Available online at ScienceDirect. Procedia Engineering 78 (2014 )

Available online at www.sciencedirect.com ScienceDirect Procedia Engineering 78 (2014 ) 178 187 Humanitarian Technology: Science, Systems and Global Impact 2014, HumTech2014 Remote sensing tools for evaluating poverty alleviation projects: A case study in Tanzania Robert Morikawa* *Plant With Purpose, 4747 Morena Blvd, Suite 100, San Diego, CA, 92117-3466, USA Abstract Identifying reliable methods to evaluate rural poverty alleviation projects is an on-going challenge. Advances in spatial analysis and remote sensing, in particular NDVI, (Normalized Difference Vegetation Index), provide new opportunities for poverty alleviation projects and rural communities to understand the context and effect of project initiatives. This study demonstrates the use of NDVI as a practical, low-cost method for evaluating the impact of rural community action and the possible association with vegetation change. 2014 Elsevier Ltd. This is an open access article under the CC BY-NC-ND license 2014 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer-review under responsibility of the Organizing Committee of HumTech2014. Selection and peer-review under responsibility of the Organizing Committee of HumTech2014 Keywords: NDVI, vegetation monitoring, impact evaluation, community development, poverty alleviation, savings groups 1. Introduction Evaluating poverty alleviation projects in rural areas is an on-going challenge, and identifying appropriate data that can effectively assess project impact is difficult [1]. Advances in spatial analysis tools have made it possible to access a wide variety of remote sensing data that can function as evaluation indicators. These remote sensing data have the potential to answer questions for rural poverty alleviation projects at a finer resolution, both spatially and temporally than may have been possible in the past. One example is vegetation monitoring data such as the Normalized Difference Vegetation Index (NDVI). * Corresponding author. Tel.: 858-274-3718. E-mail address: robert@plantwithpurpose.org 1877-7058 2014 Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer-review under responsibility of the Organizing Committee of HumTech2014 doi:10.1016/j.proeng.2014.07.055

Robert Morikawa / Procedia Engineering 78 (2014 ) 178 187 179 NDVI relies on the way that light is reflected by plants, in particular the different ways that near-infrared light and visible red light are absorbed and reflected [2]. This property of photosynthesizing plants, expressed as a numerical NDVI value has been shown to be a reliable indicator of both vegetation dynamics and intensity over a wide range of conditions of both climate and ecology [3, 4]. NDVI has been developed as a tool to measure a wide variety of vegetation-related characteristics including crop type and cropping area [5], famine and drought early warning [6], land cover types [7], soil erosion [8], and hydrological variation [9]. Not only has NDVI been used to measure basic vegetation metrics, it has also been employed as a proxy to reliably estimate biodiversity indicators such as species richness [10, 11, 12, 13], avian abundance [14, 15], tree species richness [16, 17, 18, 19, 20] and vegetation species abundance [21]. Subsistence farmers, such as those involved in this current study, commonly employ dispersed tree management systems such as agroforestry. These systems result in landscapes of complex vegetative mosaics combining trees, crops, and grazing grassland, making vegetation analysis significantly more challenging than more homogeneous cropping systems. Nevertheless NDVI has proven to be useful in mapping land use of small scale, fragmented farm systems [22, 23]. NDVI combined with slope and rainfall data provided accurate estimates of erosion risk in Uganda and resulted in a low-cost method to identify priority areas for watershed management [24]. A study in the Iberian Peninsula showed that NDVI could accurately detect dispersed tree cover in savannah woodland [25], similar to the conditions often found in subsistence farm systems. Moderate and low resolution NDVI data may be less precise in dispersed tree systems and should be supplemented with higher resolution data [26]. While a lack of precision may be a risk in some contexts, sufficient precision has been demonstrated in some studies, particularly where ground truthing and modelling is combined with NDVI estimation. In these cases, accurate estimates of tree biomass [27], as well as crop yield have been derived [28, 29, 30, 31]. Detection of long-term trends in vegetation dynamics, the primary purpose of this study, is a commonly utilized aspect of NDVI. NDVI is widely seen as a reliable way to measure surface vegetation and dynamics [32] over a wide range of climates and ecotypes [3, 4]. NDVI was used to study long-term variation in vegetation in Nigeria [33]. High resolution SPOT NDVI data has been successfully used to study vegetation dynamics in Northern China [34] as well as for long-term vegetation monitoring in Kenya [35]. Cumulated NDVI has been used to measure variability of farm systems in sub-saharan Africa [36]. MODIS NDVI, the data-type used specifically in this study, has been useful in