ABSTRACT
Crop type observation is crucial for various environmental and agricultural remote sensing applications including land use and land cover mapping, crop growth monitoring, crop modelling, yield forecasting, disease surveillance, and climate modelling. Quality-controlled georeferenced crop type information is essential for calibrating and validating machine learning algorithms. However, publicly available field data is scarce, particularly in the highly dynamic smallholder farming systems of sub-Saharan Africa. For the 2020/21 main cropping season (Meher), the Ethiopian Crop Type 2020 (EthCT2020) dataset compiled from multiple sources provides 2,793 harmonized, quality-controlled, and georeferenced in-situ samples on annual crop types (7 crop groups; 22 crop classes) at smallholder field level across the complex and highly fragmented agricultural landscape of Ethiopia. The focus was on rainfed, wheat-based farming systems. A nationwide ground data collection campaign (GDCC; Source 1) was designed using a stratification approach based on wheat crop calendar information, and 1,263 in-situ data samples were collected in selected sampling regions. This in-situ data pool was enriched with 1,530 wheat samples extracted from a) the Wheat Rust Toolbox (WRTB; Source 2; 734 samples), a database for wheat disease surveillance data [1] and b) an inhouse farm household survey database (FHSD; Source 3; 796 samples). Obtained field data was labelled according to the Joint Experiment for Crop Assessment and Monitoring (JECAM) guidelines for cropland and crop type definition and field data collection [2] and the FAO Indicative Crop Classification [3]. The EthCT2020 dataset underwent extensive processing including data harmonization, mixed pixel assessment through visual interpretation using 5 m Planet satellite image composites, and quality-control using Sentinel-2 NDVI homogeneity analysis. The EthCT2020 dataset is unique in terms of crop diversity, pixel purity, and spatial accuracy while targeting a countrywide distribution. It is representative of Ethiopia's complex and highly fragmented agricultural landscape and can be useful for developing new machine learning algorithms for land use land cover mapping, crop type mapping, agricultural monitoring, and yield forecasting in smallholder cropping systems. The dataset can also serve as a baseline input parameter for crop models, climate models, and crop disease and pest forecasting systems.
ABSTRACT
The BELSAR dataset consists of high-resolution multitemporal airborne mono- and bistatic fully-polarimetric synthetic aperture radar (SAR) data in L-band, alongside concurrent measurements of vegetation and soil biogeophysical variables measured in maize and winter wheat fields during the summer of 2018 in Belgium. Its collection was funded by the European Space Agency (ESA) to address the lack of publicly-accessible experimental datasets combining multistatic SAR and in situ measurements. As such, it offers an opportunity to advance the development of SAR remote sensing science and applications for agricultural monitoring and hydrology. This paper aims to facilitate its adoption and exploration by offering comprehensive documentation and integrating its multiple data sources into a unified, analysis-ready dataset.
ABSTRACT
Mapping crop within-field yield variability provide an essential piece of information for precision agriculture applications. Leaf Area Index (LAI) is an important parameter that describes maize growth, vegetation structure, light absorption and subsequently maize biomass and grain yield (GY). The main goal for this study was to estimate maize biomass and GY through LAI retrieved from hyperspectral aerial images using a PROSAIL model inversion and compare its performance with biomass and GY estimations through simple vegetation index approaches. This study was conducted in two separate maize fields of 12 and 20â¯ha located in north-west Mexico. Both fields were cultivated with the same hybrid. One field was irrigated by a linear pivot and the other by a furrow irrigation system. Ground LAI data were collected at different crop growth stages followed by maize biomass and GY at the harvesting time. Through a weekly/biweekly airborne flight campaign, a total of 19 mosaics were acquired between both fields with a micro-hyperspectral Vis-NIR imaging sensor ranging from 400 to 850 nanometres (nm) at different crop growth stages. The PROSAIL model was calibrated and validated for retrieving maize LAI by simulating maize canopy spectral reflectance based on crop-specific parameters. The model was used to retrieve LAI from both fields and to subsequently estimate maize biomass and GY. Additionally, different vegetation indices were calculated from the aerial images to also estimate maize yield and compare the indices with PROSAIL based estimations. The PROSAIL validation to retrieve LAI from hyperspectral imagery showed a R2 value of 0.5 against ground LAI with RMSE of 0.8â¯m2/m2. Maize biomass and GY estimation based on NDRE showed the highest accuracies, followed by retrieved LAI, GNDVI and NDVI with R2 value of 0.81, 0.73, 0.73 and 0.65 for biomass, and 0.83, 0.69, 0.73 and 0.62 for GY estimation, respectively. Furthermore, the late vegetative growth stage at V16 was found to be the best stage for maize yield prediction for all studied indices.
