An efficient visual servo tracker for herd monitoring by UAV.

It is a challenging and meaningful task to carry out UAV-based livestock monitoring in high-altitude (more than 4500 m on average) and cold regions (annual average - 4 °C) on the Qinghai Tibet Plateau. The purpose of artificial intelligence (AI) is to execute automated tasks and to solve practical problems in actual applications by combining the software technology with the hardware carrier to create integrated advanced devices. Only in this way, the maximum value of AI could be realized. In this paper, a real-time tracking system with dynamic target tracking ability is proposed. It is developed based on the tracking-by-detection architecture using YOLOv7 and Deep SORT algorithms for target detection and tracking, respectively. In response to the problems encountered in the tracking process of complex and dense scenes, our work (1) Uses optical flow to compensate the Kalman filter, to solve the problem of mismatch between the target bounding box predicted by the Kalman filter (KF) and the input when the target detection in the current frame is complex, thereby improving the prediction accuracy; (2) Using a low confidence trajectory filtering method to reduce false positive trajectories generated by Deep SORT, thereby mitigating the impact of unreliable detection on target tracking. (3) A visual servo controller has been designed for the Unmanned Aerial Vehicle (UAV) to reduce the impact of rapid movement on tracking and ensure that the target is always within the field of view of the UAV camera, thereby achieving automatic tracking tasks. Finally, the system was tested using Tibetan yaks on the Qinghai Tibet Plateau as tracking targets, and the results showed that the system has real-time multi tracking ability and ideal visual servo effect in complex and dense scenes.

The success of ecological engineering projects on vegetation restoration in China strongly depends on climatic conditions.

China has implemented extensive ecological engineering projects (EEPs) during recent decades to restore and enhance ecosystem functioning. However, the effectiveness of these interventions can vary due to factors such as local climate and specific project objectives. Here, we used two independent satellite remote sensing datasets, including the Global Inventory Monitoring and Modeling System (GIMMS) Normalized Difference Vegetation Index (NDVI) and vegetation optical depth from Ku-band (Ku-VOD), to investigate the vegetation trends in two hotspot regions of EEPs characterized by different climate conditions, i.e., the xeric/semi-xeric Loess Plateau and mesic southwest China. We found diverging vegetation greenness/biomass trend shift patterns in these two regions as a result of the combined effects of EEPs and climate variations, as indicated by changes in the Standardized Precipitation Evapotranspiration Index (SPEI). In the Loess Plateau, where no significant climate variations were observed, NDVI/Ku-VOD increased continuously after the implementation of key EEPs in 2000. Conversely, southwest China has experienced persistent drying since 2000, and vegetation greenness/biomass showed an increasing trend during the initial stages of ecological engineering implementation but subsequently reversed towards a decline due to the continued dry climatic conditions. We used the residual trend method to separate the influence of EEPs from climate variations on vegetation trends and found a positive effect of the ecological management practices in the Loess Plateau, yet a predominantly negative effect in the southwest China region, which means that projects implemented in southwest China did not lead to a long-term improvement in vegetation growth under the given climate conditions in southwest China. This adverse impact suggests that ecological engineering practices could potentially increase the ecosystem's vulnerability to droughts, owing to the increased transpirational water demands introduced by ecological engineering interventions. Our study highlights the importance of considering the expected occurrence and magnitude of climatic variability when implementing large-scale EEPs.

Procapra Przewalskii Tracking Autonomous Unmanned Aerial Vehicle Based on Improved Long and Short-Term Memory Kalman Filters.

This paper presents an autonomous unmanned-aerial-vehicle (UAV) tracking system based on an improved long and short-term memory (LSTM) Kalman filter (KF) model. The system can estimate the three-dimensional (3D) attitude and precisely track the target object without manual intervention. Specifically, the YOLOX algorithm is employed to track and recognize the target object, which is then combined with the improved KF model for precise tracking and recognition. In the LSTM-KF model, three different LSTM networks (f, Q, and R) are adopted to model a nonlinear transfer function to enable the model to learn rich and dynamic Kalman components from the data. The experimental results disclose that the improved LSTM-KF model exhibits higher recognition accuracy than the standard LSTM and the independent KF model. It verifies the robustness, effectiveness, and reliability of the autonomous UAV tracking system based on the improved LSTM-KF model in object recognition and tracking and 3D attitude estimation.

