This article was written by Matt, whom has invested hundreds of hours into researching the disappearance case of Kris Kremers and Lisanne Froon.
Matt is a member of the Imperfect Plan team and has written numerous articles, including the following:
- A Deep Analysis of The Night Photos
- Timestamps Of Missing Daytime Photos
- Forensic Analysis of Phone Data
- Missing Photo #509 – Field Testing The Canon Powershot SX270 HS
Additionally, Matt was our central point of contact and provided critical support during the team’s expedition in Panama.
If you’re new to the disappearance case of Kris Kremers and Lisanne Froon, we suggest that you check our Case Articles page.
After Kris Kremers and Lisanne Froon disappeared on the Pianista trail, their blue backpack was later found on the bank of a river by a Ngobe tribe woman. Inside of the blue backpack was Kris and Lisanne’s handheld camera, which contained numerous daytime photos and nighttime photos. Digital photos contain meta-data, called “EXIF data”, which consists of small pieces of information such as image resolution, dimensions, shudder speed, if the flash was used, etc. EXIF data is stored inside of the photos themselves, at the time that each photo is captured.
This is Matt’s further investigation into the night photos that were found on Kris and Lisanne’s camera.
This article examines specific temperature EXIF data that was recently extracted from the night photos (Image 510 to Image 610). The data was kindly provided by Jürgen Snoeren (JurgenSnoeren.com). The night images that the Imperfect Plan team previously has access to did not contain this specific EXIF data.
It should also be noted that Jürgen has published a book with Marja West about the case: “Lost in the Jungle” (Verlorenindejungle.nl).
Before looking at the actual data it is necessary to consider that this data comes from and represents a period of time of pain and anguish of two young women, Kris Kremers and Lisanne Foon. While examining this data this should be considered; and discussion of this data should be done in a respectful manner. This article intends to do this and also intends to disprove some rather creative speculations about the night photos.
Overview Of The Camera Sensor
First, let’s look at where the data comes from and how it is determined during use.
The Canon SX270 HS has a built-in temperature sensor. This sensor is most likely a simple base-level sensor. Where the sensor is located inside the body of the camera is not known. The sensor could be either a discrete sensor, mounted on the circuit board, or it could be incorporated into one of the integrated circuits itself. The sensor’s location could have an impact on how quickly the recorded temperature changes based on ambient temperature. It’s location, however, not overly relevant to analyzing this data. Overall, data recorded by the sensor was sufficient to make some important observations.
The purpose of the temperature sensor is most likely to measure the internal temperature of the camera and to prevent it from overheating, rather than taking a precise measurement of the ambient temperature. Overheating could occur when the flash is used extensively, especially in warm environments.
The camera stores a measurement taken by this sensor in the EXIF data of the image, every time a photo is captured.
Below is EXIF extract from an image taken by the Canon SX270 HS. The camera temperature is stored as an integer value, which is a whole number without decimals.
It is likely that the camera’s temperature is initially measured with higher resolution but is only stored as an integer because the temperature reading is not intended to be a precision measurement. Instead, this temperature is utilized internally to determine if the camera’s temperature exceeds a certain threshold. This prevents heat damage to the internal electronics, and permits the camera to shut down or warn the user if it is overheating.
Unfortunately, this makes this data somewhat less useful because Canon does not published precisely how the camera’s sensor data is processed. For example, is the temperature reading rounded or simply truncated?
The temperature also does not allow us to distinguish small temperature changes, because they are all rolled into the same integer.
For example the values 22.5, 22.6, 22.7, 22.8, 22.9, 23.0, 23.1, 23.2 and 23.4 would all be rounded into the same integer 23. It is furthermore possible that the decimal point is simply truncated and that the values 23.0, 23.1, 23.2, 23.4, 23.5, 23.6, 23.7, 23.8 and 23.9 would be truncated into 23.
Let’s look at an example in the table below. Please note that the data in the following table is not related to the disappearance case and only serves as an example to understand the possible inner-programming of the camera.
The first column (Decimal) are examples of data with one decimal point. In the second column (Truncated) are the same values, but the values after the decimal point are simply truncated and in the third column (Rounded) are the decimal values but rounded to no decimal point.
|#||Decimal (xx.x)||Truncated (xx)||Rounded (xx)|
It is easier to see trends by graphing numbers, so let’s graph them.
The graph below shows the decimal values and small changes in temperature:
The next graph shows the truncated data, without decimals, as is similar to the way in which the camera recorded the temperature data.
In the chart above, it’s immediately visible how a lot of information is lost by excluding the decimal values. Rather than including small changes, we’re only able to observe large changes, which are jumping between full numbers. Therefore, it is not possible to observe any small changes in temperature.
Lastly the rounded data also only shows large jumps and no small changes in the data.
All three graphs above are based on the same data. Only in the last two graphs, the data was rounded to the nearest whole number, and therefore truncated.
The final whole numbers are similar to the way in which the camera logged each photo’s EXIF temperature data.
In summary, a small change in the camera’s temperature may appear as a change of a full degree and conversely small changes in temperature may appear as no change or stable temperature. Data may show large sudden jumps in temperature, however those may in reality be only very small changes.
It is important to remember these value changes when reviewing the real temperatures so that they can be interpreted correctly.
Raw Night Photo Temperature Data
The EXIF Temperature data from every night image is shown in the table below.
The temperature of 21 C from the first image (Photo 510) is most likely the temperature of the environment that the camera was in before the first image was taken. A temperature of 21 C would be realistic for the ambient night temperature in April in these approximate regions of Panama. This temperature is similar to the night temperatures that the Imperfect Plan team encountered during their expeditions and it can be assumed the camera was in ambient temperature before Image 510 was taken. This means the camera was likely not in someone’s hand, nor in someone’s pocket. The camera must have been either placed on the ground or had been sitting in the blue backpack. Had the camera been in someone’s hand or pocket, the temperature of the camera would likely have been higher, as we’ll see in the following experiments.
|Image #||EXIF Temperature (C)|
The graph below shows the temperatures plotted over the times the image was taken.
The graph can be broken into three distinct areas:
- Area 1: Where many images are taken in short succession and where the temperature rises quickly at a very constant rate.
- Area 2: Where images are taken at a slower rate and the temperature remains reasonably stable.
- Area 3: Where images are taken at a very low rate and the temperature increases slowly.
Here are the areas on the graph:
Area 1 shows what is expected when the camera takes a long series of images with the camera’s flash. The camera heats itself from use, especially when using the flash. The camera temperature increases in increments of +1 C and many points are observed having the exact same temperature, which illustrates that the camera only stores integer values.
Area 2 shows the temperature jumping up and down slightly which at first sight appears unnatural, however this is likely also caused by the temperature being stored as integer values. A very small increase or decrease of the camera temperature, caused by the camera having enough time between images to cool down slightly, would cause the stored temperature value to change by a full 1 C.
Area 3 shows unexpected behavior where the temperature increases despite the rate, at which images are taken, is very low. This is not what is expected. It would rather be expected that the temperature drops closer to 21C as the camera is not used for longer periods of time.
Experiments To Identify Normal Temperature Behavior
To find out if the observed camera temperature behavior is normal, it is necessary to conduct an experiment.
The experiment requires 99 images to be taken at the exact same time intervals that photos 510 – 609 were captured, in a dark environment with the same ambient temperature of 21 C.
After performing the experiment, the new temperature data can then be compared to the temperature data of the real night photos that Kris and Lisanne had taken. It is necessary to perform this experiment in darkness because the camera’s flash intensity automatically adjusts based on the ambient light. In complete darkness the flash will be triggered at its maximum power and therefore leading to the quickest increase in temperature.
Constructing The Robot
This was a challenging experiment to perform, primarily due to the very random intervals that the original photos were taken, and the long span of time that the original photos were taken (approximately 2.5 hours).
Therefore, to overcome these challenges, a little robot “Rudy” was built.
Rudy is a very crude robot, that consists of a wooden frame to hold the components and camera in place, a servo to actuate the camera’s shutter button and lastly an Arduino based microprocessor that operates the servo.
Below is a picture of Rudy with the Canon SX270 HS, the same exact model camera that Kris and Lisanne used.
The microprocessor was then programmed with the duration in milli seconds between the real night pictures and instructed to actuate the servo to press the shutter button after each interval.
For the camera to take a photo in complete darkness it had to complete four steps:
- Achieve Focus
- Calculate Exposure
- Charge Flash
- Take Image
Focus: The camera has a simple autofocus based on contrast detection. Without going into too much detail, the camera will try to find the focus position that shows the highest image contrast, which in return usually means that focus is achieved. It can easily be imagined that an object with clear edges will show a high contrast edge, while if it was not in focus, it will leave a blurry edge with low contrast. In complete darkness this process will fail, unless the Autofocus assist light can provide enough light, and the camera will either default to a set focus distance or if it miscalculates the focus distance, it will focus on a random distance. What matters for our experiment is only that this takes some time, the actual focus is irrelevant.
