 pp131-140 JRC, Desert Action-LMNH Unit, (2011). NDWI: Normalized difference water index product factsheet-Europe, Version 1 December, 2011: Institute for environment and sustainability Khanna, M. (2009). Hydrological drought indices; Water technology centre, Indian Agricultural Research Institute, New Delhi, India Lanen, H. And Peters, E. (2000). Definition, effects and assessment of groundwater droughts, In: Drought and Drought Mitigation in Europe (ed. by J.V.Vogt and F.Somma), Kluwer Academic Publishers, the Netherlands,Pp 49-61 Landsat7 data users Science handbook, (2002). NASA, retrieved on 28th August 2012 from: www.gsfc.nasa.gov/IAS/handbook/handbook-toc.html Luke, J. Wersinger, J. Jianshen, S. Kopalle, D. (2001). Remote sensing applications from the Alabama Space Program Li, J. Meng,L. Chen, M. and Chen, D. (2009). Research on dynamic drought monitoring methods based on remote sensing and precipitation information: ISPRS Archives XXXVIII-8/w3 Workshop proceedings: Impacts of climate change on agriculture. pp 25-31, Retrieved from: www.earthobservations.org/document/cop/ag-gams/2009-12-17/ISRO_Final.pdf

!*" "

Markham, B. Dabney, P. Storey, J. Morfitt,R. Knight, E. Lee, G. (2008). Landsat Data Continuity Mission, Calibration and Validation. Pecora 17.The future of Land Imaging. Going Operational .November 18-20, 2008, Denver, Colorado. McGuffie, K. And Henderson-sellers, H. (2005): A climate Modeling Primer; John Wiley and Sons Ltd; ISBN 0470-85750-1 (HB); 0-470-85751-X (PB) Mokhtari, M. (2005). Agricultural drought impact assessment using remote sensing. A case study of Borkhar District, Iran. Unpublished MS Thesis, ITC Netherlands Morgan,J. (2009, Sept 28). Kenyas Heart Stop Pumping. BBC News.Retrieved on 22nd August 2012 from: www.news.bbc.co.uk/2/hi/africa/8055553.stm Mwenya, O. (2009). Flamingoes flee Lake Nakuru. Safarimate News. The Safarimate News, Retrieved on 3rd August 2012 on www.safariweb.com/safarimate/flamingo.html Mubea, K. Gachari, M. Mundia, C. (2009). Environmental assessment of Nakuru using remote sensing: A case study of Nakuru municipality. GIS Development. Retrieved from: www.gisdevelopment.net/application/environment/overview/NakuruMunicipality.html Mulanda,O and Mwangi, M. (2009, Aug 31). The Wild Animals not spared by drought. Retrieved on 20th August 2012 from: www.allafrica.com/stories/2009083/1710.html Murad, H. Islam, S. (2011). Drought assessment using remote sensing and GIS in North-West region of Bangladesh. 3rd International Conference on Water and Flood Management (ICWFM) Nguyen T, (2006). Coping with drought in central highlands Vietnam. PhD thesis submitted to Institute of Environment and Resources Technical University of Denmark, www.dtu.dk Nyangito, H. Ikiara, M. Ronge, E. (2002). Performance of Kenyas wheat industry and prospects for regional trade in wheat products. Productive sector division of Kenya Institute for Public Policy Research Analysis; KIPPRA. Discussion paper No.17, November 2002 Palmer, C. (1965). Meteorological Drought.US department of commerce, Research Paper No.45 Qinqin, S. Tan, J. Xu, Y. (2009). An ERDAS Image processing method for retrieving LST and describing urban evolution: A case study in the Pearl River delta region in South China. Environment Earth Science (2010) 59;1047-1055 Doi;101007/s1265-009-3 Roncoli, C. Okoba, B. Gathaara, V. Ngugi, J. Nganga , T. (2010). Adaptation to climate change for small holder agriculture in Kenya. Community based perspective from five districts. Retrieved on 27th July 2012 from: www.africa-adapt.net/media/rsources/410/Roncoli-report-FINAL.pdf Ron Hayes (2007). Landsat calibration and validation activities, status and issues. Landsat science team meeting, January 9th 2007.USGS Schuman, A. (2007). A multi-criteria approach in drought risk assessment. Lecture materials; Institute of Hydrology, Ruhr University, Bochum #+" "

