Towards long-multitemporal change detection using SVI differencing by integrated DWT–ISOCLUS: a model for forest temporal dynamics mapping

Abstract

Characterisation and mapping of land cover/land use within forest areas over long-multitemporal intervals is a complex task. This complexity is mainly due to the location and extent of such areas and, as a consequence, to the lack of full continuous cloud-free coverage of those large regions by one single remote sensing instrument. In order to provide improved long-multitemporal forest change detection using Landsat MSS and ETM + in part of Mt. Kenya rainforest, and to develop a model for forest change monitoring, wavelet transforms analysis was tested against the ISOCLUS algorithm for the derivation of changes in natural forest cover, as determined using four simple ratio-based Vegetation Indices: Simple Ratio (SR), Normalised Difference Vegetation Index (NDVI), Renormalised Difference Vegetation Index (RDVI) and modified simple ratio (MSR). Based on statistical and empirical accuracy assessments, RDVI presented the optimal index for the case study. The overall accuracy statistic of the wavelet derived change/no-change was used to rank the performances of the indices as: RDVI (91.68%), MSR (82.55%), NDVI (79.73%) and SR (65.34%). The integrated discrete wavelet transform–ISOCLUS (DWT–ISOCLUS) result was 42.65% higher than the independent ISOCLUS approach in mapping the change/no-change information. The methodology suggested in this study presents a cost-effective and practical method to detect land-cover changes in support of decision-making for updating forest databases, and for long-term monitoring of vegetation changes from multisensor imagery. The current research contributes to Digital Earth with regards to geo-data acquisition, data mining and representation of one forest systems.     

Identification and analysis of groundwater potential zones in Ken–Betwa river linking area using remote sensing and geographic information system

The use of remote sensing data with other ancillary data in a geographic information system (GIS) environment is useful to delineate groundwater potential zonation map of Ken–Betwa river linking area of Bundelkhand. Various themes of information such as geomorphology, land use/land cover, lineament extracted from digital processing of Landsat (ETM+) satellite data of the year 2005 and drainage map were extracted from survey of India topographic sheets, and elevation, slope data were generated from shuttle radar topography mission (SRTM) digital elevation model (DEM). These themes were overlaid to generate groundwater potential zonation (GWPZ) map of the area. The final map of the area shows different zones of groundwater prospects, viz., good (5.22% of the area), moderate (65.83% of the area) poor (15.31% of the area) and very poor (13.64% of area).     

Usefulness of the temporal analysis and the normalized difference in the study of rice by means of landsat‐5 TM images: Identification and inventory of rice fields

The multispectral and multitemporal analysis of the spectral response of rice has made it possible to determine at which point in the vegetative cycle of rice it is best to make an inventory, together with the usefulness of the normalized‐difference vegetation index for such an inventory. The results could be usefully included in any classification procedure of the TM image in order to make the inventory in a systematic way. In this case a supervised classification of the image has been made which assumes a Gaussian behaviour for each spectral class. The results obtained are, for the most part, consistent with those obtained by using traditional methods.     

Understanding long-term (1982–2013) patterns and trends in winter wheat spring green-up date over the North China Plain

Monitoring the spring green-up date (GUD) has grown in importance for crop management and food security. However, most satellite-based GUD models are associated with a high degree of uncertainty when applied to croplands. In this study, we introduced an improved GUD algorithm to extract GUD data for 32 years (1982–2013) for the winter wheat croplands on the North China Plain (NCP), using the third-generation normalized difference vegetation index form Global Inventory Modeling and Mapping Studies (GIMMS3g NDVI). The spatial and temporal variations in GUD with the effects of the pre-season climate and soil moisture conditions on GUD were comprehensively investigated. Our results showed that a higher correlation coefficient (r = 0.44, p < 0.01) and lower root mean square error (22 days) and bias (16 days) were observed in GUD from the improved algorithm relative to GUD from the MCD12Q2 phenology product. In spatial terms, GUD increased from the southwest (less than day of year (DOY) 60) to the northeast (more than DOY 90) of the NCP, which corresponded to spatial reductions in temperature and precipitation. GUD advanced in most (78%) of the winter wheat area on the NCP, with significant advances in 37.8% of the area (p < 0.05). GUD occurred later at high altitudes and in coastal areas than in inland areas. At the interannual scale, the average GUD advanced from DOY 76.9 in the 1980s (average 1982–1989) to DOY 73.2 in the 1990s (average 1991–1999), and to DOY 70.3 after 2000 (average 2000–2013), indicating an average advance of 1.8 days/decade (r = 0.35, p < 0.05). Although GUD is mainly controlled by the pre-season temperature, our findings underline that the effect of the pre-season soil moisture on GUD should also be considered. The improved GUD algorithm and satellite-based long-term GUD data are helpful for improving the representation of GUD in terrestrial ecosystem models and enhancing crop management efficiency.     

