Multi-temporal Normalized Difference Vegetation Index Research Articles

Rice Yield Estimation Using Multi-Temporal Remote Sensing Data and Machine Learning: A Case Study of Jiangsu, China

Effective estimation of crop yields at a regional scale holds significant importance in facilitating decision-making within the agricultural sector, thereby ensuring grain security. However, traditional ground-based measurement techniques suffer from inefficiencies, and there exists a need for a reliable, precise, and effective method for estimating regional rice yields. In this study, we employed four machine-learning techniques: partial least squares regression (PLSR), support vector regression (SVR), random forest regression (RFR), and back propagation neural network (BPNN). We combined these methods with multi-temporal rice NDVI (normalized difference vegetation index) data for rice yield estimation. Following an accuracy evaluation and a spatial analysis, the key findings of our study are as follows. (1) The RFR model emerged as the most accurate for rice yield estimation, achieving an R2 of 0.65, an RMSE of 388.79 kg/ha, and an rRMSE of 4.48%. While PLSR and SVR demonstrated comparable accuracy, they were both inferior to RFR. (2) Using the top seven predictors with the highest importance rankings as inputs for the RFR model (NDVI values on the 6th, 17th, 33rd, 44th, 71st, 90th, and 106th days after the rice transplanting stage) achieved comparable accuracy while reducing information redundancy. (3) The proposed model demonstrated good spatial applicability (MI = −0.03) for rice yield estimation in Jiangsu, China. (4) A high spatial resolution yearly rice yield dataset (1 km) spanning from 2001 to 2020 was generated using the proposed model and is accessible on the Zenodo database. In conclusion, this study has demonstrated the efficacy of combining multi-temporal remote sensing data with machine-learning techniques for accurate rice yield estimation, thereby aiding agricultural authorities and production enterprises in the timely formulation and refinement of cropping strategies and management policies for the ongoing season.

Assessment of vegetation cover changes and the contributing factors in the Al-Ahsa Oasis using Normalized Difference Vegetation Index (NDVI)

The abandonment of date palm grove of the former Al-Ahsa Oasis in the eastern region of Saudi Arabia has resulted in the conversion of delicate agricultural area into urban area. The current state of the oasis is influenced by both expansion and degradation factors. Therefore, it is important to study the spatiotemporal variation of vegetation cover for the sustainable management of oasis resources. This study used Landsat satellite images in 1987, 2002, and 2021 to monitor the spatiotemporal variation of vegetation cover in the Al-Ahsa Oasis, applied multi-temporal Normalized Difference Vegetation Index (NDVI) data spanning from 1987 to 2021 to assess environmental and spatiotemporal variations that have occurred in the Al-Ahsa Oasis, and investigated the factors influencing these variation. This study reveals that there is a significant improvement in the ecological environment of the oasis during 1987–2021, with increase of NDVI values being higher than 0.10. In 2021, the highest NDVI value is generally above 0.70, while the lowest value remains largely unchanged. However, there is a remarkable increase in NDVI values between 0.20 and 0.30. The area of low NDVI values (0.00–0.20) has remained almost stable, but the region with high NDVI values (above 0.70) expands during 1987–2021. Furthermore, this study finds that in 1987–2002, the increase of vegetation cover is most notable in the northern region of the study area, whereas from 2002 to 2021, the increase of vegetation cover is mainly concentrated in the northern and southern regions of the study area. From 1987 to 2021, NDVI values exhibit the most pronounced variation, with a significant increase in the “green” zone (characterized by NDVI values exceeding 0.40), indicating a substantial enhancement in the ecological environment of the oasis. The NDVI classification is validated through 50 ground validation points in the study area, demonstrating a mean accuracy of 92.00% in the detection of vegetation cover. In general, both the user’s and producer’s accuracies of NDVI classification are extremely high in 1987, 2002, and 2021. Finally, this study suggests that environmental authorities should strengthen their overall forestry project arrangements to combat sand encroachment and enhance the ecological environment of the Al-Ahsa Oasis.