detecting long-term trends in a variety of environments including, grazing lands of Mongolia [37], and Etosha National Park in Namibia [38]. Research has indicated that while MODIS data is reliable for trend detection there is greater variability of results in more humid climates [39, 40]. The use of NDVI in fields such as forestry, ecology, and agriculture is well established, but its use as an evaluation method in community development or poverty alleviation is less well known. One of the earliest published studies showed that NDVI was a predictor of poverty incidence in Kenya and dates to 2005 [41]. A more recent study indicated that there was no simple statistical relationship between land degradation expressed as AVHRR-NDVI and poverty in China [42]. NDVI data was used as a tool to evaluate watershed development projects in India [43] and ecological restoration work in China [44]. Plant With Purpose, a sponsoring organization in this current study has been using NDVI since 2011 as a practical evaluation tool for measuring project impact [45]. Tracking vegetative cover in and around subsistence farm communities can increase understanding of the effect of agricultural activities on the landscape. Are communities clearing land for cropping? Are trees being cut for firewood or charcoal production? Are land management systems such as agroforestry having a positive effect on vegetative cover? Traditional methods in poverty alleviation studies rely on household surveys, or field data collection to measure the impact of subsistence farm communities on the environment, but remote sensing data such as NDVI provide the opportunity to answer these questions at a community scale and over multiple years. Being able to understand community action at a fine resolution both temporally and spatially is unprecedented and can result in a greater understanding of the dynamics of related landscape degradation and restoration. This improved understanding can help communities and poverty alleviation projects to develop better strategies for improving farming practices resulting in better community conditions and healthier landscapes.

180 Robert Morikawa / Procedia Engineering 78 ( 2014 ) 178 187 In this current study we demonstrate a practical and low-cost example of the use of NDVI by a rural poverty alleviation project to measure impact at the community level. 2. Materials & Methods 2.1. Purpose of Study The organization involved in this study, Plant With Purpose, and its local affiliate, Floresta-Tanzania, work to alleviate poverty in rural areas through an integrated approach. The organization works from the hypothesis that poverty and environmental degradation are intimately related and each contributes to the other. Subsistence farmers are impoverished in part by the degradation of their land, and at the same time are unable to invest in the restoration of their land due to the urgency of their immediate circumstances. This degradation tends to increase in response to environmental and economic shock, for example increased tree cutting in Haiti in response to the 2010 earthquake or increased charcoal making in Ethiopia in response to drought. Since reforestation and environmental activities are voluntary rather than directly compensated, they are only engaged in when people have at least a modest surplus. Thus an increase in vegetation as measured by NDVI can be a proxy indicator for improving economic conditions. Similarly, it is hypothesized that watersheds with increasing vegetation will be more fertile and productive. Thus an increase in NDVI is an indicator for the potential for future farm health and future prosperity. Plant With Purpose provides training and technical support to help communities establish savings groups based on the Village Savings and Loans Association model, or VSLA [46]. These savings groups, through additional technical support and training from Floresta-Tanzania actively adopt sustainable farming techniques such as agroforestry, and environmental conservation practices such as reforestation and watershed protection. Groups operate autonomously and receive no direct financial support from the training organization. The purpose of this study was to use remote sensing vegetation data, in this case NDVI as a low cost method to determine if the actions of local savings groups in rural areas supported by the training organization are having a measurable impact on local environmental conditions. 2.2. Area of Study Communities targeted in this study are located in the Pangani watershed in Tanzania (see map). The Pangani river basin is approximately 44000 square kilometres in area and includes Mount Meru, Mount Kilimanjaro, as well as the Pare and Usambara mountain ranges. Approximately 130 savings groups in the Pangani watershed were part of the study representing approximately 4300 families and a geographic coverage of about 600 square kilometres. Study area is primarily agricultural and typified by complex mosaics of trees and crops associated with multi-story agroforestry systems. Farm communities in the study area often find themselves in close proximity to ecologically and touristically important areas, and therefore de facto managers of land with competing demands for limited resources.