Grassland habitat mapping by intra-annual time series analysis – Comparison of RapidEye and TerraSAR-X satellite data

Remote sensing concepts are needed to monitor open landscape habitats for environmental change and biodiversity loss. However, existing operational approaches are limited to the monitoring of European dry heaths only. They need to be extended to further habitats. Thus far, reported studies lack the exploitation of intra-annual time series of high spatial resolution data to take advantage of the vegetations’ phenological differences. In this study, we investigated the usefulness of such data to classify grassland habitats in a nature reserve area in northeastern Germany. Intra-annual time series of 21 observations were used, acquired by a multi-spectral (RapidEye) and a synthetic aperture radar (TerraSAR-X) satellite system, to differentiate seven grassland classes using a Support Vector Machine classifier. The classification accuracy was evaluated and compared with respect to the sensor type – multi-spectral or radar – and the number of acquisitions needed. Our results showed that very dense time series allowed for very high accuracy classifications (>90%) of small scale vegetation types. The classification for TerraSAR-X obtained similar accuracy as compared to RapidEye although distinctly more acquisitions were needed. This study introduces a new approach to enable the monitoring of small-scale grassland habitats and gives an estimate of the amount of data required for operational surveys.     

Spatial and temporal variations of spatial population accessibility to public hospitals: a case study of rural–urban comparison

Quantification and assessment of nationwide population access to health-care services is a critical undertaking for improving population health and optimizing the performance of national health systems. Rural–urban unbalance of population access to health-care services is widely involved in most of the nations. This unbalance is also potentially affected by varied weather and road conditions. This study investigates the rural and urban performances of public health system by quantifying the spatiotemporal variations of accessibility and assessing the impacts of potential factors. Australian health-care system is used as a case study for the rural–urban comparison of population accessibility. A nationwide travel time-based modified kernel density two-step floating catchment area (MKD2SFCA) model is utilized to compute accessibility of travel time within 30, 60, 120, and 240 min to all public hospitals, hospitals that provide emergency care, and hospitals that provide surgery service, respectively. Results show that accessibility is varied both temporally and spatially, and the rural–urban unbalance is distinct for different types of hospitals. In Australia, from the perspective of spatial distributions of health-care resources, spatial accessibility to all public hospitals in remote and very remote areas is not lower (and may even higher) than that in major cities, but the accessibility to hospitals that provide emergency and surgery services is much higher in major cities than other areas. From the angle of temporal variation of accessibility to public hospitals, reduction of traffic speed is 1.00–3.57% due to precipitation and heavy rain, but it leads to 18–23% and 31–50% of reduction of accessibility in hot-spot and cold-spot regions, respectively, and the impact is severe in New South Wales, Queensland, and Northern Territory during wet seasons. Spatiotemporal analysis for the variations of accessibility can provide quantitative and accurate evidence for geographically local and dynamic strategies of allocation decision-making of medical resources and optimizing health-care systems both locally and nationally.     