ABSTRACT
In this study, we assess the validity of an African-scale groundwater pollution model for nitrates. In a previous study, we identified a statistical continental-scale groundwater pollution model for nitrate. The model was identified using a pan-African meta-analysis of available nitrate groundwater pollution studies. The model was implemented in both Random Forest (RF) and multiple regression formats. For both approaches, we collected as predictors a comprehensive GIS database of 13 spatial attributes, related to land use, soil type, hydrogeology, topography, climatology, region typology, nitrogen fertiliser application rate, and population density. In this paper, we validate the continental-scale model of groundwater contamination by using a nitrate measurement dataset from three African countries. We discuss the issue of data availability, and quality and scale issues, as challenges in validation. Notwithstanding that the modelling procedure exhibited very good success using a continental-scale dataset (e.g. R2 = 0.97 in the RF format using a cross-validation approach), the continental-scale model could not be used without recalibration to predict nitrate pollution at the country scale using regional data. In addition, when recalibrating the model using country-scale datasets, the order of model exploratory factors changes. This suggests that the structure and the parameters of a statistical spatially distributed groundwater degradation model for the African continent are strongly scale dependent.
Subject(s)
Environmental Monitoring/methods , Groundwater/chemistry , Models, Statistical , Nitrates/analysis , Water Pollutants, Chemical/analysis , Water Pollution/analysis , Africa , Fertilizers/analysis , Humans , Multivariate Analysis , Nitrogen , Reproducibility of ResultsABSTRACT
In sub-Saharan Africa, transaction costs are believed to be the most significant barrier that prevents smallholders and farmers from gaining access to markets and productive assets. In this study, we explore the impact of social capital on millet prices for three contrasted years in Senegal. Social capital is approximated using a unique data set on mobile phone communications between 9 million people allowing to simulate the business network between economic agents. Our approach is a spatial equilibrium model that integrates a diversified set of data. Local supply and demand were respectively derived from remotely sensed imagery and population density maps. The road network was used to establish market catchment areas, and transportation costs were derived from distances between markets. Results demonstrate that accounting for the social capital in the transaction costs explained 1-9% of the price variance depending on the year. The year-specific effect remains challenging to assess but could be related to a strengthening of risk aversion following a poor harvest.
ABSTRACT
This study evaluates the potential of high resolution hyperspectral airborne imagery to capture within-field variability of durum wheat grain yield (GY) and grain protein content (GPC) in two commercial fields in the Yaqui Valley (northwestern Mexico). Through a weekly/biweekly airborne flight campaign, we acquired 10 mosaics with a micro-hyperspectral Vis-NIR imaging sensor ranging from 400-850 nanometres (nm). Just before harvest, 114 georeferenced grain samples were obtained manually. Using spectral exploratory analysis, we calculated narrow-band physiological spectral indices-normalized difference spectral index (NDSI) and ratio spectral index (RSI)-from every single hyperspectral mosaic using complete two by two combinations of wavelengths. We applied two methods for the multi-temporal hyperspectral exploratory analysis: (a) Temporal Principal Component Analysis (tPCA) on wavelengths across all images and (b) the integration of vegetation indices over time based on area under the curve (AUC) calculations. For GY, the best R2 (0.32) were found using both the spectral (NDSI-Ri, 750 to 840 nm and Rj, ±720-736 nm) and the multi-temporal AUC exploratory analysis (EVI and OSAVI through AUC) methods. For GPC, all exploratory analysis methods tested revealed (a) a low to very low coefficient of determination (R2 ≤ 0.21), (b) a relatively low overall prediction error (RMSE: 0.45-0.49%), compared to results from other literature studies, and (c) that the spectral exploratory analysis approach is slightly better than the multi-temporal approaches, with early season NDSI of 700 with 574 nm and late season NDSI of 707 with 523 nm as the best indicators. Using residual maps from the regression analyses of NDSIs and GPC, we visualized GPC within-field variability and showed that up to 75% of the field area could be mapped with relatively good predictability (residual class: -0.25 to 0.25%), therefore showing the potential of remote sensing imagery to capture the within-field variation of GPC under conventional agricultural practices.