Exposure: For the camera to take an image it will also try to calculate the correct exposure and set ISO speed, aperture and shutter speed accordingly. In complete darkness it will try to capture the maximum amount of light and set those parameters to achieve this. The details are irrelevant and it is again only important that this takes some time.
Flash: Lastly the camera will wait until the flash is charged (if flash is needed for correct exposure).
This entire process takes somewhere in the range of 3-7 seconds. The difficulty is that Rudy does not know when the camera actually takes a photo. It would have been possible to give Rudy a feedback sensor to know when the camera took an image but this would have added a lot of complexity and is not needed.
Thus, Rudy has to press the shutter long enough to assure that the camera has enough time to take the picture. Rudy was programmed to press the shutter for 3000 ms, which resulted in the shots being taken in most cases. But the real night photos that were taken so quickly that it was not possible for Rudy to take all pictures and he consistently missed images 563 and 580.
Experiment 1 – Camera Always On
Now, on to the experiment.
Camera: Canon SX270 HS
Settings: Auto, Camera prevented to go to sleep mode
Ambient Temperature: 21 C
Ambient Light: Complete Darkness
Rudy was activated and took the pictures. The camera temperature was then extracted from the EXIF data and plotted (see graph below). The camera settings were configured to prevent it from going to sleep, which it normally does on default settings. On default settings the camera goes to sleep after a duration of time. Changing these default settings was necessary to reduce the chance for Rudy to miss taking images. When the camera goes to sleep, it takes a small amount of time for the camera to wake up, which would hinder the results of the experiment.
Here are the results of the first experiment:
Note: The camera had not cooled down completely to 21 C from a previous experiment, therefore the starting temperature was 23 C. This does not make a great difference once the camera warms up, as you will see in Experiment 2.
The temperature graph looks similar to that of Kris and Lisanne’s EXIF data in some areas. For example, Area 1 shows the same steep increase in temperature as the originals. Area 2 does not show as much temperature variation as in the original data and Area 3 is very different.
Compared to the original case EXIF data, instead of an increase of temperature in Area 3, it shows the temperature hovering steadily at 26 C. The camera does not cool down further because it’s powered on, which creates heat, thus reaching an equilibrium temperature that is above the ambient temperature.
This experiment does not need to be analyzed further. This scenario is unrealistic because the camera was intentionally prevented from going into sleep mode. However, this experiment demonstrates how the camera heats itself, even when the flash is not used, while the camera is left on.
Experiment 2 – Default Settings (Sleep Mode Enabled)
The next experiment was the same as above, except on automatic settings (“Auto”), which involves the camera going into sleep-mode automatically after a while, during which time the camera no longer heats itself.
Camera: Canon SX270 HS
Ambient Temperature: 21 C
Ambient Light: Complete Darkness
Here are the results from Experiment 2:
The sensor in the Canon SX270 HS used for this experiment was very accurate and matched the ambient temperature.
The blue line between IMG 596 and IMG 600 was interpolated and shows how much the camera would cool down when not used, based on a theoretical cooling rate of 1 C in 180s. It is not easy to determine this precisely since the measurements of the camera are only in full degrees. The blue line is intended to illustrate the hypothetical cooling-rate of Kris and Lisanne’s camera if it was completely turned off, which it may have been.
The new data from Experiment 2 again illustrates that the camera warms up quickly to 29 C, while the original data shows an increase to only 27 C. Although the relationship between the data is very strong, the small variations were caused by unknown variables, which resulted in the camera warming up slightly less. For example, the experiment was performed without the camera being touched or moved at all. Kris and Lisanne’s camera was likely moved and touched when their original photos were taken, which could affect the manner in which the camera warmed.
Generally, it can be concluded that the increase in camera temperature identified in Area 1, is the typical temperature behavior, when photos are taken with the flash in quick succession.
Area 2 shows differences to the original pictures, as it is rather stable with one temperature peak at IMG 596, while the original data shows the temperature going up and down several times.
As mentioned before, the camera’s temperature differences are likely affected by the manner in which the camera’s programming rounds or truncates the recorded temperature data (see above). Furthermore, the camera’s temperature differences can be affected by other variables, such differences between the simulated conditions vs real conditions. For example, how the camera was held, if it was turned off, if it was moved, if it got wet, if there was a breeze, etc. Generally, Area 2 is reasonably similar to the original data.
Area 3 is very different to the original data. While the experiment shows that camera cooled down to ambient temperature by the time IMG 609 was taken, the data from Kris and Lisanne’s photos show that the temperature increased to 33 C, which is even far above the temperature achieved during the experiment, after prolonged flash use (during Area 1).