Robert Morikawa / Procedia Engineering 78 (2014 ) 178 187 181 2.3. Data The MODIS NDVI product, MOD13A3 [47] was used as the primary source of data for measuring vegetation change in this study. MOD13A3 provides 30 day NDVI and EVI products which are atmospherically corrected and masked for clouds, cloud shadows and other atmospheric contamination. MODIS NDVI has been shown to be a reliable estimator of vegetation dynamics and intensity over a wide range of climate and eco-types [3, 4]. Only data from MODIS was used for this study, since data across sensor systems is not easily compared directly [48, 32]. MODIS NDVI data was acquired for the period 2003 to 2013 representing one year before the inception of the project under study to present. Nation-wide community data was sourced from the Tanzanian Government National Bureau of Statistics (NBS), based on national Census data [49]. Savings group locations and data were based on GPS waypoints and field reports collected by Floresta Tanzania. 2.4. Data analysis Raw NDVI 30 day average raster files were imported into ESRI ArcGIS Desktop 10.2. Data from three specific seasonal points, mid wet season (April), mid dry season (September) and transition between wet and dry seasons (June) were aggregated to generate NDVI annual mean values using ArcGIS-Cell Statistics. Annually aggregated NDVI values have been widely used both to classify fragmented farm systems such as those found in the study area and to reliably detect long-term change in vegetation [50, 23]. Data were analyzed for the period 2003 to 2013, the period for which Floresta-Tanzania worked with savings groups in the target area. Annual NDVI values were expressed as three year rolling means in order to smooth the variability caused by short-term vegetation changes. This yielded NDVI annual means for the period 2004 to 2012. Each of the 133 savings groups as well as all official villages recognized by the Tanzanian National Bureau of Statistics, located in the Pangani watershed [49] were assigned an annual mean NDVI value for all available MODIS data for the period 2004 to 2013 using ArcGIS Extract Values. NDVI means were tested for long-term trends by four groupings shown in Table 1. Table 1 Grouping Savings-all Savings-Marangu Communities-all Communities-Marangu Description All 133 savings groups working with Floresta Tanzania in Pangani watershed A subset of 67 savings groups from Savings-all located in the Marangu area of Moshi Rural All 1432 official communities in the Pangani watershed according to official census data A subset of 115 official villages from Communities-all located in the Marangu area of Moshi Rural Groupings were selected to allow reasonable comparison between groups. That is, the Savings-all and Communities-all groupings represented similar ecological and demographic conditions for comparing vegetation change, as these points were distributed across the Pangani Watershed. The Savings-Marangu and the Communities-Marangu groupings also provided a comparison under similar conditions as the points in both these groupings were found in sub-watersheds of the Pangani located in the Marangu area at the foot of Mount Kilimanjaro. Trends were tested for significance using the Mann-Kendall Trend test [51], suitable for testing a trend over time, and where the data cannot be assumed to be normally distributed. Groupings were further analyzed for average NDVI value with respect to distance from the community. ArcGIS Focal Statistics was used to estimate mean NDVI and NDVI change values zero, one, two, five, and ten km distance