Potential of the satellite-based Dynamic Habitat Index (DHI) to capture changes in soil properties and drought conditions across Land Use/Land Cover types in a Central European landscape

This study evaluates the sensitivity of the Dynamic Habitat Index (DHI), utilizing multitemporal Normalized Difference Vegetation Index (NDVI) data, to changing environmental conditions across Land Use/Land Cover (LULC) types in a central European landscape (2017–2020). We observed distinct DHI characteristics for all LULC types, and the DHI responded to an extreme drought year in 2018 with no return to pre-drought conditions except for deciduous forests. The DHI also effectively captured spatio-temporal variability of pedo-climatic conditions. Thus, integrated with ancillary geodata, the DHI enhances traditional categorical LULC maps, offering applications in biodiversity and ecosystem research. Such integrated products could serve as valuable tools for decision makers to formulate sustainable land management strategies and contribute to Sustainable Develop Goal indicators related to land degradation, e.g. by identifying deviations from typical, context-specific DHI profiles as a response to disturbance and environmental stress.

Monitoring vegetation dynamics using multi-temporal Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI) images of Tamil Nadu

Vegetation indices serve as an essential tool in monitoring variations in vegetation. The vegetation indices used often, viz., normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) were computed from MODIS vegetation index products. The present study aimed to monitor vegetation's seasonal dynamics by using time series NDVI and EVI indices in Tamil Nadu from 2011 to 2021. Two products characterize the global range of vegetation states and processes more effectively. The data sources were processed and the values of NDVI and EVI were extracted using ArcGIS software. There was a significant difference in vegetation intensity and status of vegetation over time, with NDVI having a larger value than EVI, indicating that biomass intensity varies over time in Tamil Nadu. Among the land cover classes, the deciduous forest showed the highest mean values for NDVI (0.83) and EVI (0.38), followed by cropland mean values of NDVI (0.71) and EVI (0.31) and the lowest NDVI (0.68) and EVI (0.29) was recorded in the scrubland. The study demonstrated that vegetation indices extracted from MODIS offered valuable information on vegetation status and condition at a short temporal time period.

Multi-temporal NDVI analysis using UAV images of tree crowns in a northern Mexican pine-oak forest

Multi-temporal NDVI analysis using UAV images of tree crowns in a northern Mexican pine-oak forest

Assessment of Post-Fire Phenological Changes Using MODIS-Derived Vegetative Indices in the Semiarid Oak Forests

Wildfire has significant impact on plant phenology. The plants’ phenological variables, derived from time series satellite data, can be monitored and the changes in satellite imagery may be used to identify the beginning, peak, and end of the growing season. This study investigated the use of remote sensing data and land surface phenology (LSP) parameters to evaluate the impacts of fire. The LSP parameters included the start of growing season (SOS), the length of the growing season (LOS), the end of the growing season (EOS), maximum greenness of the season (Gmax), and minimum greenery in the season (Gmin) in the fire-impacted, semiarid oak forests of Iran. These LSP parameters were extracted from multitemporal normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI2) data, acquired from MODIS sensor images in Zagros of the Ilam province in western Iran. By extracting LSP indices from the NDVI and EVI2 data, the indices were compared between burned forest areas, areas surrounding the burned forests, and unburned areas and for timesteps representing pre-fire, fire (i.e., year of fire), and post-fire (i.e., 2 years) conditions. It was found that for the burned area, there were significant differences in Gmax and the day that Gmax occurred. Furthermore, there was also a significant difference in Gmin between the pre- and post-fire conditions when NDVI was used and a significant difference between Gmax when EVI2 was used. The results also showed that in both time series there was a significant difference between the burned and control area in terms of Gmax. In general, the results showed that the fire had a negative effect on LSP, but in the two years after the fire, there were signs of forest restoration. This study provides necessary information to inform forest and resource conservation and restoration programs.

A Global Remote-Sensing Assessment of the Intersite Variability in the Greening of Coastal Dunes

In recent decades, the vegetation on many coastal dunes has expanded spatially, which is attributed, among other things, to global-scale climate change. The intersite variability in this dune greening has not yet been substantially investigated, nor is it known whether it is consistent with intersite variability in climate change. Therefore, the objectives of this work were firstly to quantify and analyse the change in vegetation cover from multitemporal NDVI time series at a large number (186) of dune fields worldwide, calculated from Landsat satellite imagery available between 1984 and 2021 and secondly, to correlate the identified trends with trends in the main climate variables influencing vegetation growth (temperature, precipitation and wind speed). We show that greening is strongest in cool temperate climates (35° to 66.5° north/south latitudes) and that the rate of greening is accelerating at many sites. We find no dependence between the rate of greening and the local temporal change in temperature, precipitation and/or wind speed. Based on existing literature, sand supply and anthropogenic activities are discussed as possible reasons for the absence of a clear global relationship between variability in dune greening and climate change.