This substantial difference merited a third experiment.
Experiment 3 – The Impact Of Body Heat
Besides considering intrinsic issues with the experiment, the only other cause for the discrepancy in temperature could have been that the camera was exposed to a second heat-source besides the camera itself.
Alternative heat sources could have been:
- A fire
- A heated room
- Body heat
The first two possibilities were dismissed as unlikely and the experiment was repeated with the camera being exposed to body heat between images 602 and 609.
The experiment simulated a series of steps in order to mimic the original case data.
As can be seen in the original case photos (from Kris and Lisanne’s camera), the photographer took a continuous series of photos up to IMG 601 and then stopping taking images. During this interruption the camera was turned off, and either held in the photographer’s hand, put into a pocket or tucked into pants, thus providing some degree of contact with body heat.
Then, according to their EXIF data, the camera must have been warmed during the periods of time that no photos were taken.
Considering that the temperature of hands and feet usually has the lowest body temperatures, the camera was tucked into the pants of the test subject in order to duplicate the temperature conditions during the experiment.
The experiment was split into two phases.
Phase 1: Image 510 to 601, the same workflow was utilized as the previous experiments, wherein Rudy took the images according to the same time schedule.
Phase 2: Then the camera was removed from Rudy and the images were manually taken, based on cue from Rudy to ensure the correct timing. In between images the camera was turned off and exposed to the test subject’s body heat, as was previously described above. A few seconds before each shot was taken the camera was removed from the test subject, turned on and the image taken. Directly after the image was taken, the camera was turned off and placed back in position with the test subject.
The results of this experiment are illustrated in the following graph:
Note that for this graph, the data of the last 8 images that were not taken by Rudy, was overlaid to the previous graph in order to simplify the experiment and make the graph more easily comparable.
The resulting graph is now more similar to the original, however the original peak temperature is 33 C while the peak temperature of the experiment is only 30 C . It is important to remember that there are certain inaccuracies involved when the camera measures and records the temperature. A difference of 1-2 C between two different cameras is therefore not unreasonable to expect. Another unknown factor is the difference in body temperature of the original photographer and the test subject and the exact treatment of the camera when the original images were taken. A variety of subtle factors were likely the cause of the differences in the results.
Below the original data and the “body heat experiment” data were combined into one graph for easier visualization:
The two graphs match qualitatively but because the exact original conditions are unknown and because the experiment was conducted under static conditions, it is apparent that there are differences.
Conclusions & Takeaways
The experimental data confirms that the camera heated up quickly during the first roughly 70 images. Then there was a period of more or less constant camera temperature where the images were taken with less frequency and the camera was able to cool between images. Finally the last 8-10 images where the camera was exposed to an external heat source that could have been body heat.
The experiment shows that:
- The camera was turned on and off during taking of the night photos.
- The camera was not placed on the ground or in the backpack while the last 8 – 10 images were taken.
- There was an external heat source that the camera was exposed to.
- The camera temperature can occur naturally, i.e. during normal anticipated use of the camera
It can be speculated that:
- The person taking the images put the camera into a pocket or other location where it was close to body heat.
- The photographer’s body temperature was normal or could have been higher than that of the test subject
- The person taking the images was alert and conscious since they turned the camera on and off multiple times and placed the camera in a “storage location”, e.g. pocket between some of the images while it was not in use.
What can be ruled out with this experiment is:
- The camera fell onto the ground and “took” the images by itself
- The camera took the images by itself due to unspecified malfunction
- An animal “took” the images while manipulating the camera in some fashion
The data and the experiment does not show signs of data manipulation and does not show signs of influence of third parties. However, it can also not be ruled out since any person could have taken the photos as described above.
It is the author’s opinion that there is no evidence of wrong doing based on the data discussed in this article.
A Few Final Comments From Chris
I’d like to extend a big “thank you” to Matt for this excellent work and Jürgen Snoeren for collaborating with Matt to make this possible. This new information is yet another important piece of the puzzle that contributes to our understanding of the disappearance case of Kris Kremers and Lisanne Froon.
I found it particularly interesting that the initial ambient temperature was 21 C (69.8 F) and that the camera temperature increased after the initial cluster of photos, most likely from body heat. This supports our evolving theory, that maybe Kris and Lisanne had opted to go further north from River 3, towards lower elevations and higher temperatures.
Altogether, this information helps us to better understand some details about when the night photos were taken. Matt did a fantastic job at reconstructing the known conditions that were necessary to conduct these experiments.
Additionally, thank you to all of our supporters and followers as we make progress in this work.