Methods comparison for detecting trends in herbicide monitoring time-series in streams

An inadvertent consequence of pesticide use is aquatic pesticide pollution, which has prompted the implementation of mitigation measures in many countries. Water quality monitoring programs are an important tool to evaluate the efficacy of these mitigation measures. However, large interannual variability of pesticide losses makes it challenging to detect significant improvements in water quality and to attribute these improvements to the application of specific mitigation measures. Thus, there is a gap in the literature that informs researchers and authorities regarding the number of years of aquatic pesticide monitoring or the effect size (e.g., loss reduction) that is required to detect significant trends in water quality. Our research addresses this issue by combining two exceptional empirical data sets with modelling to explore the relationships between the achieved pesticide reduction levels due to mitigation measures and the length of the observation period for establishing statistically significant trends. Our study includes both a large (Rhine at Basel, ~36,300 km²) and small catchment (Eschibach, 1.2 km²), which represent spatial scales at either end of the spectrum that would be realistic for monitoring programs designed to assess water quality. Our results highlight several requirements in a monitoring program to allow for trend detection. Firstly, sufficient baseline monitoring is required before implementing mitigation measures. Secondly, the availability of pesticide use data helps account for the interannual variability and temporal trends, but such data are usually lacking. Finally, the timing and magnitude of hydrological events relative to pesticide application can obscure the observable effects of mitigation measures (especially in small catchments). Our results indicate that a strong reduction (i.e., 70-90 %) is needed to detect a change within 10 years of monitoring data. The trade-off in applying a more sensitive method for change detection is that it may be more prone to false-positives. Our results suggest that it is important to consider the trade-off between the sensitivity of trend detection and the risk of false positives when selecting an appropriate method and that applying more than one method can provide more confidence in trend detection.

Spatiotemporal variability of extreme temperature indices and their implications over the heterogeneous river basin, India

The current study on spatiotemporal variability of temperature presents a holistic approach for quantifying the joint space-time variability of extreme temperature indices over the physio-climatically heterogeneous Tapi River basin (TRB) using two unsupervised machine learning algorithms, i.e., principal component analysis (PCA) and cluster analysis. The long-term variability in extreme temperature indices, recommended by the Expert Team on Climate Change Detection and Indices (ETCCDI), was evaluated for 1951-2016. The magnitude and statistical significance of the temporal trend in extreme temperature indices were estimated using non-parametric Sen's slope estimator and modified Mann Kendall (MMK) tests, respectively. The multivariate assessment of temporal trends using PCA resulted in four principal components (PCs) encapsulating more than 90% variability. The cluster analysis of corresponding PCs resulted in two spatial clusters exhibiting homogeneous spatiotemporal variability. Cluster 1 is characterized by significantly increasing hottest, very hot, and extremely hot days with rising average maximum temperature and intraday temperature variability. On the other hand, cluster 2 showed significantly rising coldest nights, mean minimum, mean temperature, and Tx37 with significantly decreasing intraday and interannual temperature variability, very cold, and extremely cold nights with reducing cold spell durations. The summertime heat stress computation revealed that the Purna sub-catchment of the Tapi basin is more vulnerable to various health issues and decreased work performance (> 10%) for more than 45 days per year. The current study dealing with the associated effects of rising temperature variability on crop yield, human health, and work performance would help policymakers formulate better planning and management strategies to safeguard society and the environment.

Trend detection with non-detects in long-term monitoring, a mixed model approach

In long-term monitoring of contaminants in biota, a common approach is to use yearly geometric means of measured concentrations in sampled individuals as a basis for trend analysis. When some or all measurements in a particular year are reported as non-detects, it is unclear how to proceed in calculating the yearly mean. I argue that by casting the problem in terms of a mixed model, non-detects can be accounted for using statistical techniques for censored data. The approach is illustrated using data from the Swedish national monitoring programme for contaminants in biota.