Topographic effects by the Stokes–Helmert method of geoid and quasi-geoid determinations

The topographic potential and the direct topographic effect on the geoid are presented as surface integrals, and the direct gravity effect is derived as a rigorous surface integral on the unit sphere. By Taylor-expanding the integrals at sea level with respect to topographic elevation (H) the power series of the effects is derived to arbitrary orders. This study is primarily limited to terms of order H ². The limitations of the various effects in the frequently used planar approximations are demonstrated. In contrast, it is shown that the spherical approximation to power H ² leads to a combined topographic effect on the geoid (direct plus indirect effect) proportional to H˜² (where terms of degrees 0 and 1 are missing) of the order of several metres, while the combined topographic effect on the height anomaly vanishes, implying that current frequent efforts to determine the direct effect to this order are not needed. The last result is in total agreement with Bjerhammar's method in physical geodesy. It is shown that the most frequently applied remove–restore technique of topographic masses in the application of Stokes' formula suffers from significant errors both in the terrain correction C (representing the sum of the direct topographic effect on gravity anomaly and the effect of continuing the anomaly to sea level) and in the term t (mainly representing the indirect effect on the geoidal or quasi-geoidal height). Received: 18 August 1998 / Accepted: 4 October 1999     

UAV for mapping shrubland vegetation: Does fusion of spectral and vertical information derived from a single sensor increase the classification accuracy?

High quality data on plant species occurrence count among the essential data sources for ecological research and conservation purposes. Ecologically valuable small grain mosaics of heterogeneous shrub and herbaceous formations however pose a challenging environment for creating such species occurrence maps. Remote sensing can be useful for such purposes, it however faces several challenges, especially the need of ultra high spatial resolution (centimeters) data and distinguishing between plant species or genera. Unmanned aerial vehicles (UAVs) are capable of producing data with sufficient resolution; their use for identification of plant species is however still largely unexplored. A fusion of spectral data with LiDAR-derived vertical information can improve the classification accuracy, such a solution is however costly. A cheaper alternative of vertical data acquisition can be represented by the use of the structure-from-motion photogrammetry (SfM) utilizing the images taken for (multi/hyper)spectral analysis. We investigated the use of such a fusion of UAV-borne multispectral and SfM-derived vertical information acquired from a single sensor for classification of shrubland vegetation at species level and compared its accuracy with that derived from multispectral information only. Multispectral images were acquired using Tetracam Micro-MCA6 camera in the west of Czechia in a shrubland landscape protected within the NATURA 2000 network. Using (i) multispectral imagery only and (ii) multispectral-SfM fusion, we classified the vegetation into six classes representing four woody plant species and two meadow types. Our results prove that the multispectral-SfM fusion performs significantly better than multispectral only (88.2% overall accuracy, 85.2% mean producer’s accuracy and 85.7% mean user’s accuracy for fusion instead of 73.3%, 75.1% and 63.7%, respectively, for multispectral). We concluded that the fusion of multispectral and SfM information acquired from a single UAV sensor is a viable method for shrub species mapping.     

Landuse/land cover mapping and change detection using satellite data — a case study of dehlon block, district Ludhiana, Punjab

The land use information collected for Dehlon block of Ludhiana district, Punjab from the analysis of the IRS-1B LISS-II data for the year 1993 and IRS PAN data for the year 1997 and SOI topographical maps for 1964 revealed a large change in the area of different land use categories during the period from 1964 to 1997. The agricultural land covering an area of about 94.14 per cent in 1964 reduced to 90.26 per cent in 1997. while the area under rural settlements increased from 312 ha in 1964 to 1162 ha in 1997. An extra area of about 169 ha under waste land was added during the period under study making total waste land area to about 400 ha in 1997. However, the block lacks the forest cover of the required limit. Considerable change in living environment was observed in the block. Number of persons per unit settlement area (ha) being 213.3 in 1964 reduced to 97.1 in 1991; it indicate that the living standard of the people of the block has improved with the changed cropping pattern and increased agricultural production during the period from 1964 to 1991.     