ABSTRACT
A global reference data set on cropland was collected through a crowdsourcing campaign using the Geo-Wiki crowdsourcing tool. The campaign lasted three weeks, with over 80 participants from around the world reviewing almost 36,000 sample units, focussing on cropland identification. For quality assessment purposes, two additional data sets are provided. The first is a control set of 1,793 sample locations validated by students trained in satellite image interpretation. This data set was used to assess the quality of the crowd as the campaign progressed. The second data set contains 60 expert validations for additional evaluation of the quality of the contributions. All data sets are split into two parts: the first part shows all areas classified as cropland and the second part shows cropland average per location and user. After further processing, the data presented here might be suitable to validate and compare medium and high resolution cropland maps generated using remote sensing. These could also be used to train classification algorithms for developing new maps of land cover and cropland extent.
ABSTRACT
The lack of sufficient ground truth data has always constrained supervised learning, thereby hindering the generation of up-to-date satellite-derived thematic maps. This is all the more true for those applications requiring frequent updates over large areas such as cropland mapping. Therefore, we present a method enabling the automated production of spatially consistent cropland maps at the national scale, based on spectral-temporal features and outdated land cover information. Following an unsupervised approach, this method extracts reliable calibration pixels based on their labels in the outdated map and their spectral signatures. To ensure spatial consistency and coherence in the map, we first propose to generate seamless input images by normalizing the time series and deriving spectral-temporal features that target salient cropland characteristics. Second, we reduce the spatial variability of the class signatures by stratifying the country and by classifying each stratum independently. Finally, we remove speckle with a weighted majority filter accounting for per-pixel classification confidence. Capitalizing on a wall-to-wall validation data set, the method was tested in South Africa using a 16-year old land cover map and multi-sensor Landsat time series. The overall accuracy of the resulting cropland map reached 92%. A spatially explicit validation revealed large variations across the country and suggests that intensive grain-growing areas were better characterized than smallholder farming systems. Informative features in the classification process vary from one stratum to another but features targeting the minimum of vegetation as well as short-wave infrared features were consistently important throughout the country. Overall, the approach showed potential for routinely delivering consistent cropland maps over large areas as required for operational crop monitoring.
Subject(s)
Crops, Agricultural , Geographic Mapping , Geographic Information Systems , Geography , Models, Theoretical , Reproducibility of Results , South AfricaABSTRACT
We estimated vulnerability and pollution risk of groundwater at the pan-African scale. We therefore compiled the most recent continental scale information on soil, land use, geology, hydrogeology and climate in a Geographical Information System (GIS) at a resolution of 15 km × 15 km and at the scale of 1:60,000,000. The groundwater vulnerability map was constructed by means of the DRASTIC method. The map reveals that groundwater is highly vulnerable in Central and West Africa, where the watertable is very low. In addition, very low vulnerability is found in the large sedimentary basins of the African deserts where groundwater is situated in very deep aquifers. The groundwater pollution risk map is obtained by overlaying the DRASTIC vulnerability map with land use. The northern, central and western part of the African continent is dominated by high pollution risk classes and this is very strongly related to shallow groundwater systems and the development of agricultural activities. Subsequently, we performed a sensitivity analysis to evaluate the relative importance of each parameter on groundwater vulnerability and pollution risk. The sensitivity analysis indicated that the removal of the impact of vadose zone, the depth of the groundwater, the hydraulic conductivity and the net recharge causes a large variation in the mapped vulnerability and pollution risk. The mapping model was validated using nitrate concentration data of groundwater as a proxy of pollution risk. Pan-African concentration data were inferred from a meta-analysis of literature data. Results shows a good match between nitrate concentration and the groundwater pollution risk classes. The pan African assessment of groundwater vulnerability and pollution risk is expected to be of particular value for water policy and for designing groundwater resources management programs. We expect, however, that this assessment can be strongly improved when better pan African monitoring data related to groundwater pollution will be integrated in the assessment methodology.
Subject(s)
Environmental Monitoring , Groundwater/chemistry , Water Pollution/statistics & numerical data , Water Supply/statistics & numerical data , Africa , Geographic Information SystemsABSTRACT
Policies to reduce emissions from deforestation and forest degradation largely depend on accurate estimates of tropical forest carbon stocks. Here we present the first field-based carbon stock data for the Central Congo Basin in Yangambi, Democratic Republic of Congo. We find an average aboveground carbon stock of 162 ± 20 Mg C ha(-1) for intact old-growth forest, which is significantly lower than stocks recorded in the outer regions of the Congo Basin. The best available tree height-diameter relationships derived for Central Africa do not render accurate canopy height estimates for our study area. Aboveground carbon stocks would be overestimated by 24% if these inaccurate relationships were used. The studied forests have a lower stature compared with forests in the outer regions of the basin, which confirms remotely sensed patterns. Additionally, we find an average soil carbon stock of 111 ± 24 Mg C ha(-1), slightly influenced by the current land-use change.