182 Robert Morikawa / Procedia Engineering 78 ( 2014 ) 178 187 from the community grouping in question (see table 1). Distance values were tested for significance using the Mann-Whitney U Test for nonparametric data. 3. Results NDVI annual values showed a general downward trend for all four groupings, that is all savings groups combined (Savings-all), savings groups from Marangu only (Savings-Marangu), all communities in the Marangu area (Marangu-all), and all communities in the Pangani watershed (Pangani-all) as shown in figure 3. This indicates that over the ten year period vegetation likely decreased in all four groupings within the study area. The negative trend in NDVI was less severe among savings groups than communities in the Pangani watershed in general, but none of these trends were statistically significant. However, from 2009 to 2012 (see figure 4), the trend among savings groups for NDVI is positive, and this pattern is statistically significant at the 10% level. NDVI trend for other communities during the same period was neutral or negative. In particular, savings groups in the Marangu area (Savings-Marangu) showed a steady increase in the mean annual value of NDVI compared to other communities in the same subwatershed area (Communities-Marangu). This suggests that in spite of being located in similar ecological and demographic conditions, locations where savings groups are present are experiencing a measurable increase in vegetation over the period in question. While correlation does not equal causation, it is known that the savings groups in question independently plant thousands of trees per year, utilize energy saving wood stoves, and practice a variety of sustainable farming practices. Data from internal evaluations by the organization involved in this study show that farmers who participate in savings groups are planting on average twice as many trees annually as farmers who do not participate in savings groups.

Robert Morikawa / Procedia Engineering 78 (2014 ) 178 187 183 While there is no data at this time to directly link savings group tree planting activity and change in NDVI trend, figure 5 shows that growth of savings groups does coincide with vegetation change. NDVI means for 2012 as seen in Figure 6, tended to decrease with distance from any community grouping, however, savings groups showed consistently higher values of NDVI at least five km away from the community centre point. These values were tested for statistical significance at the centre point and showed that savings groups were different from communities in the Marangu area, as well as the greater Pangani watershed region. Change in NDVI over the period 2009 to 2012 was also tested for a distance effect (Figure 7). Savings groups showed a positive change in NDVI for 2009-2012 at least one km away from the centre point, while average communities in Marangu and the Pangani watershed showed neutral or negative change at the centre point and all other distances. The difference between savings groups for this distance effect and other communities was statistically significant at the 5% level or higher for both the centre point, and the one km distances.

184 Robert Morikawa / Procedia Engineering 78 ( 2014 ) 178 187 4. Discussion Trends in the data show a positive correlation between vegetative cover as expressed by NDVI mean annual values, and the presence of savings groups. While correlation does not equal causation, each of these savings groups independently plant thousands of trees per year, as well as engage in other environmentally beneficial activities such as the use of energy saving wood stoves, and agroforestry. Although the overall trend since the beginning of the project period, 2004, shows a negative trend for all grouping types, savings or not, the decrease in vegetative cover is not as severe where savings groups are present, and since 2009, there has been an increase in vegetative cover where savings groups are present. Other, non-savings group points show neutral or negative change over the same 2009 to 2012 period. Furthermore, this positive effect is observed up to at least 1 km away from where a savings group is present. While other NDVI studies have been able to accurately model tree biomass or crop yield [27, 28, 29, 30, 31], this requires field plots and ground truthing to develop reliable models. This current study can only interpret the direction of vegetative change--positive or negative--and the relative quantity of change--greater or less. Further study will be required to more precisely quantify levels of change. Nevertheless, the kind of information generated by NDVI