Spatial–temporal characteristic of BDS phase delays and PPP ambiguity resolution with GEO/IGSO/MEO satellites

GPS precise point positioning (PPP) ambiguity resolution (AR) can improve the positioning accuracy and shorten the convergence time. However, for the BeiDou Satellite Navigation System (BDS), the problems of satellite-induced code bias, imperfections in the error models and the inadequate accuracy of orbit products limit the applications of the BDS PPP AR system, which requires more than 6 h to achieve the first ambiguity-fixed solution. In this study, the accuracy of a wide-lane (WL) uncalibrated phase delay (UPD) is improved after careful consideration of the code bias and multipath. Meanwhile, the accuracy of the BDS float ambiguity is also improved by multi-GNSS fusion and improved precise orbit and clock products, which are critical for high-quality narrow-lane (NL) UPD estimations. With three tracking networks of different scales, including Hong Kong, the Crustal Movement Observation Network of China (CMONOC) and the multi-GNSS experiment (MGEX) networks, the spatial–temporal characteristics of WL and NL UPDs for BDS GEO/IGSO/MEO satellites are analyzed, and the PPP AR is performed. Numerous results show that WL and NL UPDs with a standard deviation (STD) of less than 0.15 cycles can be achieved for BDS GEO satellites, while a STD of less than 0.1 cycles can be obtained for IGSO and MEO satellites. With the precise UPD estimation, for the first time, the BDS PPP rapid ambiguity resolution for GEO/IGSO/MEO satellites is achieved. We found that the average time to first fix (TTFF) of the BDS PPP AR is shortened significantly, to approximately 40 min for Hong Kong and the CMONOC, while the TTFF was 57.4 min for the MGEX networks. With ambiguity resolution, the accuracy of the daily BDS PPP in the east, north and vertical directions improves from 1.74 cm, 1.08 cm, and 5.52 cm to 0.72 cm, 0.54 cm, and 3.21 cm for the Hong Kong network, 2.24 cm, 2.31 cm, and 5.64 cm to 1.18 cm, 0.79 cm, and 3.30 cm for the CMONOC, and 2.71 cm, 1.80 cm, and 6.00 cm to 1.58 cm, 1.15 cm, and 4.33 cm for the MGEX networks. Significant improvement is also achieved for kinematic PPP, with improvements of 40.41%, 34.33% and 37.17% in the east, north and vertical directions for the MGEX networks, respectively.     

Numerical modeling and prediction of future response of permafrost to different climate change scenarios on the Qinghai–Tibet Plateau

Projecting the future distribution of permafrost under different climate change scenarios is essential, especially for the Qinghai–Tibet Plateau (QTP). The altitude-response model is used to estimate future permafrost changes on the QTP for the four RCPs (RCP2.6, RCP4.5, RCP6.0, and RCP8.5). The simulation results show the following: (1) from now until 2070, the permafrost will experience different degrees of significant degradation under the four RCP scenarios. This will affect 25.68%, 40.54%, 45.95%, and 62.84% of the current permafrost area, respectively. (2) The permafrost changes occur at different rates during the periods 2030–2050 and 2050–2070 for the four different RCPs. (1) In RCP2.6, the permafrost area decreases a little during the period 2030–2050 but shows a small increase from 2050 to 2070. (2) In RCP4.5, the rate of permafrost loss during the period 2030–2050 (about 12.73%) is higher than between 2050 and 2070 (about 8.33%). (3) In RCP6.0, the permafrost loss rate for the period 2030–2050 (about 16.52%) is similar to that for 2050–2070 (about 16.67%). (4) In RCP8.5, there is a significant discrepancy in the rate of permafrost decrease for the periods 2030–2050 and 2050–2070: the rate is only about 3.70% for the first period but about 29.49% during the second.     

Spatial–temporal variations of water vapor content over Ethiopia: a study using GPS observations and the ECMWF model

GPS Solutions - We characterize the spatial–temporal variability of integrated water vapor (IWV) in Ethiopia from a network of global positioning system (GPS) stations and the European Center...     

