Access the full text.
Sign up today, get DeepDyve free for 14 days.
References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.
GEOLOGY, ECOLOGY, AND LANDSCAPES INWASCON https://doi.org/10.1080/24749508.2022.2144855 RESEARCH ARTICLE Comparative evaluation of NDVI strata and livelihood zones as spatial units of the practiced crop calendars Edossa Fikiru Wayima Plant Science Department, Madda Walabu University, Bale-Robe, Ethiopia ABSTRACT ARTICLE HISTORY Received 11 May 2019 Properly mapping the spatial variability of the practiced crop calendars is useful for precise Accepted 3 November 2022 planning of agricultural activities. A seasonal calendar based on the livelihood zone (LHZ) map is widely employed as a crop calendar in Ethiopia. However, using the LHZ-based crop calendar KEYWORDS for agricultural purposes may cause problems of doubtful accuracy, since non-ecological Agro-ecology; cropping parameters are included in the LHZ map. This research was conducted in the central and schedule; farmers’ crop southeastern parts of Ethiopia to compare the hypertemporal normalized difference vegeta- calendar; seasonal calendar; tion index (NDVI) and LHZ-based stratification methods in explaining the spatial variability of spatial variability; spatial precision the farmers’ actual crop calendar. A NDVI strata generated from SPOT-VGT and PROBA-V NDVI data of 16 years and the 2008 LHZ map of Ethiopia were processed to select sampling sites that represented three NDVI strata and four LHZs in common. Interview data were collected randomly from individual farmers on ploughing, planting and expected harvesting dates of teff (Eragrostis tef), wheat and barley (fields), and were analysed using non-parametric tests. The results revealed that the NDVI strata had greater spatial quality and precision than the LHZs in terms of uniqueness and internal homogeneity of the spatial units, indicating its suitability as a preferable alternative crop calendar. Introduction the United Nations [FAO], 2010) and for estimating A crop calendar is defined differently in various litera- supply, demand, and price fluctuations (MAFFIS, ture. According to Kotsuki and Tanaka (2015), a crop 2012). calendar is the period during which farmers plant and A crop calendar can be estimated using a variety of harvest crops on their cropland. Others describe it techniques, but all are associated with some specific mer- broadly as a schedule of all cropping activities starting its and demerits. For instance, the census-based from the fallow period (International Rice Research approach, which depends on the agricultural census Institute [IRRI], 2018), through land preparation and data collected by national and/or international institu- planting up to harvesting (De Bie, 2000; International tions (Fritz et al., 2019; Kotsuki & Tanaka, 2015), has high Rice Research Institute [IRRI], 2018; Van Heemst, reliability in areas with adequate census data (Portmann 1986), storage, and marketing (International Rice et al., 2010). However, this method is time-consuming Research Institute [IRRI], 2018; Van Heemst, 1986), (Fritz et al., 2019), labour-intensive and costly because it including all the required crop management opera- involves complete enumeration of the population. tions (Van Heemst, 1986). In the modelling approach, a crop calendar is esti- A crop calendar is an essential tool for all the parties mated based on crop growth models, which predict involved in the agricultural sector (Guo, 2013). It is crop growth by simulation using agricultural and meteor- useful for proper timing of planting and other agro- ological data such as soil moisture, solar radiation, and nomic practices (FAO, 2014b, 2019; Rea & Ashley, temperature (Kotsuki & Tanaka, 2015). This method can 1976), for providing a sufficient amount of input at accurately predict crop growth when parameters are well- the right time and place (Guo, 2013), for crop mon- calibrated though this is difficult in areas with insufficient itoring, for accurately estimating agricultural water census