Marshall's Journal

2023-10-02

We did it. Marsh, Meadow and dax, home alone for 48 hours after our mission abort trip to NYC due to the torrential rain and flooding. Alej went by train the next day while we three held down the fort. Very little time for my attention to wander, save nap times. During naps, I continue to think about weather forecasts, about the steep learning curve and convoluted methods of access. Of their centralization and points of failure (gov shutdown? sorry!).

The uninterrupted and intense time with Meadow went much better than expected. She ate voraciously, slept consistently and was a joy to walk and play with. It's a silly thing, but I'm proud of myself for handling it with little stress and no mishaps.

Tired now. Some things I want to write soon: 1. Art imitates life, life imitates art: videogame edition: I've been studying some of the techniques and math behind shaders, game simulations and other technological feats and thinking about how they apply to problems in my universe of interest. 2. A guide to starting a business/organization, if I were to do it again ;) I get a lot of outreach from folks asking me for "readmes" for my current co. Although I think I did quite a good job, I'd change a few things if I had the opportunity to.

I drew some goldenrod and some architectural drawings (I'm very bad at the basics, consistent perspective and scale) - but fun nevertheless.

2023-09-20

Went down a TodePond rabbit hole over a cup of tea (i.e. this live talk). Their energy and enthusiasm about spatial programming -- and their storytelling bringing us along the discovery journey -- was incredible. It was one of those mind expanding nights. I've been dealing in "spatial data" from the meteorology, satellite, and geospatial perspective for some time, but this concept of rule-based computation (like, say Orca), explained in this way, had me scribbling down sparks of thoughts, not yet ideas, about how they could relate to forecasting dynamical systems and defining environments.

Thank you, Lu!

2023-09-15

We've spent the last few months in the mountains of the western part of the state. Cool, misty mornings and sunny afternoons perfect for swimming in the many lakes and streams. My trail shoes got some solid miles; on my last run before we came home I saw three bears, a mother and two cubs rolling and playing and scratching trees. I don't think they saw me as I observed for a while before doubling back to not disturb them.

We also did a first excursion up the green river, walking in and through the riverbed itself, shifting bank to bank to avoid the strongest currents and most harrowing rapids. I'd like next time to travel the full distance upstream to see where the river takes us.

Two things I would recommend not mixing: a newborn entering your family and a complicated and prolonged corporate restructuring/financing. Thankfully, one of these things makes my heart sing. The other has been a metaphorical storm cloud; I've done a lot of journaling about it but nothing I can share yet.

I can't stop thinking about weather forecasts. A few things in particular:

1. Numerical weather predictions (NWPs) are incredibly hard to access, both due to the most common file format and their dimensionality
2. These hard to access predictions are surprisingly inaccurate (you'd think we - humans - would be better at this by now, but it's very hard)
3. The forecasting models are extremely brittle to an input being unavailable
4. NWPs, when presented, don't do a good job communicating confidence intervals

In talking to my longtime collaborator, there's the seed of an idea how how to address all of these issues. We have more thinking to do, but the benefits of an improved weather forecast are far-reaching and energizing.