Stand Structural Characteristics Derived from Combined TLS and Landsat Data Support Predictions of Mushroom Yields in Mediterranean Forest

Forest fungi provide recreational and economic services, as well as ecosystem biodiversity. Wild mushroom yields are difficult to estimate; climatic conditions are known to trigger temporally localised yields, and forest structure also affects productivity. In this work, we analyse the capacity of remotely sensed variables to estimate wild mushroom biomass production in Mediterranean Pinus pinaster forests in Soria (Spain) using generalised additive mixed models (GAMMs). In addition to climate variables, multitemporal NDVI derived from Landsat data, as well as structural variables measured with mobile Terrestrial Laser Scanner (TLS), are considered. Models are built for all mushroom species as a single pool and for Lactarius deliciosus individually. Our results show that, in addition to autumn precipitation, the interaction of multitemporal NDVI and vegetation biomass are most explanatory of mushroom productivity in the models. When analysing the productivity models of Lactarius deliciosus, in addition to the interaction between canopy cover and autumn minimum temperature, basal area (BA) becomes relevant, indicating an optimal BA range for the development of this species. These findings contribute to the improvement of knowledge about wild mushroom productivity, helping to meet Goal 15 of the 2030 UN Agenda.

Rapid method for yearly LULC classification using Random Forest and incorporating time-series NDVI and topography: a case study of Thanh Hoa province, Vietnam

Land-use and land-cover (LULC) mapping in the complex area is a challenging task due to the mixed vegetation patterns, and rough mountains with fast-flowing rivers. In Vietnam, LULC update is not frequently. In this study, we applied a supervised machine learning (Random forest—RF) approach to mapping LULC in Thanh Hoa province, Vietnam from 2011 to 2015 utilizing multi-temporal Normalized Difference Vegetation Index (NDVI) data from MODIS, combined with topographic features. Random forest classification (RFC) reached a total prediction accuracy of 91% and Kappa coefficient (K) of 0.89 across eight LULCs. Besides, the results showed that the features extracted from time-series NDVI comprising the mean of yearly NDVI, the sum of NDVI, and the topography were the important variables controlling the LULC classification. For similar studies on the distribution of LULC, the method proposed in this study could be helpful.

Remote Sensing and Phytoecological Methods for Mapping and Assessing Potential Ecosystem Services of the Ouled Hannèche Forest in the Hodna Mountains, Algeria

Regardless of their biogeographic origins or degree of artificialization, the world’s forests are a source of a wide range of ecosystem services (ES). However, the quality and quantity of these services depend on the type of forest studied and its phytogeographic context. Our objective is to transpose the concept of ES, in particular, the assessment of forest ES, to the specific Mediterranean context of the North African mountains, where this issue is still in its infancy and where access to the data needed for assessment remains difficult. Our work presents an introductory approach, allowing us to set up methodological and scientific milestones based on open-access remote sensing data and already tested geospatial processing associated with phytoecological surveys to assess the ES provided by forests in an Algerian study area. Specifically, several indicators used to assess (both qualitatively and quantitatively) the potential ES of the Ouled Hannèche forest, a forest located in the Hodna Mountains, are derived from LANDSAT 8 OLI images from 2017 and an ALOS AW3D30 DSM. The qualitative ES typology is jointly based on an SVM classification of topographically corrected LANDSAT images and a geomorphic-type classification using the geomorphon method. NDVI is a quantitative estimator of many plant ecosystem functions related to ES. It highlights the variations in the provision of ES according to the types of vegetation formations present. It serves as a support for estimating spectral heterogeneity through Rao’s quadratic entropy, which is considered a relative indicator of biodiversity at the landscape scale. The two previous variables (the multitemporal NDVI and Rao’s Q), completed by the Shannon entropy method applied to the geomorphon classes as a proxy for topo-morphological heterogeneity, constitute the input variables of a quantitative map of the potential supply of ES in the forest determined by Spatial Multicriteria Analysis (SMCA). Ultimately, our results serve as a useful basis for land-use planning and biodiversity conservation.

