With my work on mapping distribution of plant species I have thematic, spatial as well as temporal uncertainties. It is difficult to know where to draw boundary lines between species as their distribution overlaps. I have to define the criteria by which I make this choice. There is spatial uncertainty in the GPS coordinates due to instrument accuracy which also varies temporally depending on satellite contacts. There is additional temporal uncertainty with seasonal and yearly changes in vegetation cover. It is a dynamic system.
Based on that figure/resources, I found that there has a:
- Spatial uncertainty: how to make a line of reef area. Sometimes in the area, there were may have blank reef (sand) but still filled in the map
Temporal uncertainty: reef rapid change - This inevitable, because health reef needs to identify by human. And human usually can’t precisely describe the area because of the reef area complexity
- No, just due to lack of data
- This may affect the different method for maintaining reef health. Because the map only describes the overall coral reef cover, without know how the health status. Because coral reef not only covered by healthy coral. But also dead coral, DCA, rubble, or algae.
- We can only use it for general purpose. Take the own sampling on the focused location/cover are needed for more accurate result.
When conducting waterbird surveys in wetlands, some uncertainties may arise as a result of changes/impact in climatic conditions. During flooded periods as a result of high periods of rainfall, survey locations become inaccessible. As a result, one is forced to conduct surveys at defined periods. Such an approach may not show a peculiar trend in the activities of waterbirds over time. Unfortunately, due to the inevitability of these uncertainties, we are left with no choice but to document them during reporting.
There is definitely a lot of temporal uncertainty in my samples. Macroinvertebrate populations have been shown to change over time, especially seasonally. I sampled in summer, so my results will not be applicable to winter assemblages. There is also spatial uncertainty, as seagrass beds show patchy distribution so one cannot assume the full extent of the beds.
I basically have the same challenges as Greg describes - seagrass beds are dynamic and change quite rapidly due to sediment shift and seasonal growth patterns (temporal uncertainty). On the high resolution sattelite images I am using, it is also sometimes difficult to decide where to draw the boundaries of the meadows, as there are sparse transition zones as well as a degree of overlap with algal species (thematic uncertainty). As I am limited in the number of water samples I can collect for pollutant levels, there will inevitably be the risk of missing spikes in pollutant levels if these are episodic events (temporal uncertainty). One would need to take these limitations into account when interpreting the data.
In my data there are multiple sources of uncertainty! I have multiple people recording the data and human error/accuracy can affect the dataset. Collared animal data varies in the frequency of points collected (depending on the situation of the animal, e.g. whether it is a frequent or infrequent conflict individual). Territory and presence/absence is determined using spoor and camera trap data, as well as actual sightings and identification (of spoor, individuals) can be inaccurate. To minimise these we try to have standardised data collection protocols and train regularly, and in some cases generalise conflict hotspot areas over larger temporal scales to build a general picture (which can then vary with spatial features such as river flood level).
I think the only certainty is that there will be uncertainty but it is try to limit it where it is in your power to do so. I will be recording the number of river invertebrates at a number sites which are fixed so my uncertainty that I will be dealing with is the time of day, time of year, river flow rate and temperature which I believe are classified as temporal uncertainties.
IN biogeography applied to herpetology we can have the 3 types of uncertainty:
Define this is tremendously explanatory for the presence of amphibians and reptiles:
Spatial for limits of land use, where is the exact limit between agricultural and non-agricultural field?
Thematic for issues relationed with habitats, that is not clerar (for example if its mediterranean forest or just scrub)
Temporal also with distribution data in raster grids, since many species of reptiles have mobility and species in grids not previously mentioned can be cited.
It is very difficult to represent these variable in a totally exact way, so the good researcher must always live with a certain degree of uncertainty, and always take these deviations into account, science is never an absolute truth.
Collection of environmental data from surveys comes with inevitable uncertainty due to human error, and potential difficulty in categorising things, e.g. from looking at a field, is it 50% or 70% harvested. This can be minimised by having the same surveyor to reduce subjective variation. I found with my data the categories were quite wide, so the generalisation reduced thematic detail
We use a lot of citizen science input for our species distribution , and when the reportings come without photos we can’t always be fully confident that the people did indeed see the correct species. I don’t think this falls under any of the three types of uncertainty though in the end while putting together data maps it might be thematic uncertainty at that point.