Robert Morikawa / Procedia Engineering 78 (2014 ) 178 187 185 analysis gives a poverty alleviation organization and farm communities unprecedented insight into their ecological context and allows more informed project evaluation and planning. 5. Conclusion Many factors other than the presence of an active tree planting community (savings group) may explain changes in vegetation and there is a risk of interpreting NDVI change as directly attributable to community action. However, selecting comparison groups in the same watershed, with the same ecotype, and similar populations, reduces the possibility of variance caused by other factors. More sophisticated sampling and modelling can and perhaps should be done to distinguish one effect from another. Promising new developments may help rural poverty alleviation projects develop new sources of data for understanding project and community action. For example, LIDAR and UAV (Unmanned Aerial Vehicles) allow high resolution data collection [52]. High resolution poverty maps are being developed as better modelling techniques are developed and as more spatially referenced data becomes available [53, 54, 55, 56]. Thus while NDVI provides an opportunity for better understanding of poverty and the effects of poverty alleviation efforts in rural areas, rapid changes in technology are creating new opportunities for analysis and understanding. This study demonstrates the use of NDVI as a practical low cost method to assess community action in rural areas. Results show a strong association between the presence of savings groups and a positive change in vegetation over from 2009-2012. This information, while not proof of cause, better informs community groups and rural poverty alleviation projects and allows improved project evaluation and planning. Acknowledgements To the dedicated farmers and staff of Floresta-Tanzania for their sacrificial efforts to better the economic and environmental conditions of northern Tanzania. To the staff and supporters of Plant With Purpose for making this study possible. Thanks to ESRI Non-Profit program and the Conservation GIS program for their generous support in providing ArcGIS licenses. References [1] Perrin B. (2012). Impact Evaluation Notes: Linking monitoring and evaluation to impact evaluation. No. 2. April 2012. [2] Weier J, Herring D. (2000). Measuring Vegetation (NDVI & EVI). NASA Earth Observatory. http://earthobservatory.nasa.gov/features/measuringvegetation/ [3] Huete A, Didan K, Miura T, Rodriguez EP, Gao X, Ferreira LG. (2002). Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sensing of Environment, 83(2-Jan), 195-213. [4] Fensholt R, Sandholt I. (2005). Evaluation of MODIS and NOAA AVHRR vegetation indices with in situ measurements in a semi-arid environment. International Journal of Remote Sensing; 2005. 26(12):2561-2594 [5] Vintrou E, Desbrosse A, Begue A, Traore S, Baron C, LoSeen D. (2012b).Crop area mapping in West Africa using landscape stratification of MODIS time series and comparison with existing global land products. International Journal of Applied Earth Observation and Geoinformation; 2012. 14(1):83-93. [6] Funk CC, Brown M E. (2006). Intra-seasonal NDVI change projections in semi-arid Africa. Remote Sensing of Environment; 2006. 101(2):249-256 [7] Combalicer MS, DongYeob K, Combalicer LD, Cruz EA, Im SangJun RVO. (2011). Changes in the forest landscape of Mt. Makiling Forest Reserve, Philippines. Forest Science and Technology; 2011. 