Multi-temporal Landsat-derived NDVI for vegetation cover degradation for the period 1984-2018 in part of the Arganeraie Biosphere Reserve (Morocco)

Biosphere Reserves (BRs) are especial zones established to conserve the biodiversity (ecosystem, species or genetic). Therefore, evaluating and analyzing the ecosystem degradation in these zones are crucial for protecting them. This study attempts to assess and analyze the vegetation cover degradation during the last three decades (period between 1984 and 2018) in part of the Arganeraie Biosphere Reserve (ABR) of Morocco. The study aims to describe the ecosystem degradation scenario by using multi-temporal Landsat-derived NDVI (Normalized Difference Vegetation Index), and to highlight the reasonable factors causing its degradation. Long-term and short-term NDVI changes were investigated using image difference change detection technique and IsoData classification. The obtained results confirm that about 3644.64 ha of the vegetation cover was degraded during last decade. Field investigations confirm that the anthropogenic activities caused vegetation cover degradation that last a long time (2007–2018) as confirmed by long-term analysis. Forest fire, due to charcoal production, caused degradation of about 2989.59 ha of the vegetation cover during the period 2010–2018. This study has highlighted the importance to use long-term NDVI change detection to map anthropogenic-induced vegetation cover degradation and to use short-term NDVI change detection for best chronological analysis of ecosystem degradation. In addition, the easiest proposed method offers a user-friendly tool to monitor ecosystem degradation.

RELATIONSHIPS BETWEEN VEGETATION INDICES AND RAINFALL AND PET AT DIFFERENT TIME-LAGS: A STUDY AT A MEDITERRANEAN TO ARID GRADIENT

Abstract. The climatic gradient between the Judean Lowland and the Negev Desert in Central Israel represents a transition zone between dense shrublands in the North, and desert fringe Batha and Irano-Turanian vegetation in the South, characterizing wide Mediterranean Type Climate regions around the world. Understanding the expected response of these water-limited ecosystems to climate change presents a significant challenge due to the high geodiversity of Mediterranean environments. Studying relationships between vegetation patterns and climatic parameters is fundamental for this purpose, and remote sensing provides a valuable tool for investigating these relationships over large regions. This study aims at examining the relationships between NDVI extracted from Sentinel2 and rainfall and PET accumulated over 1 to 6 months. The analysis was first conducted for 38 sites (100×100 meters) across the climatic gradient for three years representing high (2016), low (2017), and average (2018) rainfall. Results indicate that the highest correlation between NDVI and climatic parameters is achieved for accumulation interval of two months. Least-squares analysis was then utilized for calculating the per-pixel regression coefficients between NDVI and corresponding rainfall and PET. Classification of the multi-temporal NDVI (2016–2018) and of the linear regressions’ coefficients between NDVI and rainfall and PET at accumulation interval of 2 months yielded both high accuracies. Since these slope and intercept coefficients can be perceived as representing the water-use regime at each pixel, the similarity between the classification results suggests that multi-temporal NDVI typologies correspond water-use regime typologies across desert fringe ecosystems at the margins of Mediterranean regions.

Synergetic use of multi-temporal Sentinel-1, Sentinel-2, NDVI, and topographic factors for estimating soil organic carbon

Soil organic carbon (SOC) is a reliable indicator of soil productivity and land management. Despite the widespread use of remote sensing data in estimating SOC variation, the predictive power of single-date and multi-temporal as well as the individual and combined effects of radar and optical images have rarely been investigated. This study created six two-month interval composite layers including VH and VV polarizations of Sentinel-1 (S1), spectral bands of Sentinel-2 (S2), normalized difference vegetation index (NDVI), slope and altitude for the target year of 2019–2020 to estimate the SOC variation. Based on 80 soil samples (0–20 cm) taken in a field survey from northern Iran, random forest (RF) and support vector regression (SVR) were calibrated to estimate the SOC content. All models were fine-tuned using grid search and evaluated using spatial cross-validation. The results showed that multi-temporal data were more predictive than single-date data, and SVR outperformed the RF algorithm. SVR model using the multi-temporal S2 achieved the highest R2 and RMSE values of 57.59% and 0.94%, respectively. Feature selection has shown no benefit of adding S1 data and identified band-2 (green), band-3 (red), and band-10 (shortwave infrared) of S2 imagery as the most explanatory variables of SOC variation. The results of this study highlight the potential of freely available high-resolution S2 for mapping SOC variation using the SVR model. SVR is particularly suggested in small sample size studies, an issue which is dominant in soil studies.