Subject(s)
Carbon/metabolism , Trees/anatomy & histology , Climate , Congo , Conservation of Natural Resources , Geography , Models, Biological , Regression Analysis , Soil/chemistryABSTRACT
BACKGROUND: Southeast Asia presents a high diversity of Anopheles. Environmental requirements differ for each species and should be clarified because of their influence on malaria transmission potential. Monitoring projects collect vast quantities of entomological data over the whole region and could bring valuable information to malaria control staff but collections are not always standardized and are thus difficult to analyze. In this context studying species associations and their relation to the environment offer some opportunities as they are less subject to sampling error than individual species. METHODS: Using asymmetrical similarity coefficients, indirect clustering and the search of indicator species, this paper identified species associations. Environmental influences were then analysed through canonical and discriminant analysis using climatic and topographic data, land cover in a 3 km buffer around villages and vegetation indices. RESULTS: Six groups of sites characterized the structure of the species assemblage. Temperature, rainfall and vegetation factors all play a role. Four out of the six groups of sites based on species similarities could be discriminated using environmental information only. CONCLUSIONS: Vegetation indices derived from satellite imagery proved very valuable with one variable explaining more variance of the species dataset than any other variable. The analysis could be improved by integrating seasonality in the sampling and collecting at least 4 consecutive days.
Subject(s)
Anopheles/physiology , Biota , Animals , Anopheles/growth & development , Asia, Southeastern , Climate , Geography , HumansABSTRACT
Methods derived from ecological niche modeling allow to define species distribution based on presence-only data. This is particularly useful to develop models from literature records such as available for the Anopheles dirus complex, a major group of malaria mosquito vectors in Asia. This research defines an innovative modeling design based on presence-only model and hierarchical framework to define the distribution of the complex and attempt to delineate sibling species distribution and environmental preferences. At coarse resolution, the potential distribution was defined using slow changing abiotic factors such as topography and climate representative for the timescale covered by literature records of the species. The distribution area was then refined in a second step using a mask of current suitable land cover. Distribution area and ecological niche were compared between species and environmental factors tested for relevance. Alternatively, extreme values at occurrence points were used to delimit environmental envelopes. The spatial distribution for the complex was broadly consistent with its known distribution and influencing factors included temperature and rainfall. If maps developed from environmental envelopes gave similar results to modeling when the number of sites was high, the results were less similar for species with low number of recorded presences. Using presence-only models and hierarchical framework this study not only predicts the distribution of a major malaria vector, but also improved ecological modeling analysis design and proposed final products better adapted to malaria control decision makers. The resulting maps can help prioritizing areas which need further investigation and help simulate distribution under changing conditions such as climate change or reforestation. The hierarchical framework results in two products one abiotic based model describes the potential maximal distribution and remains valid for decades and the other including a biotic mask easy to update with frequently available information gives current species distribution.
Subject(s)
Animal Distribution , Anopheles/physiology , Ecosystem , Insect Vectors/physiology , Spatial Analysis , Animals , Anopheles/classification , Asia , Climate , Insect Vectors/classification , Malaria/prevention & control , Phylogeny , TemperatureABSTRACT
BACKGROUND: The Anopheles dirus complex includes efficient malaria vectors of the Asian forested zone. Studies suggest ecological and biological differences between the species of the complex but variations within species suggest possible environmental influences. Behavioural variation might determine vector capacity and adaptation to changing environment. It is thus necessary to clarify the species distributions and the influences of environment on behavioural heterogeneity. METHODS: A literature review highlights variation between species, influences of environmental drivers, and consequences on vector status and control. The localisation of collection sites from the literature and from a recent project (MALVECASIA) produces detailed species distributions maps. These facilitate species identification and analysis of environmental influences. RESULTS: The maps give a good overview of species distributions. If species status partly explains behavioural heterogeneity, occurrence and vectorial status, some environmental drivers have at least the same importance. Those include rainfall, temperature, humidity, shade, soil type, water chemistry and moon phase. Most factors are probably constantly favourable in forest. Biological specificities, behaviour and high human-vector contact in the forest can explain the association of this complex with high malaria prevalence, multi-drug resistant Plasmodium falciparum and partial control failure of forest malaria in Southeast Asia. CONCLUSION: Environmental and human factors seem better than species specificities at explaining behavioural heterogeneity. Although forest seems essential for mosquito survival, adaptations to orchards and wells have been recorded. Understanding the relationship between landscape components and mosquito population is a priority in foreseeing the influence of land-cover changes on malaria occurrence and in shaping control strategies for the future.