Sinha, R. (2000). Role of drought early warning system for sustainable agricultural research in India. Journal of Climate. Pp.131-146 Sobrino, A. Jimenez -Munoz, J. Paolini, L. (2004). Land surface temperature retrieval from Landsat TM5. Remote sensing of environment 90 (2004)434-440;doi;10.1016/j.rse2004.02.003 South Asia Disaster Management Centre (SDMC), (2012). The drought. Document retrieved on July 29th 2012 from: www.saarc.smdc.nic.in/pdf/drought.pdf Sun, Y. Gottsche, F. Olesen, F.(2002). Retrieval of land surface temperature from combined AVHRR data. Annales Geophysicae, 20: 1257-1259. European Geophysical Society, 2002 Tallaksen, L. Madsen, H. and Clausen, B. (1997). On the definition and modeling of stream flow drought duration and deficit volume. Hydrological Science Journal, 42(1), Pp15-33 Thomas, J. Daoyi, C. Cosh, M. Li, F. Anderson, M. Walthall, C. Doriaswamy, P. Hunt, R. (2004). Vegetation water content mapping using LANDSAT data derived normalized difference water index for corn and soya beans. Remote Sensing of Environment (92), Pp 475-482 Thome, W. (2009, Dec 30). Flamingoes affected by drought at Lake Nakuru, eTN Global Travel Industry News. Retrieved on 17th August 2012 from: www.eturbonews/13563/flamingo-affectedbydrought-lake-Nakuru Tsegaye T. (1998).Improving drought management and planning through better monitoring in Africa. Drought Network News (1994 2001) paper 81: http://digitalcommons.unl.edu/droughtnetnews181 Tychon. B, Pecheur, C, and Ozer, P. (2007). The NDWI as a drought index applied to Belgium and Heilongjiang in Belgian and Chinese crop growth monitoring systems: Comparison, adaptation and Improvement, Tychon B. (ed), FUL, Arlon, Belgium. Pp111-120 United Nations Food and Agricultural Organization, (FAO,2011). Drought emergency; The Emergency Ministerial level meeting Paper, Rome 2011 United Nations Environmental program, (UNEP, 2000). A devastating drought, environmental impacts and response. Nairobi Kenya. Pp17 Valor, E. and Caselles, V. (1996). Mapping Land surface temperature emissivity from NDVI. Application to Europe, African and South America Areas. Remote Sensing of Environment (57), Pp167-184 Wegulo, F. Wandhwa, P. Shivanga, W. Tabu, I. Muhia, N. and Inoti, S. (2007). Engaging communities in soil fertility management for sustainable agricultural production: A case study of Kakamega and Nakuru Districts. Institute for Environment and Development (IIED) www.pubs.iied.org Walubengo, D, (2007). Community led action to use of forestry in building resilience to climate change: A Kenyan case study, Njoro division, Nakuru District. Institute for Environment and Development (IIED): www.pubs.iied.org/pdfs/G02310.pdf Weng, Q. Lu, D. Schubring , J. (2003). Estimation of land surface temperature vegetation abundance relationship for urban heat Island studies. Remote Sensing of Environment (89), Pp 467- 483 #!" "

Wilhite, D. and Glantz, M. (1985). Understanding the drought phenomenon: The role of definitions. Water International, 10(3); Pp111 -120 WMO-No. 989, (2005). Climate and land degradation. World Meteorological Organization, WMO.ISBN92-6310989-3: www.wmo.int/web/wcp/agm/agmp.html Xiao, X. He,L. Salas, W.Li, C. Moore, B.Zhao, R. Frolking, S. and Boles, S. (2002).Quantitative relationships between field measured leaf area index derived from vegetation images for paddy rice fields. International Journal of Remote Sensing, ISSN 0143-1161 print/ISSN 1366-5901Taylor and Francis ltd: http://www.tandf.co.uk/journals.doi:1080/01431160110115799

This work is licensed under the Creative Commons Attribution 3.0 License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/legalcode!

!
Proceedings of Global Geospatial Conference 2013 Addis Ababa, Ethiopia, 4-8 November 2013!

##" "

Appendices Moisture Levels, 2000 The Map and chart below represents the spatial outcome of the analysis.

Moisture levels, 2000

Figure Map of Nakuru showing Moisture Levels, 2000

Moisture Levels, 2010

Chart 4.1.4: Moisture levels, 2010

#$" "

Figure 4.1.3: Map of Nakuru showing Moisture Levels, 2010

Moisture Levels, 2011

Figure 4.1.6: Moisture levels, 2011

Figure 4.1.5: Map of Nakuru showing Moisture Levels, 2011 Temperature spatial variation in the year 2000

Figure 4.3.2: Graph of LST Proportion, 2000

#%" "

Figure 4.3.1: Map of Nakuru Showing LST variations, Temperature spatial variation in the year 2010

Figure 4.3.5: Graph of LST Proportion, 2010

Figure 4.3.4: Map of Nakuru Showing LST variations 2010 Temperature spatial variation in the year, 2011

#&" "

Figure 4.3.8: Graph of LST variation 2010

Figure 4.3.7: Map of Nakuru Showing LST variations 2010

#'" "