Development of a New Phenology Algorithm for Fine Mapping of Cropping Intensity in Complex Planting Areas Using Sentinel-2 and Google Earth Engine

Cropping intensity is a key indicator for evaluating grain production and intensive use of cropland. Timely and accurately monitoring of cropping intensity is of great significance for ensuring national food security and improving the level of national land management. In this study, we used all Sentinel-2 images on the Google Earth Engine cloud platform, and constructed an improved peak point detection method to extract the cropping intensity of a heterogeneous planting area combined with crop phenology. The crop growth cycle profiles were extracted from the multi-temporal normalized difference vegetation index (NDVI) and land surface water index (LSWI) datasets. Results show that by 2020, the area of single cropping, double cropping, and triple cropping in the Henan Province are 52,236.9 km2, 74,334.1 km2, and 1927.1 km2, respectively; the corresponding producer accuracies are 86.12%, 93.72%, and 91.41%, respectively; the corresponding user accuracies are 88.99%, 92.29%, and 71.26%, respectively. The overall accuracy is 90.95%, and the Kappa coefficient is 0.81. Using the sown area in the statistical yearbook data of cities in the Henan Province to verify the extraction results of this paper, the R2 is 0.9717, and the root mean square error is 1715.9 km2. This study shows that using all the Sentinel-2 data, the phenology algorithm, and cloud computing technology has great potential in producing a high spatio-temporal resolution dataset for crop remote sensing monitoring and agricultural policymaking in complex planting areas.

Spatial Analysis of Agronomic Data and UAV Imagery for Rice Yield Estimation

In this study, a spatial analysis of agronomic and remote sensing data is carried out to derive accurate rice crop yield estimation. The variability of a series of vegetation indices (VIs) was calculated from remote sensing data obtained via a commercial UAS platform (e-Bee) at four dates (per stage of development), and the development of estimation models was conducted. The study area is located in the region of Chalastra (municipality of Thessaloniki, North Greece) and the primary data were obtained during the 2016 growing season. These data include ultra-high resolution remote sensing multispectral images of 18 plots totaling 58 hectares of Ronaldo and Gladio rice crop varieties, 97 sample point data related to yield, and many other pieces of information recorded in the producer’s field log. Ten simple and compound VIs were calculated, and the evolution of their values during the growing season as well as their comparative correlation were studied. A study of the usability of each VI was conducted for the different phenological stages of the cultivation and the variance of VIs and yield; the more correlated VIs were identified. Furthermore, three types of multitemporal VI were calculated from combinations of VIs from different dates, and their contribution to improving yield prediction was studied. As Ronaldo is a Japonica type of rice variety and Gladio is Indica type, they behave differently in terms of maturation time (Gladio is approximately 20 days earlier) and the value of every VI is affected by changes in plant physiology and phenology. These differences between the two varieties are reflected in the multitemporal study of the single-date VIs but also in the study of the values of the multitemporal VIs. In conclusion, Ronaldo’s yield is strongly dependent on multitemporal NDVI (VI6th July + VI30 Aug, R2 = 0.76), while Gladio’s yield is strongly dependent on single-date NDVI (6 July, R2 = 0.88). The compound VIs RERDVI and MCARI1 have the highest yield prediction (R2 = 0.77) for Ronaldo (VI6th July + VI30 Aug) and Gladio (R2 = 0.95) when calculated in the booting stage, respectively. For the Ronaldo variety, the examination of the multitemporal VIs increases yield prediction accuracy, while in the case of the Gladio variety the opposite is observed. The capabilities of multitemporal VIs in yield estimation by combining UAVs with more flights during the different growth stages can improve management and the cultivation practices.