Improved mapping of coastal salt marsh habitat change at Barnegat Bay (NJ, USA) using object-based image analysis of high-resolution aerial imagery

Tidal wetlands are valued for the ecosystem services they provide yet are vulnerable to loss due to anthropogenic disturbances such as land conversion, hydrologic modifications, and the impacts of climate change, especially accelerating rates of sea level rise. To effectively manage tidal wetlands in face of multiple stressors, accurate studies of wetland extent and trends based on high-resolution imagery are needed. We provide salt marsh delineations for Barnegat Bay, New Jersey, by means of object-based image analysis of high-resolution aerial imagery and digital elevation models. We performed trends analyses of salt marsh extent from 1995 to 2015 and estimated drivers of marsh area change. We found that in 1995, 8830 ± 390 ha were covered with marsh vegetation, while in 2015 only 8180 ± 380 ha of salt marsh habitat remained. The resulting net loss rate of 0.37% yr-1 is equivalent to historic loss rates since the 1970s, indicating that despite regionally accelerating relative sea level rise and purported eutrophication, salt marsh loss rates at Barnegat Bay remain steady. The main drivers of salt marsh loss are excavations for mosquito control (409 ha), edge erosion (303 ha) and ponding (240 ha). Upland migration of salt marsh did not completely mitigate these losses but accounted for a gain of 147 ha of tidal marsh habitat. The methodology presented herein yielded accurate salt marsh delineations (>90%) and trend detection (85%), outperforming low-resolution wetland delineations used in coastal management. This study demonstrates the suitability of high-resolution imagery for the detection of open water features. For the purposes of salt marsh change detection and the identification of change drivers, management and conservation agencies should make use of high-resolution imagery whenever feasible.

The confounding effect of snow cover on assessing spring phenology from space: A new look at trends on the Tibetan Plateau

The Tibetan Plateau is the highest and largest plateau in the world, hosting unique alpine grassland and having a much higher snow cover than any other region at the same latitude, thus representing a "climate change hot-spot". Land surface phenology characterizes the timing of vegetation seasonality at the per-pixel level using remote sensing systems. The impact of seasonal snow cover variations on land surface phenology has drawn much attention; however, there is still no consensus on how the remote sensing estimated start of season (SOS) is biased by the presence of preseason snow cover. Here, we analyzed SOS assessments from time series of satellite derived vegetation indices and solar-induced chlorophyll fluorescence (SIF) during 2003-2016 for the Tibetan Plateau. We evaluated satellite-based SOS with field observations and gross primary production (GPP) from eddy covariance for both snow-free and snow covered sites. SOS derived from SIF was highly correlated with field data (R² = 0.83) and also the normalized difference phenology index (NDPI) performed well for both snow free (R² = 0.77) and snow covered sites (R² = 0.73). On the contrary, normalized difference vegetation index (NDVI) correlates only weakly with field data (R² = 0.35 for snow free and R² = 0.15 for snow covered sites). We further found that an earlier end of the snow season caused an earlier estimate of SOS for the Tibetan Plateau from NDVI as compared to NDPI. Our research therefore adds new evidence to the ongoing debate supporting the view that the claimed advance in land surface SOS over the Tibetan Plateau is an artifact from snow cover changes. These findings improve our understanding of the impact of snow on land surface phenology in alpine ecosystems, which can further improve remote sensing based land surface phenology assessments in snow-influenced ecosystems.

Of power and despair in cetacean conservation: estimation and detection of trend in abundance with noisy and short time-series

Many conservation instruments rely on detecting and estimating a population decline in a target species to take action. Trend estimation is difficult because of small sample size and relatively large uncertainty in abundance/density estimates of many wild populations of animals. Focusing on cetaceans, we performed a prospective analysis to estimate power, type-I, sign (type-S) and magnitude (type-M) error rates of detecting a decline in short time-series of abundance estimates with different signal-to-noise ratio. We contrasted results from both unregularized (classical) and regularized approaches. The latter allows to incorporate prior information when estimating a trend. Power to detect a statistically significant estimates was in general lower than 80%, except for large declines. The unregularized approach (status quo) had inflated type-I error rates and gave biased (either over- or under-) estimates of a trend. The regularized approach with a weakly-informative prior offered the best trade-off in terms of bias, statistical power, type-I, type-S and type-M error rates and confidence interval coverage. To facilitate timely conservation decisions, we recommend to use the regularized approach with a weakly-informative prior in the detection and estimation of trend with short and noisy time-series of abundance estimates.