data. Furthermore, this approach has difficulty in demand, and for food production or yield forecasting identifying the planting dates (Portmann et al., 2010) if (FAO, 2014a; Kotsuki & Tanaka, 2015). Thus, it sig- the cultivation period is significantly influenced by factors nificantly contributes to reduced production costs other than those considered in the model. (International Rice Research Institute [IRRI], 2018) The satellite-based method estimates a crop calen- and optimum yield. A crop calendar can also be dar by using hypertemporal remote sensing data. applied in seed relief and rehabilitation programs Time-series data of vegetation indices show the spatial (FAO, 2019; Food and Agriculture Organization of and temporal dynamics of vegetation on the land CONTACT Edossa Fikiru Wayima firstname.lastname@example.org Plant Science Department, Madda Walabu University, Bale-Robe, Ethiopia © 2022 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of the International Water, Air & Soil Conservation Society(INWASCON). This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 2 E. F. WAYIMA surface (Fatiha et al., 2013), and are thus significantly Ethiopia and the USAID was used for this study. The useful for estimating crop calendars (Kotsuki & shape file of the LHZ map was downloaded from the Tanaka, 2015; Zheng et al., 2016). This approach has web page of the FEWSNET (2009). On the other hand, several advantages, including the use of near-real-time a total of 576 SPOT-VGT and PROBA-V 10-daily max- data with consistent spatial and temporal coverage imum value composite NDVI products at 1 km resolution (Khorram et al., 2012), and the ability to work at an spanning from January 1999 to December 2014 (http:// agro-ecology level (Mumby et al., 1999). However, the land.copernicus.eu/global/) were used for this study. The inability to obtain data through the cloud (National image dataset was classified using the ISODATA (ERDAS Research Council, 2001) and the need for a highly Inc., Norcross, GA, USA) in a series of unsupervised skilled workforce for data processing and analysis are classification runs with a pre-specified number of classes among the drawbacks of the remote sensing-based (10 to 100), and a run with 60 classes was chosen for technique (Umali & Exconde, 2003). further study. Seasonal calendars are extensively used as reference crop calendars for cropping operations in Ethiopia. The study area A seasonal calendar is a chart that depicts the distribution of social and natural events that vary seasonally across Two neighbouring LHZs with two dominant NDVI months of the year (Sontheimer et al., 1999). It is esti- classes in common were selected from the central and mated by locally knowledgeable people through rapid southeastern parts of Ethiopia as a study area (Figure 1). rural appraisal or participatory rural appraisal techniques The area selected from central Ethiopia consisted of two (Sontheimer et al., 1999). As a result, a seasonal calendar LHZs named “Ambo-Selale-Gindeberet teff and wheat can be prepared rapidly at low cost (Sontheimer et al., (AMT)” and “Selale-Ambo highland barley, wheat and 1999) and tends to have high accuracy (Heaver, 1992). horse bean belt (SAW).” The area chosen from the south- The seasonal calendars are prepared using the livelihood eastern part of the country is composed of three polygons zones (LHZs) as spatial units (MoARD, 2010b). The classified into two LHZs named “Arsi-Bale wheat, barley LHZs are the result of the combined effect of geographic and potato (ABW)” and “Robe, Chole, Sude and Seru teff , factors such as altitude, rainfall and population density as maize and haricot bean belt (RCS).” well as market forces (MoARD, 2010a). The LHZ is distinguished at the level of regional states of Ethiopia, Description of the LHZs and kebele (or neighbourhood) is its smallest adminis- The AMT LHZ, with two rainy seasons named arfasa/belg trative unit (MoARD, 2010a). Therefore, the seasonal (short rainy season extending from March to April) and calendar based on LHZ is confounded by non- ganna/kiremt/meher (main rainy season that lasts from ecological factors such as population density, market June to September), is known for the production of teff force and administrative boundaries at regional state and wheat as major crops. The SAW LHZ, where barley, and kebele levels. In addition, the seasonal calendar is wheat, oats and horse beans are the dominant crops, also generalized to a monthly basis and tends to exclude the has two rainy seasons with similar naming and durations inter-annual variability as local analysts are unlikely to to the AMT LHZ. The ABW LHZ, characterized by bimo- remember the long-term trend of seasonal variation. dal rainfall distribution distinguished as ganna and bona NDVI data can also be used to estimate crop phenolo- seasons, is known for the cultivation of wheat, barley, gical attributes such as ploughing, planting, and harvesting pulses, rapeseed and flax. The RCS LHZ, where maize, dates because these phenological stages are associated with sorghum, haricot beans, teff and khat (Catha edulis) are the specific NDVI values. Thus, an NDVI data-based stratifi - major crops, also has two seasons: arfasa (March–May/ cation approach could serve as a viable alternative to the June) and ganna (July–November), with the latter being LHZ-based technique for estimating the crop calendars the major (MoARD, 2010b). practiced by farmers. A comparative assessment of these two mapping methods is necessary to identify a preferable Description of the NDVI strata method for preparing crop calendars with better spatial The long-term average NDVI profiles of the NDVI accuracy. Thus, the objective of this research was to com- strata studied are presented in Figure 2. The NDVI pare NDVI map-based strata with the LHZ map-based profiles of the three NDVI strata are typically charac- zonation in terms of accurately representing the spatial terized by a double peak, indicating the presence of variability of the practiced crop calendars. two cropping seasons. The minor and the major peaks represent the short and main cropping seasons of the area, as described in (MoARD, 2010b). Stratum 23 is Materials and methods characterized by low level of greenness as well as early start and end of the two seasons because the peaks of Data description the seasons occur earlier in stratum 23 than in the The LHZ map published in 2010 jointly by the Ministry of others. Although the beginning of the shorter season is Agriculture and Rural Development (MoARD) of about the same, the major season starts and ends GEOLOGY, ECOLOGY, AND LANDSCAPES 3 Figure 1. Map of the study area. Figure 2. Average long term (1999–2014) NDVI profile of the NDVI classes studied. earlier in stratum 29 than in 43. Late seasons, a high Holeta) were used to access the sampling sites, called survey amount of greenness and a relatively long duration of strata, for field data collection. A stratified random sampling the short season are all characteristics of stratum 43. strategy was employed for selecting the sampling sites. Sampling strategy Field data collection Seven towns that were found in the selected units of the two Farmers from the specific sampling strata were ran- maps (Bekoji, Gasera, Gorfo, Muger, Chole, Chancho, and domly selected and interviewed about the ploughing, 4 E. F. WAYIMA planting and expected harvesting dates of teff , wheat technique were significantly different from one and barley for the 2015 major cropping season (known another. The Mann-Whitney U test was used to eval- as ganna in the AMT and SAW LHZs and bona in the uate if two spatial units significantly varied from one ABW and RCS LHZs or keremt. Interview data were another. The median absolute deviation (MAD) and collected as absolute dates or as dekadal data, desig- interquartile range (IQR) were used to compare the nated as early (1–10), mid (11–20) and end (21–30) of homogeneity of the spatial units of NDVI and LHZ in a month in the Ethiopian Calendar, during the 28th of order to identify the stratification technique that