Google Earth Engine for large-scale land use and land cover mapping: an object-based classification approach using spectral, textural and topographical factors

ABSTRACT Mapping the distribution and type of land use and land cover (LULC) is essential for watershed management. The Tigris-Euphrates basin is a transboundary region in the Middle East shared between six countries, but a recent fine-scale LULC map of the area is lacking. Using Landsat-8 time series, a 30-m resolution LULC map was produced for the Tigris-Euphrates basin. In total, 1184 Landsat scenes were processed within the Google Earth Engine (GEE). For the collection of ground truth data, differential manifestations of green cover were considered by dividing the study area into five climatic regions and the training samples were taken from each sub-region. To account for the temporal variation of LULC types, six two-month interval composite layers, including the spectral and thermal bands of Landsat-8, texture and spectral indices, as well as topographic factors were created for the target year 2019. Image segmentation and classification were performed using the simple non-iterative clustering (SNIC) and Random Forest (RF) algorithms, respectively. A computationally effective parallel processing approach was developed, which created a number of tiles and sub-tiles and a bulk command was converted into smaller parallel commands. The generated LULC map showed a satisfactory overall accuracy of 91.7%, with the highest User’s accuracy in water and wetland, and the lowest in rainfed crop and rangeland and the highest Producer’s accuracy in water and barren areas, and the lowest in garden and rangeland. This study highlights the necessity of using multi-temporal data for LULC mapping, in particular, multi-temporal NDVI, for the separation of different green cover types in arid and semi-arid environment.

Analysis of Optimum Growth Stages for Winter Crop Separability using Multi-Temporal NDVI Profiles in Vijapur Taluka, Gujarat State

The multi-date Remote sensing satellite data has shown its great potential in agricultural applications. The crop monitoring throughout its growing season requires up-to-date and timely information for better crop growth management. In the present study analysis of growth stages of different winter crops grown in Vijapur Taluka was carried for determining optimum growth period for crop separability using multi-temporal NDVI profiles. The multi-temporal Sentinel-2 multi-spectral data (spatial resolution 10-m) covering study area for the winter crop period from November-2018 to March-2019 was downloaded from https://earthexplorer.usgs.gov/. Ground Truth (GT) data on winter crops grown in different villages of Vijapur Taluka was collected at multiple locations coinciding with different growth stages of winter crops. The spectral behavior of wheat, cotton, potato, fennel and castor crops during active growth stages was studied and it was observed that the spectral response of these crops is quite distinct. The maximum spectral difference was observed in the NIR spectral band as compared to other visible spectral bands of Sentinel-2 multi-spectral data. However, wheat and potato have almost similar spectral response during the month of January when wheat was at flag leaf to flowering stage and potato was at maximum vegetative growth stage. This suggests that, single-date Sentinel-2 multispectral digital data of active vegetative growth stages of these two crops during the month of January is not appropriate and sufficient for unique discrimination of wheat and potato. Therefore, Normalized Difference Vegetation Index (NDVI) profiles of these major winter crops were generated to monitor crop growth stages where maximum spectral differences can be observed. The results of NDVI growth profiles indicated that, during second fortnight of February wheat was grain filling and grain-milk stage and potato was at maximum vegetative stage with tuber maturity and at the same time other crops namely cotton, castor and fennel were at the maturity stages, have quite distinct NDVI values. During this period major winter crops with different growth stages in this study area will have quite distinct spectral behaviour which indicates that the period of second fortnight of February is quite appropriate and unique for discriminating these winter crops using single date satellite data.

A two-step mapping of irrigated corn with multi-temporal MODIS and Landsat analysis ready data

