Search

Hail represents a major threat to agriculture in Switzerland, and assessments of current and future hail risk are of paramount importance for decision-making in the insurance industry and the agricultural sector. However, relating observational information on hail with crop-specific damage is challenging. Here, we build and systematically assess an open-source model to predict hail damage footprints for field crops (wheat, maize, barley, rapeseed) and grapevine from the operational radar product Maximum Expected Severe Hail Size (MESHS) at different spatial resolutions. To this end, we combine the radar information with detailed geospatial information on agricultural land use and geo-referenced damage data from a crop insurer for 12 recent hail events in Switzerland. We find that for field crops model skill gradually increases when the spatial resolution is reduced from 1 km down to 8 km. For even lower resolutions, the skill is diminished again. In contrast, for grapevine, decreasing model resolution below 1 km tends to reduce skill, which is attributed to the different spatial distribution of field crops and grapevine in the landscape. It is shown that identifying a suitable MESHS thresholds to model damage footprints always involves trade-offs. For the lowest possible MESHS threshold (20 mm) the model predicts damage about twice as often as observed (high frequency bias and false alarm ratio), but it also has a high probability of detection (80 %). The frequency bias decreases for larger thresholds and reaches an optimal value close to 1 for MESHS thresholds of 30–40 mm. However, this comes at the cost of a substantially lower probability of detection (around 50 %), while overall model skill, as measured by the Heidke skill score (HSS), remains largely unchanged (0.41–0.44). We argue that, ultimately, the best threshold therefore depends on the relative costs of a false alarm versus a missed event. Finally, the frequency of false alarms is substantially reduced and skill is improved (HSS = 0.54) when only areas with high cropland density are considered. Results from this simple, open-source model show that modelling of hail damage footprints to crops from single-polarization radar in Switzerland is skilful and is best done at 8 km resolution for field crops and 1 km for grapevine. [ABSTRACT FROM AUTHOR]

Hail remains a major threat to agriculture in Switzerland and beyond and assessments of current and future hail risk are of paramount importance for decision-making in the insurance industry and the agricultural sector. However, relating observational information on hail with crop-specific damages is challenging. Here, we build and systematically assess a model to predict hail damage footprints for field crops (wheat, maize, barley, rapeseed) and grapevine from the operational radar product Maximum Expected Severe Hail Size (MESHS) at different spatial resolutions. To this end, we combine the radar information with detailed geospatial information on agricultural land use and geo-referenced damage data from a crop insurer for 12 recent hail events in Switzerland. We find that for field crops, model skill gradually increases when the spatial resolution is reduced from 1 km down to 8 km. For even lower resolutions, the skill is diminished again. On the contrary, for grapevine, a lower model resolution tends to reduce skill, which is attributed to the different spatial distribution of field crops and grapevine in the landscape. It is shown that identifying a suitable MESHS thresholds to model damage footprints always involves trade-offs. For the lowest possible MESHS threshold (20 mm) the model predicts damage about two times too often (high frequency bias and number of false alarms) but also has a high probability of detection (80 %). The frequency bias decreases for larger thresholds and reaches an optimal value close to 1 for MESHS thresholds of 30–40 mm. However, this comes at the cost of a substantially lower probability of detection (around 50 %) while overall model skill remains largely unchanged. We argue that, ultimately, the best threshold selection therefore depends on the user need and the costs of a false alarm or a missed event. Finally, the frequency of false alarms can be substantially reduced when only areas with high cropland density are considered. Results from this simple, open-source model show that modelling of hail damage footprints to crops from single-polarization radar in Switzerland is skillful and is best done at 8 km resolution for field crops and 1 km for grapevine. They further allow different users of such models to identify the suitable threshold for their application, taking into account associated trade-offs. [ABSTRACT FROM AUTHOR]

In the semi-arid Peruvian Andes, the growing season is mostly determined by the timing of the onset and retreat of the wet season, to which annual crop yields are highly sensitive. Recently, local farmers in the Rio Santa basin (RSB) reported more erratic rainy season onsets and further challenges related to changes in rainfall characteristics. Previous studies based on local rain gauges, however, did not find any significant long-term rainfall changes, potentially linked to the scarce data basis and inherent difficulties in capturing the highly variable rainfall distribution typical for complex mountain terrain. To date, there remains considerable uncertainty in the RSB regarding changes in plant-available water over the last decades. In this study, we exploit satellite-derived information of high-resolution vegetation greenness as an integrated proxy to derive variability and trends of plant water availability. By combining MODIS Aqua and Terra vegetation indices (VIs), datasets of precipitation (both for 2000–2020) and soil moisture (since 2015), we explore recent spatio-temporal changes in the vegetation growing season. We find the Normalized Difference Vegetation Index (NDVI) to be coupled to soil moisture on a sub-seasonal basis, while NDVI and rainfall only coincide on interannual timescales. Over 20 years, we find significant greening in the RSB, particularly pronounced during the dry season (austral winter), indicating an overall increase in plant-available water over the past 2 decades. The start of the growing season (SOS) exhibits high interannual variability of up to 2 months compared to the end of the growing season (EOS), which varies by up to 1 month, therefore dominating the variability of the growing season length (LOS). The EOS becomes significantly delayed over the analysis period, matching the observed dry-season greening. While both in situ and gridded rainfall datasets show incoherent changes in annual rainfall for the region, Climate Hazards InfraRed Precipitation with Station data (CHIRPS) rainfall suggests significant positive dry-season trends for 2 months coinciding with the most pronounced greening. As the greening signal is strongly seasonal and reaches high altitudes on unglaciated valley slopes, we cannot link this signal to water storage changes on timescales beyond one rainy season, making interannual rainfall variability the most likely driver. Exploring El Niño–Southern Oscillation (ENSO) control on greening, we find an overall increased LOS linked to an earlier SOS in El Niño years, which however cannot explain the observed greening and delayed EOS. While our study could not corroborate anecdotal evidence of recent changes, we confirm that the SOS is highly variable and conclude that rainfed farming in the RSB would profit from future efforts being directed towards improving medium-range forecasts of the rainy season onset. [ABSTRACT FROM AUTHOR]