pro- September to the 26th of October 2015. Converting duced more homogeneous spatial units. The statistical the unit of the field data to early, middle and end of summary and distribution of the data on first plough- a month was necessary to change the absolute dates to ing, planting and expected harvesting dates were the same unit (i.e., decad) with the other field data, and visualized in a nested box plot using RStudio to align the length of duration of the field data with the (RStudio Team, 2015). NDVI data. Results Data analysis Description of the field data The GPS coordinate data were converted from DMS (degrees, minutes and seconds) to DD (decimal The distribution of the sampling points is shown degree) and the sample points were mapped with Figure 3 and Figure 4. The practiced crop calendar ArcGIS software (ESRI, 2016). The data, collected in data were collected through interview from 53, 89 and the Ethiopian Calendar (E.C.), were converted to the 64 fields for teff , wheat and barley, respectively. Gregorian Calendar and then to Julian date, such However, some crops were grown at a few sampling that day 1 represented 1 January 2015. Furthermore, sites during the major cropping season of the dekadal data was converted to absolute dates by repre- survey year. For instance, wheat was not planted at senting each dekad by its fifth day. For analysis, the one survey stratum (RCS 29), while barley was not data were arranged following the classifications of the cultivated at two sampling sites (AMT 29 and RCS 29). two mapping methods under study, NDVI strata and Furthermore, teff was not planted in half of the total LHZs. locations sampled (ABW 29, ABW 43, RCS 43 and The data were tested for normality and homogene- SAW 23). ity of variance with SPSS software Version 23 (IBM, Among a total of 222 crop fields for which data 2016) using the Shapiro-Wilk normality test (Yap & were collected, 203 fields that were not fallowed in Sim, 2011) and Levene’s test, respectively. The 2014 (2006/07 E.C.) were used for final analysis. The Kruskal-Wallis one-way ANOVA was computed to data from 19 crop fields that were fallowed in 2014 assess whether the spatial units of each mapping were not considered for analysis because the Figure 3. Distribution of sample points from the survey strata in North and West Shewa Zones. GEOLOGY, ECOLOGY, AND LANDSCAPES 5 Figure 4. Distribution of sample points from the survey strata in Arsi, West Arsi and Bale Zones. ploughing dates of such fields differed from those that general linear model analysis (Nancy & Anthony, were not fallowed. 2002). Thus, nonparametric tests were employed. The data collected were in different units of date, which necessitated their conversion to the same unit. About two-thirds of the data were in dekad, while the Overall comparison between the spatial units of remaining were in absolute date. In terms of the vari- each mapping method ables studied, 90% and 67% of the data on first plough- Establishing whether the spatial units of the two map- ing and expected harvesting dates respectively were in ping methods were statistically different from one dekad, while 75% of the planting date data were in another was of paramount importance in determining absolute date. Thus, the data collected in absolute the variability of the crop calendars practiced in sepa- dates were converted to decadal data by categorizing rate units of each classification method. In this regard, days 1–10 as early, 11–20 as middle, and 20-end of the Kruskal-Wallis one-way ANOVA revealed highly month as late, so that the units of the field data aligned significant (p < 0.01) differences between all NDVI and fit with the time units of the remote-sensing data. classes and all LHZs (Table 1). This suggests that different crop calendars were practiced by farmers in different NDVI strata as well as LHZs emphasizing the Pre-processing of the practiced crop calendar data need for pairwise comparison between the spatial The data of all variables for the NDVI classes and the units of each mapping