7(2):60-67 [8] Butt MJ. Waqas A, Mahmood R. (2010). The combined effect of vegetation and soil erosion in the water resource management. Water Resources Management; 2010. 24(13):3701-3714 [9] Poveda G, Jaramillo A, Gil MM, Quiceno N, Mantilla RI. (2001). Seasonality in ENSO-related precipitation, river discharges, soil moisture, and vegetation index in Colombia. Water Resources Research; 2001. 37(8):2169-2178 [10] Gaitan JJ, Bran D, Oliva G, Ciari G, Nakamatsu V, Salomone J, Ferrante D, Buono G, Massara V, Humano G, Celdran D, Opazo W, Maestre FT. (2013). Evaluating the performance of multiple remote sensing indices to predict the spatial variability of ecosystem structure and functioning in Patagonian steppes. Ecological Indicators; 2013. 34:181-191 [11] Pau S, Gillespie TW, Wolkovich EM. (2012). Dissecting NDVI-species richness relationships in Hawaiian dry forests. Journal of Biogeography; 2012. 39(9):1678-1686 [12] Psomas A, Kneubuhler M, Huber S, Itten K, Zimmermann NE. (2011). Hyperspectral remote sensing for estimating aboveground biomass and for exploring species richness patterns of grassland habitats. International Journal of Remote Sensing; 2011. 32(24):9007-9031

186 Robert Morikawa / Procedia Engineering 78 ( 2014 ) 178 187 [13] Parviainen M, Luoto M, Heikkinen RK. (2009). The role of local and landscape level measures of greenness in modelling boreal plant species richness. Ecological Modelling; 2009. 220(20):2690-2701 [14] Mcfarland T M, Riper,. III van Johnson GE. (2012). Evaluation of NDVI to assess avian abundance and richness along the upper San Pedro River. Journal of Arid Environments; 2012. 77:45-53 [15] Oindo BO. (2008). Understanding the relationship between environmental energy availability and bird species richness in Kenya using remote sensing and ancillary data.open Remote Sensing Journal; 2008. 1:1-6 [16] Hernandez-Stefanoni JL, Gallardo-Cruz JA, Meave JA, Dupuy JM. (2011). Combining geostatistical models and remotely sensed data to improve tropical tree richness mapping.ecological Indicators; 2011. 11(5):1046-1056 [17] Mohammadi J, Shataee S. (2010). Possibility investigation of tree diversity mapping using Landsat ETM+ data in the Hyrcanian forests of Iran. Remote Sensing of Environment; 2010. 114(7):1504-1512 [18] Gillespie TW, Saatchi S, Pau S, Bohlman S, Giorgi AP, Lewis S. (2009). Towards quantifying tropical tree species richness in tropical forests. International Journal of Remote Sensing; 2009. 30(5/6):1629-1634 [19] Levin N, Shmida A, Levanoni O, Tamari H, Kark S. 2007. Predicting mountain plant richness and rarity from space using satellite-derived vegetation indices. Diversity and Distributions; 2007. 13(6):692-703 [20] Bawa K, Rose J, Ganeshaiah KN, Barve N, Kiran MC.(2002). Umashaanker, R.Assessing biodiversity from space: an example from the Western Ghats, India. Conservation Ecology; 2002. 6(2):7 [21] Virtanen R. Luoto M. Rama T. Mikkola K. Hjort J. Grytnes JA. Birks HJB. (2010). Recent vegetation changes at the high-latitude tree line ecotone are controlled by eomorphological disturbance, productivity and diversity. Global Ecology and Biogeography; 2010. 19(6):810-821 [22] Bie CA, de Khan JM, Smakhtin MR, Venus VU, Weir V, Smaling MJC. (2011). Analysis of multi-temporal SPOT NDVI images for smallscale land-use mapping, International Journal of Remote Sensing; 2011. 32(21):6673-6693. [23] Vintrou E, Soumare M, Bernard S, Begue A, Baron C, Lo Seen D. (2012a). Mapping Fragmented Agricultural Systems in the Sudano- Sahelian Environments of Africa Using Random Forest and Ensemble Metrics of Coarse Resolution MODIS Imagery. Photogrammetric Engineering and Remote Sensing, 78(8), 839-848. [24] Mutekanga FP, Visser SM, Stroosnijder L. (2010). A tool for rapid assessment of erosion risk to support decision-making and policy development at the Ngenge watershed in Uganda.Geoderma; 2010. 