In the semi-arid Peruvian Andes, the growing season is mostly determined by the timing of the onset and retreat of the wet season, to which annual crop yields are highly sensitive. Recently, local farmers in the Rio Santa basin (RSB) reported decreasing predictability of the onset of the rainy season and further challenges related to changes in rainfall characteristics. Previous studies based on time series of local rain gauges however, did not find any significant changes in either the timing or intensity of the wet season. Both in-situ and satellite rainfall data for the region lack the necessary spatial resolution to capture the highly variable rainfall distribution typical for complex terrain, and are often questionable in terms of quality and temporal consistency. To date, there remains considerable uncertainty in the RSB regarding hydrological changes over the last decades. In this study, we overcome this limitation by exploiting satellite-derived information on vegetation greenness to reveal a robust and highly resolved picture of recent changes in rainfall and vegetation phenology across the region: As the semi-arid climate causes water availability (i.e. precipitation) to be the key limiting factor for plant growth, patterns of precipitation occurrence and the seasonality of vegetation indices (VIs) are tightly coupled. Therefore, VIs can serve as an integrated proxy of rainfall. By combining MODIS Aqua and Terra VIs for 2000-2020 and several datasets of precipitation, we explore recent spatio-temporal changes in vegetation and water availability. Furthermore, we examine their links to El Niño Southern Oscillation (ENSO). While different rainfall datasets tend to be incoherent in the period of observation, we find significant greening over the majority of the RSB domain in VI data, particularly pronounced during the dry season (Austral winter). This indicates an overall increase of plant available water over time. The rainy season onset and consequently the start of the growing season (SOS) exhibits high inter-annual variability and dominates the growing season length (LOS). The end of the growing season (EOS) is significantly delayed in the analysis which matches the observed dry-season greening. By partitioning the results into periods of three stages of ENSO (neutral, Niño, Niña), we find an earlier SOS and an overall increased season length in years associated with El Niño. However, the appearance of Niño/Niña events during the analysed period cannot explain the observed greening and delayed EOS. While our study could not corroborate anecdotal evidence for recent changes in the SOS, we confirm that the SOS is highly variable and conclude that rainfed farming in the RSB would profit from future efforts being directed towards improving medium-range forecasts of the rainy season onset. [ABSTRACT FROM AUTHOR]

Large variability in N2O emissions from managed grasslands may occur because most emissions originate in surface litter or near-surface soil where variability in soil water content (θ) and temperature (Ts) is greatest. To determine whether temporal variability in θ and Ts of surface litter and near-surface soil could explain this in N2O emissions, a simulation experiment was conducted with ecosys, a comprehensive mathematical model of terrestrial ecosystems in which processes governing N2O emissions were represented at high temporal and spatial resolution. Model performance was verified by comparing N2O emissions, CO2 and energy exchange, and θ and Ts modelled by ecosys with those measured by automated chambers, eddy covariance (EC) and soil sensors on an hourly timescale during several emission events from 2004 to 2009 in an intensively managed pasture at Oensingen, Switzerland. Both modelled and measured events were induced by precipitation following harvesting and subsequent fertilizing or manuring. These events were brief (2-5 days) with maximum N2O effluxes that varied from < 1 mgNm-2h-1 in early spring and autumn to > 3 mgNm-2h-1 in summer. Only very small emissions were modelled or measured outside these events. In the model, emissions were generated almost entirely in surface litter or near-surface (0-2 cm) soil, at rates driven by N availability with fertilization vs. N uptake with grassland regrowth and by O2 supply controlled by litter and soil wetting relative to O2 demand from microbial respiration. In the model, NOx availability relative to O2 limitation governed both the reduction of more oxidized electron acceptors to N2O and the reduction of N2O to N2, so that the magnitude of N2O emissions was not simply related to surface and near-surface θ and Ts. Modelled N2O emissions were found to be sensitive to defoliation intensity and timing which controlled plant N uptake and soil θ and Ts prior to and during emission events. Reducing leaf area index (LAI) remaining after defoliation to half that under current practice and delaying harvesting by 5 days raised modelled N2O emissions by as much as 80% during subsequent events and by an average of 43% annually. Modelled N2O emissions were also found to be sensitive to surface soil properties. Increasing near-surface bulk density by 10% raised N2O emissions by as much as 100% during emission events and by an average of 23% annually. Relatively small spatial variation in management practices and soil surface properties could therefore cause the large spatial variation in N2O emissions commonly found in field studies. The global warming potential from annual N2O emissions in this intensively managed grassland largely offset those from net C uptake in both modelled and field experiments. However, model results indicated that this offset could be adversely affected by suboptimal land management and soil properties. [ABSTRACT FROM AUTHOR]