method. LHZs, except for NDVI class 29 and RCS LHZ for The other important point was to determine which planting date, showed significant (p < 0.05) deviation of the two mapping methods maximized the variabil- from normality. Similarly, the variances of the data for ity between the corresponding spatial units. The var- many spatial units were not homogeneous, while only iances of the NDVI classes, represented by chi square a very few were homogeneous (p < 0.05). When the values, were higher than those of the LHZs for the dependent variable deviates from normality, the non- three variables studied correspondingly (Table 1). This parametric statistic is preferred over the parametric suggests that the NDVI stratification method was Table 1. Kruskal-Wallis rank sum test between all NDVI classes and LHZs. Group Variable Chi-squared df Sig. Between all NDVI strata Ploughing date 92.732 2 0.000 Planting date 80.220 2 0.000 Harvesting date 67.725 2 0.000 Between all LHZs Ploughing date 79.832 3 0.000 Planting date 36.683 3 0.000 Harvesting date 37.942 3 0.000 6 E. F. WAYIMA better than the livelihood-based zoning technique in planting and harvesting dates when tested with the maximizing the variability between the resulting spa- Mann-Whitney U test, though significant (p < 0.05) tial units. variation was observed between all the rest for all variables (Table 2). This suggests that the NDVI stra- tification method outperformed the livelihood-based Comparison of the NDVI strata versus the LHZs zoning technique in maximizing the variability between classes. A pair-wise comparison between the NDVI strata and The other important point was to identify LHZs was required to determine if there were differ - a mapping method that was better in maximizing the ences between the spatial units of the two classification internal homogeneity of its corresponding spatial methods. This was accomplished through the use of units. For this purpose, the median absolute deviation a Mann-Whitney U test, which revealed a significant (MAD) was mostly used to assess the variability of the (P < 0.05) difference between the corresponding spatial units of the two mapping methods. However, in NDVI classes and LHZs in many cases (Table 2). In some cases, although the Mann-Whitney U test the relevant NDVI class versus LHZ comparisons, showed significant differences between the relevant 75% (6/8) of ploughing and planting dates as well as spatial units of the two mapping techniques, the 62.5% (5/8) of expected harvesting date revealed sig- MAD of the units were equal. This could be attributed nificant (p < 0.05) difference, while the rest did not to the data unit, dekad. In such cases, the interquartile show statistical difference (Table 2). This indicates the range (IQR) was used to assess the internal homoge- existence of variation between the two mapping meth- neity of each spatial unit. Comparison of the MAD ods in terms of explaining the spatial variability of the and IQR of the spatial units of the two maps (Table 3) practiced crop calendars. Therefore, it is worthwhile to revealed that the NDVI strata were significantly identify a method with better spatial quality in terms (p < 0.05) better in homogeneity than the LHZs in of precisely representing the farmers’ crop calendars. 10 out of 24 (41.67%) NDVI class vs livelihood zone comparisons, while the LHZs were significantly (p < 0.05) better than the NDVI classes by 29.17% Pair-wise comparison of the spatial units of each (7/24). The strata of the two maps were statistically mapping method equally homogeneous in 7 out of 24 (29.17%) relevant A pair-wise comparison of the spatial units of the two comparisons. Thus, the NDVI strata-based crop calen- mapping methods separately is useful to determine dars (Figure 5 and Figure 6) were prepared since the a stratification method that classifies the study area NDVI strata were better than the LHZs in maximizing into mutually exclusive units. In this case, the Mann- the heterogeneity between classes and homogeneity Whitney U test revealed