160(2):165-174. [25] Carreiras JMB, Pereira JMC, Pereira JS. (2006). Estimation of tree canopy cover in evergreen oak woodlands using remote sensing. Forest Ecology and Management; 2006. 223(1/3):45-53 [26] Munyati C, Mboweni G. (2013). Variation in NDVI values with change in spatial resolution for semi-arid savanna vegetation: a case study in northwestern South Africa. International Journal of Remote Sensing; 2013. 34(7):2253-2267 [27] Migliavacca M, Meroni M, Busetto L, Colombo R, Zenone T, Matteucci G, Manca G, Seufert G. (2009). Modeling gross primary production of agro-forestry ecosystems by assimilation of satellite-derived information in a process-based model. Sensors; 2009. 9(2):922-942 [28] Basso B, Cammarano D, Carfagna E. (2013). Review of Crop Yield Forecasting Methods and Early Warning Systems. FAO archives. http://www.fao.org/fileadmin/templates/ess/documents/meetings_and_workshops/gs_sac_2013/improving_methods_for_crops_estimates/ Crop_Yield_Forecasting_Methods_and_Early_Warning_Systems_Lit_review.pdf [29] Fermont A, Benson T. (2011). Estimating Yield of Food Crops Grown bysmallholder Farmers A Review in the Uganda Context. IFPRI Discussion Paper 01097. http://www.ifpri.org/sites/default/files/publications/ifpridp01097.pdf [30] Matinfar, HR. (2013). Modeling wheat yield estimation base upon spectral data and field measurement, case study: Razan plain, Iran. Technical Journal of Engineering and Applied Sciences; 2013. 3(17):2123-2130. 16 [31] Zand F, Matinfar HR. (2012). Winter wheat yield estimation base upon spectral data and ground measurement. Annals of Biological Research, 2012, 3 (11):5169-5177(http://scholarsresearchlibrary.com/archive.html) [32] Tarnavsky E, Garrigues S, Brown ME. (2008). Multiscale geostatistical analysis of AVHRR, SPOT-VGT, and MODIS global NDVI products. Remote Sensing of Environment; 2008. 112(2):535-549. http://scholar.google.ca/scholar_url?hl=en&q=http://www.researchgate.net/publication/222888866_multiscale_geostatistical_analysis_of_a VHRR_SPOT- VGT_and_MODIS_global_NDVI_products/file/9c96051a8d3fb8c11a.pdf&sa=X&scisig=AAGBfm3IifHYQ3Xhb62wTWUgS1KQxNDRN w&oi=scholarr&ei=s0eouqvcjc6bqaghg4cida&ved=0ccwqgamoadaa [33] Aweda ED, Adeyewa ZD. (2011). Inter annual variation of vegetation anomaly over Nigeria using satellite-derived index. Advances in Applied Science Research; 2011. 2(3):468-475. http://pelagiaresearchlibrary.com/advances-in-applied-science/vol2-iss3/aasr-2011-2-3-468-475.pdf [34] ZhanDong S, NiBin C, Opp C. (2010). Using SPOT-VGT NDVI as a successive ecological indicator for understanding the environmental implications in the Tarim River Basin, China.Journal of Applied Remote Sensing; 2010. 4:043554 [35] Kinyanjui MJ. (2011). NDVI-based vegetation monitoring in Mau forest complex, Kenya. African Journal of Ecology; 2011. 49(2):165-174 [36] Vrieling A, de Beurs KM, Brown ME. (2011). Variability of African farming systems from phenological analysis of NDVI time series. Climatic Change. December 2011, Volume 109, Issue 3-4, pp 455-477 [37] Yin H. Udelhoven T. Fensholt R. Pflugmacher D. Hostert P. (2012). How normalized difference vegetation index (NDVI) trends from advanced very high resolution radiometer (AVHRR) and Systeme Probatoire d'observation de la Terre VEGETATION (SPOT VGT) time series differ in agricultural areas: an Inner Mongolian case study. Remote Sensing; 2012. 4(11):3364-3389 [38] Wagenseil H, Samimi C. (2005). Use of multitemporal satellite data for vegetation change detection in Namibia. New strategies for European remote sensing. Proceedings of the 24th Symposium of the European Association of Remote Sensing Laboratories, Dubrovnik, Croatia, 25-27 May 2004; 2005. :183-190