Comparison of variability and change rate in tropospheric NO2 column obtained from satellite products across China during 1997–2015

ABSTRACT

Tropospheric NO₂ column (TNC) products retrieved from five satellites including GOME/ERS-2 (H, 1997–2002), SCIAMACHY (S, 2003–2011), OMI (O, 2005–2015), GOME-2/METOP_A (A, 2007–2013) and GOME-2/METOP_B (B, 2013–2015) were compared in terms of their spatiotemporal variability and changes over China. The temporal series of H suggested an increasing trend of TNC from 1997 to 2002, those of S, O and A revealed further increasing trends until the highest level of TNC was reached in 2011, but decreasing trends were detected by those of O and B from 2011 to 2015. Seasonally, TNC was the highest in winter and the lowest in summer. Variability and changes from satellite TNC products are also analyzed in different regions of China. Spatially, it was the highest in North China and the lowest in Tibetan Plateau based on five datasets. Overall, TNCs from A, B and S were higher than that from O; and TNC from S was larger than that from A at the country level. The higher TNC the region has, the larger difference satellite products would show. However, different datasets reached a good agreement in the spatial pattern of trends in TNC with highly significant increasing trends detected in North China.     

The nature and possible causes of land cover change (1984 –1996) along a rainfall gradient in southeastern Botswana

Land cover change analysis was undertaken in semi-arid southeastern Botswana. The aim was to determine how remotely sensed data could be applied over time and under different rainfall regimes to help assess the relative significance of biophysical and human factors in causing land cover change in a rapidly evolving developing world context. To this purpose, land cover changes were studied along an east (hardveld)-west (sandveld) gradient of decreasing rainfall and decreasing population density. Three years of Thematic Mapper imagery from 1984, 1994 and 1996, covering the period from the 1980s drought to the 1990s ‘normal’ rainfall regime were analysed using supervised classification techniques. Land cover change analysis revealed that over a large part of the study area the dry and more biophysically vulnerable western sandveld showed greater vegetation recovery than the eastern hardveld with its more productive soils and higher rainfall. Underlying causes behind this apparent reversal of trends are inferred to be mainly socioeconomic in nature and particularly related to higher population density due to the rise of salaried urban occupation opportunities in the hardveld. This work concludes that, while biophysical causes of change are important, the human dimension is regarded as being more significant especially where human factors negate otherwise positive biophysical effects in an agrarian developing country.     

Fine-scale simulation of spatial population distribution by coupling GA–ABM and big data: A case study of Dongguan,China

Capturing spatial population distribution can offer useful information for urban planning to promote reasonable population distribution and allocate urban resource. Agent-based model (ABM) based on the modeling idea of “bottom-up” can offer the ability to simulate the complex individual behaviors that generate spatial population distribution. Previous ABMs were unable to be extended for simulation of spatial population distribution at a fine scale due to the shortage of fine characterization of the urban environment and the calibration of agents' behavior. This study filled these gaps by proposing a genetic algorithm-ABM (GA–ABM) for fine-scale simulation of spatial population distribution in a manufacturing metropolis. In this model, the employment and residential choice behaviors of agents were defined by the labor economic theory and discrete selection model. Multisource geospatial big data such as enterprise points-of-interest big data and building footprints data were used to finely characterize the labor market and urban environment to reflect the impact of agents' employment choices on their residential decision. Furthermore, the grid-scale population investigation big data were combined with the GA to calibrate the agents' residential decision behaviors. The proposed model was used in Dongguan, the typical manufacturing metropolis in China. As a comparison, the expert-experience-based method-ABM (EEBM–ABM) was also conducted by using the same data set. Through the comparison of the results produced by these two models, it was demonstrated that the model coefficient calibrated by GA could effectively reflect the agents' residential decisions. The calibrated GA–ABM is more capable than EEBM–ABM in simulating spatial population distribution in a manufacturing metropolis. Hence, the proposed model can be used to simulate spatial population distribution in a manufacturing metropolis which helps the urban planner to conduct scientific urban planning.     