Spatiotemporal dynamics of forest ecosystem carbon budget in Guizhou: customisation and application of the CBM-CFS3 model for China.

BACKGROUND: Countries seeking to mitigate climate change through forests require suitable modelling approaches to predict carbon (C) budget dynamics in forests and their responses to disturbance and management. The Carbon Budget Model of the Canadian Forest Sector (CBM-CFS3) is a feasible and comprehensive tool for simulating forest C stock dynamics across broad levels, but discrepancies remain to be addressed in China. Taking Guizhou as the case study, we customised the CBM-CFS3 model according to China's context, including the modification of aboveground biomass C stock algorithm, addition of C budget accounting for bamboo forests, economic forests, and shrub forests, improvement of non-forest land belowground slow dead organic matter (DOM) pool initialisation, and other model settings. RESULTS: The adequate linear relationship between the estimated and measured C densities (R2 = 0.967, P < 0.0001, slope = 0.904) in the model validation demonstrated the high accuracy and reliability of our customised model. We further simulated the spatiotemporal dynamics of forest C stocks and disturbance impacts in Guizhou for the period 1990-2016 using our customised model. Results shows that the total ecosystem C stock and C density, and C stocks in biomass, litter, dead wood, and soil in Guizhou increased continuously and significantly, while the soil C density decreased over the whole period, which could be attributed to deforestation history and climate change. The total ecosystem C stock increased from 1220 Tg C in 1990 to 1684 Tg C in 2016 at a rate of 18 Tg C yr-1, with significant enhancement in most areas, especially in the south and northwest. The total decrease in ecosystem C stock and C expenditure caused by disturbances reached 97.6 Tg C and 120.9 Tg C, respectively, but both represented significant decreasing trends owing to the decline of disturbed forest area during 1990-2016. Regeneration logging, deforestation for agriculture, and harvest logging caused the largest C stock decrease and C expenditure, while afforestation and natural expansion of forest contributed the largest increases in C stock. CONCLUSIONS: The forests in Guizhou were a net carbon sink under large-scale afforestation throughout the study period; Our customised CBM-CFS3 model can serve as a more effective and accurate method for estimating forest C stock and disturbance impacts in China and further enlightens model customisation to other areas.

Cultivation of non-irrigated spring wheat in temperate free-grazing steppe improved both ecosystem and canopy water use efficiency.

The temperate steppe in northern China is important for sandstorm control and food/livestock production. Understanding the influence and regulatory control of cultivation on the water balance and water use efficiency (WUE) of this water-limited region would promote the sustainability of local ecosystem and food supply. This study combined eddy covariance system observational data and the Shuttleworth-Wallace model to investigate evapotranspiration (ET) and its composition in paired sites, including a free-grazing steppe site and an adjacent site reclaimed for spring wheat cultivation in Xilinhot, Inner Mongolia. Further, analysis of the WUE of both the ecosystem (WUEE) and the canopy (WUEC) under the two sites showed that the mean daily gross primary productivity (GPP) of the cultivation site was 3.84 gC·m-2·d-1, i.e., 15.7% higher than that of the free-grazing site (3.32 gC·m-2·d-1). Compared with the free-grazing site (1.76 kgH2O·m-2·d-1), the mean daily ET of the cultivation site (1.40 kgH2O·m-2·d-1) was reduced by 20.7%. The difference in ET was due mainly to suppression of evaporation at the cultivation site from increased shading associated with a higher leaf area index (LAI). The largely increased GPP of the cultivation site fundamentally contributed to the 54.7% higher WUEC (4.75 gC·kg-1H2O) in comparison with the free-grazing site (3.08 gC·kg-1H2O). The WUEE of the cultivation site was 57.9% higher than that of the free-grazing site. The variation of transpiration of the free-grazing site explained 64% of the change of WUEC. These results indicate that land use differences in the temperate steppe area changed vegetation productivity substantially. Moreover, ecosystem ET and its composition, as well as large-scale land use change, might influence the regional water use pattern and mass balance. Our findings help clarify the impact of typical land use change on regional WUE, and could promote development of visionary and effective strategies for the use of the limited resources in arid-semiarid regions.