highly significant (p < 0.01) within classes. differences between all pairs of the NDVI classes for all the variables studied (Table 2). This indicates that the NDVI stratification method classified the study area Discussion into independent units or classes for all the variables studied. In contrast, some livelihood zones (such as A crop calendar is an important tool in the agricultural ABW vs RCS and AMT vs SAW) were not signifi - sector because it defines the optimal duration for cantly (p < 0.05) different from one another for various agricultural activities and, hence, is useful for Table 2. Mann-Whitney U test between NDVI classes and LHZs. Comparison Ploughing date Planting date Harvesting date Mapping methods Spatial units MWU Sig. MWU Sig. MWU Sig. NDVI vs LHZ Class 23 vs AMT 9336.50 0.000 1085.00 0.001 1211.50 0.013 Class 23 vs SAW 5887.50 0.000 1107.50 0.001 1476.50 0.152 Class 29 vs ABW 23921.00 0.124 1643.50 0.004 1396.50 0.000 Class 29 vs AMT 19640.50 0.000 1567.00 0.000 2508.50 0.551 Class 29 vs RCS 12722.00 0.000 1541.50 0.127 1373.50 0.016 Class 29 vs SAW 23547.50 0.035 2030.50 0.004 1867.00 0.000 Class 43 vs ABW 14556.00 0.806 1087.50 0.422 1026.50 0.214 Class 43 vs RCS 7874.50 0.000 618.00 0.005 473.00 0.000 NDVI class vs NDVI class Class 23 vs 29 12,018.00 0.000 870.50 0.000 1635.00 0.000 Class 23 vs 43 7333.50 0.000 256.00 0.000 328.00 0.000 Class 29 vs 43 24,943.50 0.066 1265.50 0.000 837.00 0.000 LHZs vs LHZ ABW vs AMT 11295.00 0.009 677.50 0.000 799.50 0.000 ABW vs RCS 6899.00 0.000 739.00 0.069 725.50 0.051 ABW vs SAW 11261.50 0.001 744.50 0.000 557.00 0.000 AMT vs RCS 9773.50 0.004 488.50 0.000 809.50 0.029 AMT vs SAW 9395.00 0.000 1518.00 0.453 1355.50 0.093 RCS vs SAW 6117.00 0.000 683.50 0.001 560.00 0.000 MWU = Mann-Whitney U; Sig. = Asymp. Sig. (2-tailed) GEOLOGY, ECOLOGY, AND LANDSCAPES 7 Table 3. Strata homogeneity statistics for NDVI classes and livelihood zones. Variable Group Mean Standard deviation Median MAD IQR Ploughing date Class-23 103.39 35.407 103.0 38.548 46.7 Class-29 136.62 31.513 143.0 31.135 50.0 Class-43 134.41 37.136 133.0 29.652 38.0 ABW 136.86 32.480 133.0 29.652 38.0 AMT 120.99 39.596 123.0 44.478 60.0 RCS 110.58 29.870 113.0 29.652 40.0 SAW 140.93 36.185 163.0 25.204 50.0 Planting date Class-23 175.32 15.506 172.0 13.343 19.5 Class-29 198.26 15.790 197.0 20.756 25.0 Class-43 223.10 31.053 230.0 44.478 45.0 ABW 218.49 35.670 230.0 51.891 60.0 AMT 185.61 16.483 188.5 21.498 31.0 RCS 202.62 14.644 203.0 13.343 15.5 SAW 188.29 20.827 188.0 28.169 37.5 Expected harvesting date Class-23 318.75 30.186 329.0 29.652 52.5 Class-29 340.15 13.283 339.0 14.826 20.0 Class-43 362.92 20.590 359.0 23.722 27.0 ABW 358.16 24.228 349.0 29.652 37.0 AMT 331.39 30.384 339.0 14.826 20.0 RCS 345.67 11.202 349.0 14.826 10.0 SAW 327.36 21.757 329.0 29.652 37.0 MAD = median absolute deviation; IQR = Interquartile range Figure 5. A NDVI strata-based calendar of the study area for ploughing date using interview data of 2015. The box plot is based on median. effective and resource-efficient planning as well as limited by administrative boundaries (MoARD, crop monitoring and rehabilitation programs. A crop 2010a). Hence, it may be spatially imprecise. This calendar that is prepared based on agro-ecological study has demonstrated that NDVI strata were better parameters reflects the optimum crop phenology, than the LHZs in terms of precisely representing and which in turn implies the precise timing for the rele- explaining the spatial variability of the farmers’ crop vant agronomic practices. However, the LHZ-based calendars. reference crop calendar of Ethiopia is estimated by The purpose of stratification is to maximize the using ecological (altitude and rainfall) as well as non- variation between strata and minimize the variation ecological (population density and market forces) within strata (Cox & Cohen, 1985), so that each stra- parameters, in addition to its weakness that it is tum is unique and internally homogeneous. The 8 E. F. WAYIMA Figure 6. A NDVI strata-based calendar