Airborne remote sensing of spatiotemporal change (1955–2004) in indigenous and exotic forest cover in the Taita Hills,Kenya

We studied changes in area and species composition of six indigenous forest fragments in the Taita Hills, Kenya using 1955 and 1995 aerial photography with 2004 airborne digital camera mosaics. The study area is part of Eastern Arc Mountains, a global biodiversity hot spot that boasts an outstanding diversity of flora and fauna and a high level of endemism. While a total of 260 ha (50%) of indigenous tropical cloud forest was lost to agriculture and bushland between 1955 and 2004, large-scale planting of exotic pines, eucalyptus, grevillea, black wattle and cypress on barren land during the same period resulted in a balanced total forest area. In the Taita Hills, like in other Afrotropical forests, indigenous forest loss may adversely affect ecosystem services.     

Prediction of UT1–UTC,LOD and AAM χ<Subscript>3</Subscript> by combination of least-squares and multivariate stochastic methods

This article presents the application of a multivariate prediction technique for predicting universal time (UT1–UTC), length of day (LOD) and the axial component of atmospheric angular momentum (AAM χ ₃). The multivariate predictions of LOD and UT1–UTC are generated by means of the combination of (1) least-squares (LS) extrapolation of models for annual, semiannual, 18.6-year, 9.3-year oscillations and for the linear trend, and (2) multivariate autoregressive (MAR) stochastic prediction of LS residuals (LS + MAR). The MAR technique enables the use of the AAM χ ₃ time-series as the explanatory variable for the computation of LOD or UT1–UTC predictions. In order to evaluate the performance of this approach, two other prediction schemes are also applied: (1) LS extrapolation, (2) combination of LS extrapolation and univariate autoregressive (AR) prediction of LS residuals (LS + AR). The multivariate predictions of AAM χ ₃ data, however, are computed as a combination of the extrapolation of the LS model for annual and semiannual oscillations and the LS + MAR. The AAM χ ₃ predictions are also compared with LS extrapolation and LS + AR prediction. It is shown that the predictions of LOD and UT1–UTC based on LS + MAR taking into account the axial component of AAM are more accurate than the predictions of LOD and UT1–UTC based on LS extrapolation or on LS + AR. In particular, the UT1–UTC predictions based on LS + MAR during El Niño/La Niña events exhibit considerably smaller prediction errors than those calculated by means of LS or LS + AR. The AAM χ ₃ time-series is predicted using LS + MAR with higher accuracy than applying LS extrapolation itself in the case of medium-term predictions (up to 100 days in the future). However, the predictions of AAM χ ₃ reveal the best accuracy for LS + AR.     

Using Landsat Thematic Mapper records to map land cover change and the impacts of reforestation programmes in the borderlands of southeast Yunnan,China: 1990–2010

At the beginning of the new millennium, after a severe drought and destructive floods along the Yangtze River, the Chinese government implemented two large ecological rehabilitation and reforestation projects: the Natural Forest Protection Programme and the Sloping Land Conversion Programme. Using Landsat data from a decade before, during and after the inception of these programmes, we analyze their impacts along with other policies on land use, land cover change (LULCC) in southwest China. Our goal is to quantify the predominant land cover changes in four borderland counties, home to tens of thousands of ethnic minority individuals. We do this in three time stages (1990, 2000 and 2010). We use support vector machines as well as a transition matrix to monitor the land cover changes. The land cover classifications resulted in an overall accuracy and Kappa coefficient for forested area and cropland of respectively 91% (2% confidence interval) and 0.87. Our results suggest that the total forested area observed increased 3% over this 20-year period, while cropland decreased slightly (0.1%). However, these changes varied over specific time periods: forested area decreased between 1990 and 2000 and then increased between 2000 and 2010. In contrast, cropland increased and then decreased. These results suggest the important impacts of reforestation programmes that have accelerated a land cover transition in this region. We also found large changes in LULC occurring around fast growing urban areas, with changes in these peri-urban zones occurring faster to the east than west. This suggests that differences in socioeconomic conditions and specific local and regional policies have influenced the rates of forest, cropland and urban net changes, disturbances and net transitions. While it appears that a combination of economic growth and forest protection in this region over the past 20 years has been fairly successful, threats like drought, other extreme weather events and land degradation remain.     