Timely and reliable information about irrigated croplands is important for crop water stress analysis and studies of water, energy, and food security. This study mapped irrigated and non-irrigated corn at 30 m resolution for the state of Nebraska using a two-step multi-temporal image classification of MODIS and Landsat Analysis Ready Data (ARD). Starting from the drought year of 2012, when there was a high contrast between irrigated and non-irrigated fields, we first conducted image classification using the 250 m MODIS multi-temporal NDVI data. Training pixels were automatically derived, based on counties with predominant irrigated and non-irrigated cornfields. The MODIS-derived irrigated vs. non-irrigated map was further spatially filtered to generate training data covering the entire Nebraska to support automated Landsat ARD classification, footprint-by-footprint. Three classification algorithms of multi-layer perceptron (MLP) neural network, Random Forest (RF), and Support Vector Machine (SVM) were implemented to classify all available Landsat ARD images within the growing season (i.e. May to November). Given the issues of scanline corrector (SLC) error and cloud contamination, the provisional Landsat-based classifications were finally gap-filled to generate a seamless state-wise irrigation map guided by decreasing cross-validation accuracy. Pixel-wise accuracy assessments showed similar overall accuracies of 89.6%, 89.3%, and 90.0% for MLP, RF, and SVM, respectively. They are 3-6% higher than a commonly used gap-filing procedure based on valid (cloud free) pixel count for growing season images. The estimated areas of irrigated corn from Landsat-based mapping were consistent with the 2012 USDA county level census data (R2 = 0.97 and RMSE = 37.70 km2). Using the 2012 Landsat-derived irrigation map and the USDA’s annual Cropland Data Layer as inputs, we further developed training data for annual irrigation mapping between 2013 and 2018. Pixel-wise assessment of the 2016 map showed reasonable overall accuracies of 78.4–79.6% for three classification algorithms. The annual maps yielded R2 of 0.94–0.98 and RMSE values of 37.70–57.62 km2 for various mapping years compared with USDA county statistics. These results suggest that our proposed two-step analytical method has a high potential for automated annual irrigation mapping at 30 m spatial resolution (especially for the arid and semi-arid western U.S.), providing clear field boundaries and irrigation frequency information that are vitally important for accurate agricultural water use analysis.

Possibilities of “Vega” satellite monitoring services for arable land use assessment on the example of Smolensk region, Russia

The paper shows the practical use possibilities of satellite services “Vega” for arable land use determining. The services provide access to satellite data and different tools for its processing for solving this problem. The article shows examples of agricultural land use types interpretation using different approaches such as multi-temporal RGB composites analysis and NDVI vegetation index dynamics analysis. The results of object-based classification of arable land use assessment are demonstrated in the paper on the example of Safonovsky District (Smolensk region, Russia). It is shown that the reliability of the proposed approach reaches more than 90%.

Remote sensing-based framework to predict and assess the interannual variability of maize yields in Pakistan using Landsat imagery

Predicting crop yields and their spatio-temporal variability under a changing climate is a challenging but essential undertaking for crop management and policymaking purposes. The availability of information on risks associated with effects of climatic variability on agricultural activity outcomes is critical for stakeholders ranging from individual landowners to national economists alike. This research was conducted as a pilot study to (1) develop satellite remote sensing based estimates of maize acreage in a typical Maize growing region in Pakistan, (2) to develop a statistical-empirical model for prediction of maize yields, and finally, (3) to assess the influence of temperature on inter-annual variability in maize yields across a decade. A total of eight machine learning algorithms were tested for identifying maize growing operations in the Faisalabad district of Pakistan using Landsat 8 imagery. Classification models were evaluated via 200 randomly selected ground-verified points across the study region. Results of the maize mapping exercise were used to estimate interannual maize yields using Landsat-derived multi-temporal normalized difference vegetation index (NDVI) and land surface temperature (LST) data as predictors. Predictors for the yield forecasting model were selected via principal component screening and were fed into a least absolute shrinkage and selection (LASSO) regression model. The yield model thus developed was applied to 10 years of past data (2006-2017) and validated against data recorded by government sources. Finally, predictions spanning the ten years were tested for effects of temperature variability to find evidence of influence of ambient temperature on maize yields. Results indicate that support vector machine classifiers work the best in this landscape (accuracies >90%) and reveal that maize cropping area may be underestimated in government sources by as much as 14%. The LASSO regression models also showed very good fits (validation R2 = 0.95) and were fairly accurate in tracking interannual variations in maize yields (R2 = 0.78.) Results also indicate that the maximum temperature has significant negative influence (R2 = 0.76, P < 0.0001) on maize yields in Faisalabad district. Methods presented in this study should be of use to policymakers for better formulating export-import policies and decisions governing food security issues in the larger region.