of the study area for planting and expected harvesting dates using interview data of 2015. The box plot is based on median. NDVI classification method showed high variability temporal coverage of the field data was limited to between NDVI classes and low variability within one year of survey work. NDVI classes as compared to the level of variability revealed by the LHZ approach between and within the Disclosure statement corresponding LHZs. The improvement in spatial quality of the NDVI strata could be explained by the No potential conflict of interest was reported by the advantages offered by the NDVI data and the author(s). ISODATA clustering method. Temporal NDVI varia- tion corresponds to vegetation change and growth, and hence, areas that have the same vegetation ORCID dynamics exhibit similar NDVI curves (Zhao et al., Edossa Fikiru Wayima http://orcid.org/0000-0003-0116- 2017) or vegetation phenology. On the other hand, the 961X in-built techniques of the ISODATA clustering method maximize the homogeneity within segments and the heterogeneity between segments (Baatz & References Schäpe, 2000) of the time-series NDVI data that were used as input in the unsupervised classification Baatz, M., & Schäpe, A. (2000). Multiresolution segmenta- tion: An optimization approach for high quality multi- process. scale image segmentation. In Strobl, J., Blaschke, T., & Although this study has clearly demonstrated the Griesbner, G. (Eds.), Angewandte Geographische improvements that the NDVI map-based crop calen- Informationsverarbeitung XI (p. 270). Karlsruhe, dar can offer over the LHZ map-based crop calendar, Germany: Herbert Wichmann Verlag. the research was associated with some limitations: Cox, B. G., & Cohen, S. B. (1985). Methodological issues for only a small number of interview data (about 10–17) health care surveys. CRC Press. de Bie, C. A. J. M. (2000). Comparative performance analysis were collected for each crop; most of the sample points of agro-ecosystems (Wageningen University) [Doctoral were closer to one another; only the major crops in the dissertation]. https://www.wur.nl survey strata were sampled, despite the fact that farm- ESRI. (2016). ArcGIS desktop 10.5 ers grew multiple crops next to one another; the num- FAO. (2014a). Crop forecasting: Inputs. FAO (Food and ber of pixels sampled per strata was very small in agriculture organization of the united nations). http:// www.fao.org/nr/climpag/agromet/inputs_en.asp number (approximately between 3–10); and the GEOLOGY, ECOLOGY, AND LANDSCAPES 9 FAO. (2014b). Length of growing season. FAO (Food and Mumby, P. J., Green, E. P., Edwards, A. J., & Clark, C. D. Agriculture Organization of the United Nations). http:// (1999). The cost-effectiveness of remote sensing for tro- www.fao.org/nr/climpag/about_en.asp pical coastal resources assessment and management. FAO. (2019). AGP - information resources. FAO (Food and Journal of Environmental Management, 55(3), 157–166. agriculture organization of the united nations). http://www. https://doi.org/10.1006/jema.1998.0255 fao.org/agriculture/crops/information-resources/en/ Nancy, L. L., & Anthony, O. J. (2002) A call for greater use of Fatiha, B., Abdelkader, A., Latifa, H., & Mohamed, E. nonparametric statistics [Paper presentation]. Paper pre- (2013). Spatio temporal analysis of vegetation by vegeta- sented at the paper presented at the annual meeting of the tion indices from multi-dates satellite images: Mid-South educational research association, November Application to a semi arid area in Algeria. Energy 6-8, 2002. Chattanooga, TN, 26 (Research in the Schools/ Procedia, 36, 667–675. https://doi.org/10.1016/j.egypro. MSERA). https://eric.ed.gov/?id=ED471346 . 