Community Composition and Structure Affect Ecosystem and Canopy Water Use Efficiency Across Three Typical Alpine Ecosystems.

Unique ecosystems distributed in alpine areas of the Qinghai-Tibetan Plateau play important roles in climate change mitigation, local food supply, and conservation of species diversity. To understand the water use efficiency (WUE) of this fragile and sensitive region, this study combined observed data from the eddy covariance system and the Shuttleworth-Wallace (S-W) model to measure the continuous mass exchange, including gross primary productivity (GPP), evapotranspiration (ET), and canopy transpiration (T) throughout 2 or 3 years (2016-2018) in three common alpine ecosystems (i.e., alpine steppe, alpine meadow, and alpine swamp). These ecosystems represent a water availability gradient and thus provide the opportunity to quantify environmental and biological controls on WUE at various spatiotemporal scales. We analyzed the ecosystem WUE (WUEe; defined as the ratio of GPP to ET) and canopy WUE (WUEc; defined as the ratio of GPP and canopy T). It was found that the yearly WUEe was 1.40, 1.63, and 2.16 g C kg-1 H2O, and the yearly WUEc was 8.93, 2.46, and 5.19 g C kg-1 H2O in the three typical ecosystems, respectively. The controlling factors of yearly WUE diverged between WUEe and WUEc. We found that plant functional group proportion (e.g., gramineous and Cyperaceae) highly explained the yearly WUEe variation across sites, and a good correlation was observed between community species diversity and WUEc. These findings suggest that community composition and trait change are critical in regulating WUEe and WUEc across different alpine ecosystems and that the regulation mechanisms may differ fundamentally between WUEe and WUEc.

Integrating satellite and unmanned aircraft system (UAS) imagery to model livestock population dynamics in the Longbao Wetland National Nature Reserve, China.

The collection of field-based animal data is laborious, risky and costly in some areas, such as various nature reserves. Although multiple studies have used satellite imagery, aerial imagery, and field data individually for some animal species surveys, several technical issues still need to be addressed before full standardization of remote sensing methods for modeling animal population dynamics over large areas. This study is the first to model the population dynamics of livestock in the Longbao Wetland National Nature Reserve, China by utilizing yak estimations from Worldview-2 satellite imagery (0.5 m) collected in 2010 and yaks counted in a ground-based survey conducted in 2011 in combination with the animal population structure precisely extracted from UAS imagery captured in 2016. As a consequence, 5501, 5357, and 5510 yaks were estimated to appear in the reserve in 2010, 2011 and 2016, respectively. In total, 1092, 1062 and 1092 sheep were estimated to appear in the reserve in 2010, 2011 and 2016, respectively. The uncertainty of the presented method is also discussed. Primary experiments show that both the satellite imagery and UAS imagery are promising for use in yak censuses, but no sheep were observed in the satellite imagery because of the low resolution. Compared to the ground-based survey conducted in 2011, the UAS image estimate and satellite imagery count deviated in yak quantity by 2.69% and 2.86%, respectively. UASs are a reliable and low-budget alternative to animal surveys. No discernable changes in animal behaviors and animal distributions were observed as the UAS passed at a height of 700 m, and the accuracy of UAS imagery counts were not significantly affected by the short-distance animal movement and image mosaicking errors. The experimental results illustrate the advantages of the combination of satellite and UAS imagery in modeling animal population dynamics.

A Population Census of Large Herbivores Based on UAV and Its Effects on Grazing Pressure in the Yellow-River-Source National Park, China.

Using the Yellow-River-Source National Park (YRSNP) as a study site, an unmanned aerial vehicle (UAV) remote sensing and line transect method was used to investigate the number of wild herbivorous animals and livestock, including the kiang (Equus kiang) and Tibetan gazelle (Procapra picticaudata). A downscaling algorithm was used to generate the forage yield data in YRSNP based on a 30-m spatial resolution. On this basis, we estimated the forage-livestock balance, which included both wild animals and livestock, and analyzed the effects of functional zone planning in national parks on the forage-livestock balance in YRSNP. The results showed that the estimates of large herbivore population numbers in YRSNP based on population density in the aerial sample strips, which were compared and validated with official statistics and warm season survey results, indicated that the numbers of kiangs and Tibetan gazelles in the 2017 cold season were 12,900 and 12,100, respectively. The numbers of domestic yaks, Tibetan sheep, and horses were 53,400, 76,800, and 800, respectively, and the total number of sheep units was 353,200. The ratio of large wild herbivores and livestock sheep units was 1:5. Large wild herbivores have different preferences for functional zones, preferring ecological restoration areas consisting mainly of sparse grassland. The grazing pressure indices of the core reserve areas and ecological restoration areas were 0.168 and 0.276, respectively, indicating that these two regions still have high grazing potential. However, the grazing pressure index of the traditional utilization areas was 1.754, indicating that these grasslands are severely overloaded. After the planning and implementation of functional zones, the grazing pressure index of YRSNP was 1.967. Under this measure, the number of livestock was not reduced and the grazing pressure nearly doubled, indicating that forage-livestock conflict has become more severe in YRSNP.