Dynamics of carbon and CO<Subscript>2</Subscript> removals by Brazilian forest plantations during 1990–2016

Background

We analyzed the dynamics of carbon (C) stocks and CO₂ removals by Brazilian forest plantations over the period 1990–2016. Data on the extent of forests compiled from various sources were used in the calculations. Productivities were simulated using species-specific growth and yield simulators for the main trees species planted in the country. Biomass expansion factors, root-to-shoot ratios, wood densities, and carbon fractions compiled from literature were applied. C stocks in necromass (deadwood and litter) and harvested wood products (HWP) were also included in the calculations.

Results

Plantation forests stocked 231 Mt C in 1990 increasing to 612 Mt C in 2016 due to an increase in plantation area and higher productivity of the stands during the 26-year period. Eucalyptus contributed 58% of the C stock in 1990 and 71% in 2016 due to a remarkable increase in plantation area and productivity. Pinus reduced its proportion of the carbon storage due to its low growth in area, while the other species shared less than 6% of the C stocks during the period of study. Aboveground biomass, belowground biomass and necromass shared 71, 12, and 5% of the total C stocked in plantations in 2016, respectively. HWP stocked 76 Mt C in the period, which represents 12% of the total C stocked. Carbon dioxide removals by Brazilian forest plantations during the 26-year period totaled 1669 Gt CO_2-e.

Conclusions

The carbon dioxide removed by Brazilian forest plantations over the 26 years represent almost the totality of the country´s emissions from the waste sector within the same period, or from the agriculture, forestry and other land use sector in 2016. We concluded that forest plantations play an important role in mitigating GHG (greenhouse gases) emissions in Brazil. This study is helpful to improve national reporting on plantation forests and their GHG sequestration potential, and to achieve Brazil’s Nationally Determined Contribution and the Paris Agreement.

     

Spectral matching techniques (SMTs) and automated cropland classification algorithms (ACCAs) for mapping croplands of Australia using MODIS 250-m time-series (2000–2015) data

ABSTRACT

Mapping croplands, including fallow areas, are an important measure to determine the quantity of food that is produced, where they are produced, and when they are produced (e.g. seasonality). Furthermore, croplands are known as water guzzlers by consuming anywhere between 70% and 90% of all human water use globally. Given these facts and the increase in global population to nearly 10 billion by the year 2050, the need for routine, rapid, and automated cropland mapping year-after-year and/or season-after-season is of great importance. The overarching goal of this study was to generate standard and routine cropland products, year-after-year, over very large areas through the use of two novel methods: (a) quantitative spectral matching techniques (QSMTs) applied at continental level and (b) rule-based Automated Cropland Classification Algorithm (ACCA) with the ability to hind-cast, now-cast, and future-cast. Australia was chosen for the study given its extensive croplands, rich history of agriculture, and yet nonexistent routine yearly generated cropland products using multi-temporal remote sensing. This research produced three distinct cropland products using Moderate Resolution Imaging Spectroradiometer (MODIS) 250-m normalized difference vegetation index 16-day composite time-series data for 16 years: 2000 through 2015. The products consisted of: (1) cropland extent/areas versus cropland fallow areas, (2) irrigated versus rainfed croplands, and (3) cropping intensities: single, double, and continuous cropping. An accurate reference cropland product (RCP) for the year 2014 (RCP2014) produced using QSMT was used as a knowledge base to train and develop the ACCA algorithm that was then applied to the MODIS time-series data for the years 2000–2015. A comparison between the ACCA-derived cropland products (ACPs) for the year 2014 (ACP2014) versus RCP2014 provided an overall agreement of 89.4% (kappa?=?0.814) with six classes: (a) producer’s accuracies varying between 72% and 90% and (b) user’s accuracies varying between 79% and 90%. ACPs for the individual years 2000–2013 and 2015 (ACP2000–ACP2013, ACP2015) showed very strong similarities with several other studies. The extent and vigor of the Australian croplands versus cropland fallows were accurately captured by the ACCA algorithm for the years 2000–2015, thus highlighting the value of the study in food security analysis. The ACCA algorithm and the cropland products are released through http://croplands.org/app/map and http://geography.wr.usgs.gov/science/croplands/algorithms/australia_250m.html