2013.07.077 National Research Council. (2001). Transforming remote FEWSNET. (2009). Ethiopia - livelihood zone map. http:// sensing data into information and applications. National www.fews.net/east-africa/ethiopia/livelihood-zonemap Academy Press. /november-2009 Portmann, F. T., Siebert, S., & Döll, P. (2010). MIRCA2000 - Food and Agriculture Organization of the United Nations global monthly irrigated and rainfed crop areas around [FAO]. (2010). Crop calendar. http://www.fao.org/agricul the year 2000: A new high-resolution data set for agricul- ture/seed/cropcalendar/welcome.do tural and hydrological modeling. Global Biogeochemical Fritz, S., See, L., Bayas, J. C. L., Waldner, F., Jacques, D., Cycles, 24(1). https://doi.org/10.1029/2008GB003435 Becker-Reshef, I., . . . McCallu, I. (2019). A comparison of Rea, J., & Ashley, M. (1976). Phenological evaluations using global agricultural monitoring systems and current gaps. landsat—1 sensors. International Journal of Agricultural Systems, 168, 258–272. https://doi.org/10. Biometeorology, 20, 240–248. https://doi.org/10.1007/ 1016/j.agsy.2018.05.010 BF01553667 Guo, Z. (2013). Mapping the planting dates: An effort to RStudio Team. (2015). RStudio: Open source and retrive crop phenology information from MODIS NDVI enterprise-ready professional software for R. https://www. time series in Africa. IEEE International Geoscience and rstudio.com/ Remote Sensing Symposium - IGARSS, 21-26 July 2013 Sontheimer, S., Callens, K., & Seiffert, B. (1999). Conducting (pp. 3281-3284). Melbourne, VIC, Australia: IEEE. a PRA training and modifying PRA tools to your needs: An https://doi.org/10.1109/IGARSS.2013.6723528 example from a participatory household food security and Heaver, R. (1992). Participatory rural appraisal: Potential nutrition project in Ethiopia - PRA tool box. http://www. applications in family planning, health and nutrition fao.org/3/x5996e/x5996e00.htm programmes. RRA Notes, (16), 13–21. https://www.iied. Umali, C. G., & Exconde, A. B. (2003) Problems in using org/rra-notes-16-applications-for-health remote sensing and GIS in the Philippines [Paper IBM. (2016). IBM SPSS statistics for windows, version 24.0. presentation]. Paper presented at the XII world forestry International Rice Research Institute [IRRI]. (2018). How to congress ,21-28 September 2003, Quebec, Canada. FAO. develop a crop calendar. http://www.knowledgebank.irri. http://www.fao.org/3/XII/0261-B1.htm org/step-by-step-production/pre-planting/crop-calendar van Heemst, H. D. J. (1986). Crop calendar, workability and Khorram, S., Koch, F. H., van der Wiele, C. F., & labour requirements. In H. van Keulen & J. Wolf (Eds.), Nelson, S. A. C. (2012). Remote sensing: SpringerBriefs Modelling of agricultural production: weather, soils and in space development. Springer. crops (pp. 251–262). Pudoc. Kotsuki, S., & Tanaka, K. (2015). SACRA – Global data sets Yap, B. W., & Sim, C. H. (2011). Comparisons of various of satellite-derived crop calendars for agricultural simula- types of normality tests. Journal of Statistical tions: An estimation of a high-resolution crop calendar Computation and Simulation, 81, 2141–2155. https:// using satellite-sensed NDVI. Hydrology and Earth System doi.org/10.1080/00949655.2010.520163 Sciences Discussions, 12, 1329–1353. https://doi.org/10. Zhao, L., Zhang, P., Ma, X., & Pan, Z. (2017). Land cover 5194/hessd-12-1329-2015 information extraction based on daily NDVI time series MAFFIS. (2012). AMIS crop calendar. MAFFIS (Mexican and multiclassifier combination. Mathematical Problems agri-food and fishery information service). Agricultural in Engineering, 2017, 1–13. https://doi.org/10.1155/2017/ Market Informtion System [AMIS]. 6824051 MoARD. (2010a). An atlas of Ethiopian livelihoods. MoARD Zheng, Y., Wu, B., Zhang, M., & Zeng, H. (2016). Crop (Ministry of agriculture and rural development). phenology detection using high spatio-temporal resolu- MoARD. (2010b). Oromia region livelihood profiles. MoARD tion data fused from SPOT5 and MODIS products. (Ministry of agriculture and rural development). Sensors, 16(12), 2099. https://doi.org/10.3390/s16122099
Geology Ecology and Landscapes – Taylor & Francis
Published: Nov 9, 2022
Keywords: Agro-ecology; cropping schedule; farmers’ crop calendar; seasonal calendar; spatial variability; spatial precision
Access the full text.
Sign up today, get DeepDyve free for 14 days.