For satellites I think it would be temporal uncertainty, or just showing thematic uncertainty when it shows images of the environment .
There is quite a lot of uncertainty from my undergraduate research. It definitely has temporal uncertainty as I worked with frequency-dependent selection and as we did not have working cameras to closely monitor each location, there could definitely have been variables such as multiple birds consuming prey at one site as well as differences in weather across the sites as not all sites went through trials at the same time due to empty nest boxes. Within the funds I worked with, these uncertainties were inevitable, however my research would have greatly benefited from camera monitoring of the sites. However, with the weather issues those were inevitable due to our inability to control when birds nested in the nest boxes. These conditions were not deliberate generalizations or simplifications as they are the result of working within the field. That being said, they may affect our conclusions that were based on few birds present at the nesting locations if in fact there was theoretically a higher population at the nest sites.
Doing field research study comes with different uncertainty. When using satellite imagery there some possibilities that you will have a low resolution because it was not updated and may be it is affected the weather condition on that area. Also, seasonal changes and types of vegetation cover can affect the data collected. Due to this situation it is important to indicate the year, seasons and weather condition in the data. To reduce the uncertainty we can make gathering more data by taking measurement and field study regularly either monthly, yearly, or during the seasonal changes and migration time.
Based on my previous data collection, I did tracking on spiny hill turtle, uncertainty arise when accuracy of GPS device is >+5 (due to canopy cover and weather conditions). Hence the readings are inaccurate at certain times, making it difficult to predict species movements and extant of their habitat. What I did was simply re-calibrate the GPS device multiple times just to make sure the readings are near constant on each visit. Also I use satellite maps and contour maps as reference during tracking, which sometimes hard to tell. Good thing the turtle doesn’t move much at certain period of the months, and they kinda have their preferred resting spots.
My recent project was on Cnemaspis gekkonid distribution. So sometimes, the satellite imagery is not updated and some areas around the karst landscape are converted into conventional plantation. So again, making it hard to predict the extant of the population and habitat.
from the data set given the uncertainty will be the differences in the resolutions
There is always uncertainty with satellite images which mainly comes from weather changes eg rainfalls, clouds effects, etc. Therefore, when selecting images it is necessary to identify these effects and try properly model them in an innovative way.
A specific field data concerned with my work is sand mining on the nesting sites. when collected from general satellite imagery the uncertainty can be observed as thematic and spatial. Most of the sand mining activity is illegal so we need to rely heavily on satellite imagery which often lacks fine resolution resulting in no demarkation in normal and sand mined areas of the beach. It is inevitable due to deliberate generalisation which affects the management recommendation. I can’t figure out how to reduce the sources of uncertainty except to go for expensive high-resolution satellite imagery.
In my work with seagrasses, I am mainly collecting my data using a handheld GPS device. There is always uncertainty due to the error/accuracy in the device. We have already experienced some issues with some overlap of point data due to the inaccuracy of the device. There is also uncertainty temporally. Seagrass cover can change drastically due to storms, human disturbances, etc, so the map I may create this summer, might not be accurate next summer.
I think the biggest source of uncertainty in my data is temporal uncertainty. I’m dealing with climatic data in a place that is highly seasonal, and some of the variation & range in the data is definitely being lost by generalizing the data for use in gis. A lot of this is deliberate because we have to match the temporal scales between datasets, and some datasets were collected at a very coarse grain while others were collected at a fine grain. It could definitely impact the outcome of our research & the kind of recommendations we make - ideally, we’d work on creating matching datasets at a finer grain but we’re contricted by the inability to travel right now.
Both temporal and spatial uncertainty would be an issue if I was doing a GIS-related project at my field site. Controlled burns are conducted annually in the grasslands, so the habitat can be very different depending on the time of year. Similarly, the social groups of my focal species vary seasonally, so to get a full picture of their behavior they would need to be monitored year-round. For spatial uncertainty, we typically only take one GPS point per day in the field, which marks where we began our sampling that day. If we were using this data for a project, we would probably want to take more points so we can see the full extent of our daily sampling rather than just the start.
Spatial uncertainty for Marine Mammals stranding map. When we do for social media search through Facebook, if we know the location of stranding, we reference from MIMU(Myanmar Information Management Unit) or just point out the estimate places for its coordinates (latitude and longitude). It is difficult to contact each person to take the exact places of stranding.