Rainfall erosivity in Slovenia: Sensitivity estimation and trend detection

Slovenia is one of the EU countries with the largest values and largest amounts of variability in rainfall erosivity, with maximum annual values exceeding 10,000 MJ mm ha^-1 h^-1 yr^-1. Five-minute rainfall data was analysed from 10 Slovenian rainfall stations with data-length availability longer than 25 years with a maximum data length of 69 years and a total data-station length equal to 443 years. Trends in the rainfall erosivity R-factor were detected for four different sub-samples using monthly, half-year, and annual rainfall erosivity values. The results indicate that rainfall erosivity trends for the selected Slovenian stations are mostly statistically insignificant, with the selected significance level of 0.05. However, a larger share of identified trends are positive than negative. The maximum annual rainfall erosivity values were obtained for one specific mountain station. Furthermore, a sensitivity analysis regarding the rainfall erosivity factor R calculation showed that the rainfall threshold parameter (12.7 mm) that is used to remove the small-magnitude rainfall events in order to reduce the computational burden can attribute up to 10% of the average annual R-values in cases where this threshold is not used. Other parameters have, on average, a smaller impact on the calculated rainfall erosivity. Furthermore, the application of local kinetic energy equations resulted in, on average, about 20% higher annual rainfall erosivity values compared to the equation that is proposed by the Revised Universal Soil Loss Equation (RUSLE) manual and was not developed specifically for this region.

Predicting ROSC in out-of-hospital cardiac arrest using expiratory carbon dioxide concentration: Is trend-detection instead of absolute threshold values the key?

AIM

Guidelines recommend detecting return of spontaneous circulation (ROSC) by a rising concentration of carbon dioxide in the exhalation air. As CO₂ is influenced by numerous factors, no absolute cut-off values of CO₂ to detect ROSC are agreed on so far. As trends in CO₂ might be less affected by influencing factors, we investigated an approach which is based on detecting CO₂-trends in real-time.

METHODS

We conducted a retrospective case-control study on 169 CO₂ time series from out of hospital cardiac arrests resuscitated by Muenster City Ambulance-Service, Germany. A recently developed statistical method for real-time trend-detection (SCARM) was applied to each time series. For each series, the percentage of time points with detected positive and negative trends was determined.

RESULTS

ROSC time series had larger percentages of positive trends than No-ROSC time series (p=0.003). The median percentage of positive trends was 15% in the ROSC time series (IQR: 5% to 23%) and 7% in the No-ROSC time series (IQR: 3% to 14%). A receiver operating characteristic (ROC) analysis yielded an optimal threshold of 13% to differentiate between ROSC and No-ROSC cases with a specificity of 58.4% and sensitivity of 73.9%; the area under the curve was 63.5%.

CONCLUSION

Patients with ROSC differed from patients without ROSC as to the percentage of detected CO₂ trends, indicating the potential of our real-time trend-detection approach. Since the study was designed as a proof of principle and its calculated specificity and sensitivity are low, more research is required to implement CO₂-trend-detection into clinical use.

Decadal and seasonal trends of nutrient concentration and export from highly managed coastal catchments

Understanding anthropogenic and hydro-climatic influences on nutrient concentrations and export from highly managed catchments often necessitates trend detection using long-term monitoring data. This study analyzed the temporal trend (1979-2014) of total nitrogen (TN) and total phosphorus (TP) concentrations and export from four adjacent coastal basins in south Florida where land and water resources are highly managed through an intricate canal network. The method of integrated seasonal-trend decomposition using LOESS (LOcally weighted regrESSion) was employed for trend detection. The results indicated that long-term trends in TN and TP concentrations (increasing/decreasing) varied with basins and nutrient species, reflecting the influence of basin specific land and water management practices. These long-term trends were intervened by short-term highs driven by high rainfall and discharges and lows associated with regional droughts. Seasonal variations in TP were more apparent than for TN. Nutrient export exhibited a chemostatic behavior for TN from all the basins, largely due to the biogenic nature of organic N associated with the ubiquity of organic materials in the managed canal network. Varying degrees of chemodynamic export was present for TP, reflecting complex biogeochemical responses to the legacy of long-term fertilization, low soil P holding capacity, and intensive stormwater management. The anthropogenic and hydro-climatic influences on nutrient concentration and export behavior had great implications in nutrient loading abatement strategies for aquatic ecosystem restoration of the downstream receiving waterbody.