Modeling Aboveground Biomass in Hulunber Grassland Ecosystem by Using Unmanned Aerial Vehicle Discrete Lidar.

Accurate canopy structure datasets, including canopy height and fractional cover, are required to monitor aboveground biomass as well as to provide validation data for satellite remote sensing products. In this study, the ability of an unmanned aerial vehicle (UAV) discrete light detection and ranging (lidar) was investigated for modeling both the canopy height and fractional cover in Hulunber grassland ecosystem. The extracted mean canopy height, maximum canopy height, and fractional cover were used to estimate the aboveground biomass. The influences of flight height on lidar estimates were also analyzed. The main findings are: (1) the lidar-derived mean canopy height is the most reasonable predictor of aboveground biomass (R² = 0.340, root-mean-square error (RMSE) = 81.89 g·m-2, and relative error of 14.1%). The improvement of multiple regressions to the R² and RMSE values is unobvious when adding fractional cover in the regression since the correlation between mean canopy height and fractional cover is high; (2) Flight height has a pronounced effect on the derived fractional cover and details of the lidar data, but the effect is insignificant on the derived canopy height when the flight height is within the range (<100 m). These findings are helpful for modeling stable regressions to estimate grassland biomass using lidar returns.

Comparison of evapotranspiration components and water-use efficiency among different land use patterns of temperate steppe in the Northern China pastoral-farming ecotone.

Water-use efficiency (WUE), which links carbon and water cycles, is an important indicator of assessing the interactions between ecosystems and regional climate. Using chamber methods with and without plant removal treatments, we investigated WUE and evapotranspiration (ET) components in three ecosystems with different land-use types in Northern China pastoral-farming ecotone. In comparison, ET of the ecosystems with grazing exclusion and cultivating was 6.7 and 13.4 % higher than that of the ecosystem with free grazing. The difference in ET was primarily due to the different magnitudes of soil water evaporation (E) rather than canopy transpiration (T). Canopy WUE (WUEc, i.e., the ratio of gross primary productivity to T) at the grazing excluded and cultivated sites was 17 and 36 % higher than that at the grazing site. Ecosystem WUE (WUEnep, i.e., the ratio of net ecosystem productivity to ET) at the cultivated site was 34 and 28 % lower in comparison with grazed and grazing excluded stepped, respectively. The varied leaf area index (LAI) of different land uses was correlated with microclimate and ecosystem vapor/carbon exchange. The LAI changing with land uses should be the primary regulation of grassland WUE. These findings facilitate the mechanistic understanding of carbon-water relationships at canopy and ecosystem levels and projection of the effects of land-use change on regional climate and productivity.

Effects of grassland restoration programs on ecosystems in arid and semiarid China.

We explored the ecological effects of grassland restoration programs using satellite imagery and field plots sampling data and analyzing the patterns and mechanisms of land cover change and vegetation activities in arid and semiarid China during the period from 1982 to 2008. The grassland cover in the 1980s, 2000 and 2005 was compared before and after the restoration programs. The variability of net primary production (NPP) and rain use efficiency (RUE) were analyzed as indicators of vegetation productivity. Our study showed that changes in grassland cover were closely related to the relative area of farmland, with increases in grassland being caused by returning farmland to grassland and decreases being caused by reclamation for agriculture. The results of NPP and RUE measurements over the past 30 years showed systematic increases in the area of grassland in most regions, especially from 2000 to 2008. This fact was reflected by intensified vegetation activity and cannot be completely explained by the warmer and wetter climate, which suggested a contribution from restored, ungrazed grasslands. Our analysis indicates that both vegetation activity and grassland cover increased in regions in which grassland and rangeland restoration programs were implemented.