Robert Morikawa / Procedia Engineering 78 (2014 ) 178 187 187 [39] Hmimina G, Dufrene E, Pontailler JY, Delpierre N, Aubinet M, Caquet B, Grandcourt A, de Burban B, Flechard C, Granier A, Gross P, Heinesch B, Longdoz B, Moureaux C, Ourcival JM, Rambal S, Saint Andre L, Soudani K. (2013). Evaluation of the potential of MODIS satellite data to predict vegetation phenology in different biomes: an investigation using ground-based NDVI measurements. Remote Sensing of Environment; 2013. 132:145-158 [40] Fensholt R. Rasmussen K, Nielsen TT, Mbow C. (2009). Evaluation of earth observation based long-term vegetation trends Intercomparing NDVI time series trend analysis consistency of Sahel from AVHRR GIMMS, Terra MODIS and SPOT VGT data. Remote Sens. Environ. 2009, 113,1886 1898 [41] Kristjanson P, Radeny M, Baltenweck I, Ogutu J, Notenbaert A. (2005). Livelihood mapping and poverty correlates at a meso-level in Kenya. Food Policy 30 (2005) 568 583 [42] ZhanGuo B, Dent D. (2009). Recent land degradation and improvement in China.Ambio; 2009. 38(3):150-156 [43] Venkateswarlu B, Rao KV, Kaushalya Ramachandran Mandal UK. (2010). Application of geomatics in watershed prioritization, monitoring and evaluation - CRIDA's experience. Use of high science tools in integrated watershed management. Proceedings of the National Symposium, NASC Complex, New Delhi, India, 1-2 February 2010; 2011. :49-65 [44] Zhang GD, Xiao JW, Hu ZhongMin XM, Sheldon, S. (2012). Effectiveness of ecological restoration projects in Horqin Sandy Land, China based on SPOT-VGT NDVI data. Ecological Engineering; 2012. 38(1):20-29 [45] Kaiser ET, Roberts K, Pawkham J, Thongdi B, Morikawa R, Satre D. (2012). Healthy Forest, Healthy People: Remote Sensing and Monitoring Support Implementation of Community Forestry in Thailand. Global Humanitarian Technology Conference (GHTC), 2012 IEEE [46] Allen, H. (2006). Village Savings and Loans Associations sustainable and cost-effective rural finance. Small Enterprise Development, Volume 17, Number 1, March 2006: 61-68(8) [47] NASA. (2013). Land Processes Distributed Active Archive Center (LP DAAC). MOD13A3. USGS/Earth Resources Observation and Science (EROS) Center, Sioux Falls, South Dakota. [48] Gonsamo A., Chen JM. (2013). Spectral response function comparability among 21 satellite sensors for vegetation monitoring. IEEE Transactions on Geoscience and Remote Sensing; 2013. 51(3 Part 1):1319-1335 [49] NBS. (2002). Tanzania Village 2002. http://www.nbs.go.tz [50] Forkel M, Carvalhais N, Verbesselt J, Mahecha MD, Neigh CSR, Reichstein M. (2013). Trend Change Detection in NDVI Time Series: Effects of Inter-Annual Variability and Methodology. Remote Sens. 2013, 5, 2113-2144. [51] Mann HB. 1945. Non-parametric tests against trend. Econometrica. Vol. 13, No. 3 (Jul., 1945), pp. 245-259 [52] Dufour S, Bernez I, Betbeder J, Corgne S, Hubert-Moy L, Nabucet J, Rapinel S, Sawtschuk J, Trolle C. (2013). Monitoring restored riparian vegetation: how can recent developments in remote sensing sciences help? Knowledge and Management of Aquatic Ecosystems; 2013. (410):10 [53] Tatem A, Gething P, Pezzulo C, Weiss D, Bhatt S. (2013). Development of Pilot High-Resolution Gridded Poverty Surfaces: Methods working paper. WorldPop.org.uk. http://www.worldpop.org.uk/resources/docs/worldpop-poverty-mapping-methods.pdf [54] Grolle J. (2012). High-resolution mapping of rural poverty and famine vulnerability in the Sahel: a possible approach for the Republic of Niger. Population and Environment (2012): 1-30 [55] Szonyi J. Pauw E, de Rovere, R, la Aw-Hassan, A. (2010). Mapping natural resource-based poverty, with an application to rural Syria. Food Policy; 2010. 35(1):41-50 [56] Bellon MR, Hodson D, Bergvinson D, Beck D, Martinez-Romero E, Montoya Y.(2005). Targeting agricultural research to benefit poor farmers: relating poverty mapping to maize environments in Mexico. (Special issue: Poverty and food security mapping) Food Policy; 2005, 30(5/6):476-492