[Spatiotemporal dynamics of forest carbon storage in Taihe County of Jiangxi Province in 1985-2030].

Based on the sixth forest inventory data of Taihe County, Jiangxi Province, this paper analyzed the curve relations between the carbon densities and ages of major forest types by using Logistic equation, and estimated the total amounts and change trends of the biomass and carbon storage of forest vegetation from 1985 to 2003 by the method of biomass expansion factor. The carbon storage in 2020 and 2030 was estimated by setting 2003 as the baseline year and assuming that the area of forest vegetation remained stable and without consideration of forest rotation. In 2003, the total forest area of Taihe County was 15.74 x 10(4) hm2, the total biomass was 6.71 Tg, the vegetation carbon storage was 4.14 Tg C, and the average carbon density was 26.31 t C x hm(-2). In 1985, 1994, 2003, 2020, and 2030, the forest carbon storage was 1.06, 2.83, 4.14, 5.65, and 6.35 Tg C, respectively. The carbon density of the forest vegetation in Taihe County decreased from the eastern and western regions to the central. Artificial afforestation contributed significantly to the increase of forest stand area, and consequently, to the improvement of forest carbon sequestration capacity.

[Carbon losses from forest fire in Jiangxi Province, China in 1950-2008].

The annual occurrence of forest fire in Jiangxi Province of China in 1950-2008 was averaged 762 times and covering 1.578 x 10(4) hm2. Based on the forest fire statistic data, meteorological data, forest inventory data, and other parameters, this paper analyzed the characters of forest fire in the province, and estimated its carbon release and transfer. In the province, the total forest biomass loss from forest fire in 1950-2008 was about 61. 155 Tg, and the biomass carbon loss was about 30.993 Tg C, accounting for 15.92% of the vegetation carbon pool in the whole province. Before the 1970s, the biomass carbon loss occupied 74.3% of the total loss in 1950-2008, but after the 1990s, this loss decreased rapidly, with an annual average carbon loss less than 0.097 Tg C. The released CO2, CH4, and CO from the forest fire were about 5.408 Tg, 0.047 Tg, and 0.486 Tg, respectively, and led to 22.436 Tg of biomass carbon transferred into soil carbon. The snow and ice disaster in the beginning of 2008 resulted in high frequency forest fire, and led to a 0.463 Tg biomass carbon loss, which was 2. 56 times of the average value (0.181 Tg) in 2003-2007.

Assessment of effects of climate change and grazing activity on grassland yield in the Three Rivers Headwaters Region of Qinghai-Tibet Plateau, China.

Inter-annual dynamics of grassland yield of the Three Rivers Headwaters Region of Qinghai-Tibet Plateau of China in 1988-2005 was analyzed using the GLO-PEM model, and the herbage supply function was evaluated. The results indicate that while grassland yield in the region showed marked inter-annual fluctuation there was a trend of increased yield over the 18 years of the study. This increase was especially marked for Alpine Desert and Alpine Steppe and in the west of the region. The inter-annual coefficient of variation of productivity increased from the east to the west of the region and from Marsh, Alpine Meadow, Alpine Steppe, Temperate Steppe to Alpine Desert grasslands. Climate change, particularly increased temperatures in the region during the study period, is suggested to be the main cause of increased grassland yield. However, reduced grazing pressure and changes to the seasonal pattern of grazing could also have influenced the grassland yield trend. These findings indicate the importance of understanding the function of the grassland ecosystems in the region and the effect of climate change on them especially in regard to their use to supply forage for animal production. Reduction of grazing pressure, especially during winter, is indicated to be critical for the restoration and sustainable use of grassland ecosystems in the region.

[Spatio-temporal problems of geographic information system in marine fishery].

In marine fisheries, it is very important to understand and grasp the spatio-temporal nature. Geographical Information System (GIS) has been applied to describe or forecast the dynamic trend of resources or to set up evaluation model, which is one of high technologies in modern marine fisheries. Based on the review of the development of marine fishery GIS (MFGIS), four spatio-temporal problems it occurred were discussed, and the possible resolutions were prospected.